The Timing of IFNβ Production Affects Early Innate Responses to Listeria monocytogenes and Determines the Overall Outcome of Lethal Infection

Dendritic cells (DCs) and natural killer (NK) cells are essential components of the innate immunity and play a crucial role in the first phase of host defense against infections and tumors. Listeria monocytogenes (Lm) is an intracellular pathogen that colonizes the cytosol of eukaryotic cells. Recent findings have shown Lm specifically in splenic CD8a+ DCs shortly after intravenous infection. We examined gene expression profiles of mouse DCs exposed to Lm to elucidate the molecular mechanisms underlying DCs interaction with Lm. Using a functional genomics approach, we found that Lm infection induced a cluster of late response genes including type I IFNs and interferon responsive genes (IRGs) in DCs. Type I INFs were produced at the maximal level only at 24 h post infection indicating that the regulation of IFNs in the context of Lm infection is delayed compared to the rapid response observed with viral pathogens. We showed that during Lm infection, IFNγ production and cytotoxic activity were severely impaired in NK cells compared to E. coli infection. These defects were restored by providing an exogenous source of IFNβ during the initial phase of bacterial challenge. Moreover, when treated with IFNβ during early infection, NK cells were able to reduce bacterial titer in the spleen and significantly improve survival of infected mice. These findings show that the timing of IFNβ production is fundamental to the efficient control of the bacterium during the early innate phase of Lm infection.


Introduction
Protective immunity requires the coordinated activation of both the innate and adaptive immune systems. Interactions between innate and adaptive immune effectors are essential for the efficient control of pathogens and tumors and often play an important role in ending immune responses which would otherwise eventually be harmful to the host. DCs and NK cells play an essential role in the first phase of pathogen infection. Mice infected with the intracellular Gram-positive bacterium Listeria monocytogenes (Lm) are often used as a model to study innate and adaptive immunity [1,2]. Major sites of infection with Lm are the spleen and liver, where bacteria are found preferentially within the cytosol of antigen-presenting cells (APC) and hepatocytes. Lm can spread from cell to cell without leaving the intracellular compartment, which is the main reason why Lm-specific CD8+ T cells are needed for bacterial clearance and for providing protective immunity. In the absence of T cells, chronically persistent infections can develop [3]. Bacterial growth during the first days after infection is mainly controlled by cells belonging to the innate immune system, including neutrophils, NK cells, and macrophages [4,5]. Upon infection in the murine spleen, Lm is first found within macrophages and DCs in the marginal zone between the T cellrich white pulp and the B cell-rich red pulp [5,6,7]. These infected cells then migrate to the white pulp region and form the beginning of a focal infection that expands as neighboring cells become infected by the intercellular spread of bacteria. The cytokine IFNc plays a major role in the control of Lm infection during both the innate and adaptive immune responses. NK cells and NKT cells secrete IFNc and are thought to limit exponential growth of the bacteria primarily by activating macrophages during the first few days of infection [8,9]. IFNc secretion by these cells is also thought to promote a Th1-type response against Lm by increasing MHC class I expression.
In addition to 'bridging' innate and adaptive immunity, DCs may also contribute to primary resistance against infection. Recent reports have suggested that DCs themselves may be involved in innate defense against infections [10]. In Lm infection, previous findings have suggested that a CD11c + DCs subset characterized by the production of tumor necrosis factor alpha (TNF-a) and inducible nitric oxide synthase (iNOS) is essential for early control of bacterial growth [11].
In addition to the major role of DCs during T-cell priming, a recent study reported an important role for DCs in CD8 + memory T-cell responses upon secondary infection with Lm [12,13]. Localization of Lm to the cytosol is required for virulence and for recognition of the bacterium by intracellular DNA sensing machinery. Lm produces a hemolysin, listeriolysin O (LLO), which permits the bacterium to destroy phagosomes and escape into the cytosol of infected cells. Consequently, strains lacking expression of LLO (DHly) are avirulent. In addition, Lm DHly fails to elicit the production of IFN-ab by infected macrophages [14]. Production of IFN-ab during Lm infection is thought to be dependent on the detection of microbial products by a receptor present in the host cell cytosol [15]. Nevertheless, IFN-b induction by a TLR2dependent mechanism has been also reported [16]. The DCs cytosol is a major host-pathogen interface during Lm infection. We analyzed events during host-pathogen interactions, by comparing the DCs response to two Listeria strains, Lm and the non pathogenic Listeria innocua (Li). Lm localizes to the cytosol whereas Li does not. Entry into the DCs cytosol activates a complex gene reprogramming that results in the release of proinflammatory cytokines, type I IFNs and chemokines. We describe a Lm-specific delay of type I IFN induction compared to E.coli in DCs and show that this delay has marked effects on NK cell activation and on mouse survival to lethal Lm infections. Thus, Lm may evade innate immune surveillance specifically by disruption of the DCs-NK cell interaction and by modification on other regulatory mechanisms through an improper induction of type I IFNs.

In vitro Production of Type I IFNs by DCs Infected with Listeria Monocytogenes is Delayed Compared to DCs Infected with other Bacteria
We studied the host responses to Gram-positive bacteria of the genus Listeria, specifically, the pathogenic species Listeria monocytogenes (Lm) and the non-pathogenic species Listeria innocua (Li). Genome organization is conserved among many Listeria species and they frequently have a high number of genes in common. However, some Listeria species, such as Lm, carry specific genes that allow them to cause disease in humans and animals [17]. We investigated the interactions between these two Listeria strains and DCs using microarray technology. We compared the transcriptomes of Lmand Li-infected D1 cells, a spleen-derived murine DC line [18], at 2, 4, 8, 12 and 24 hours post-infection (p.i.). The D1 cells were exposed to Lm or Li at a multiplicity of infection (MOI) that was optimal for triggering the up-regulation of co-stimulatory molecules. Examples of flow cytometry profiles of D1 cells infected with Lm, Li and E. coli are shown in Figure S1. After normalization and filtering for gene quality control in the microarray analyses, 10,347 probe sets remained for further evaluation. We selected only those genes for which the differential expression changes were 2.0-fold or greater and were statistically significant (one-way ANOVA, p,0.0001). We found 1,775 genes in the Lm-infected group and 2,219 genes in the Li-infected group which had at least a 2.0-fold change in transcription level relative to those in the uninfected control cells. These differentially expressed genes were studied in further detail in an effort to understand the mechanisms of the host response to these Gram-positive bacteria.
Consistent with previous studies of macrophages [19], we found that the transcriptional responses could be broadly divided into ''early'' (224 hr p.i.), ''middle'' (8212 hr p.i.) and ''late'' (24 hr p.i.) responses. At early time points, Li infection induced a stronger transcriptional response (824 genes) than that induced by Lm infection (119 genes). Only at 24 hr was the response to Lm infection (1,508) higher than the response to Li infection (1,227) ( Figure 1). An unsupervised analysis revealed groups of genes with similar changes in expression. A large number of genes encoding receptors, signaling molecules and transcription factors, as well as genes encoding adhesion molecules, enzymes and anti-apoptotic molecules and genes involved in tissue remodeling were differentially regulated in DCs infected with Listeria. Particularly, gene clusters corresponding to the early response were enriched in genes involved in the ''cytokine-cytokine receptor interaction'' (KEGG pathway number 04060). The response of Li-infected DCs (29 genes; p = 9.15610 29 ) was more highly enriched in these genes than was the response of Lm-infected DCs (14 genes; p = 1.92610 27 ) ( Figure S2). GO functional term annotation revealed different enrichment profiles for the two different Listeria-infected DCs, which reflected the different numbers of genes induced at the early time points. The biological processes most enriched in the Lm-infected cells were ''locomotion'' (GO0040011; p = 2.63610 210 ) and ''chemotaxis'' (GO0006935; p = 1.53610 29 ). In contrast, ''immune system process'' (GO0002376; p = 5.59610 29 ) and ''immune response'' (GO0006955; p = of 6.03610 28 ) were the biological processes most enriched in the Li-infected cells ( Figure S3). The data revealed that at early times (224 hr p.i.), Li bacteria induced a greater transcriptional response in DCs, which was characterized by a broad induction of immune system genes. In contrast, the functional classes of genes that were most enriched among those induced by Lm bacteria were specifically related to chemotaxis and locomotion.
Next, we compiled a list of all the genes induced by the two bacteria and identified those that were induced by both species (1,361) and those that were specifically induced by Lm (414) or by Li (858). Functional annotation of the genes specifically induced by Li revealed statistically significant enrichments for ''phosphate metabolic process'', ''regulation of apoptosis'' and the ''MAP kinase signaling pathway'' ( Figure S4 and data not shown). The genes induced specifically by Lm were mainly part of the ''response to virus'' class (Oas1b, ISG15, Tgtp2, Ifna2, Ifna4, Ifna5, Ifnb, Mx2 and POLR3C). In particular, the induction of these genes was most evident in the middle time points (8 hr and 12 hr p.i.) and included a large number of genes induced by type I IFN signaling. This finding suggests that the detection of Lm in the host cytosol is associated with a type I IFN response in DCs, as has been reported previously for macrophages and other cells [14,20,21]. The late cluster of the Lm-infected DCs was characterized by the presence of interferon-regulated genes (IRG) ( Figure 2A) and by different type I IFN subtype genes ( Figure 2B). Changes in IFNb gene transcription were first observed at 4 hr p.i., whereas an effect on IFNa genes was first detected at 8 hr p.i. ( Figure 2B).
To confirm data obtained from the transcriptional analysis, we measured IFNb and IFNa production levels in infected D1 cells. Variable levels of IFNa and IFNb proteins were released during Lm infection, and peak production occurred at 24 hr p.i. ( Figure 3A). Because the anti-IFNa antibody used in the ELISA was unable to distinguish between different IFNa subtypes, we used quantitative real-time reverse transcription PCR (qRT-PCR) to analyze the expression of IFNa4, IFNa9, IFNa2 and IFNa5, and we used an RT-PCR analysis to detect IFNa1 and IFNa6. We were unable to analyze IFNa7, IFNa11, IFNa12 and IFNa13 expression due to the absence of oligonucleotide sequence specificity. Transcripts for IFNa4, IFNa2, IFNa5 and IFNa9 were all induced at 12 hr and 24 hr p.i. The highest level produced was for IFNa4 (12,000 times the IFNa4 transcript level in untreated controls) and the lowest level of induction was for IFNa9 (1,400 times the level in the untreated controls) ( Figure 3B). IFNa1 and IFNa6 gene induction were also detected in Lm infections in DCs at 8 hr, 12 hr and 24 hr p.i. ( Figure 3C).

Type I IFNs are Induced in Bone-marrow Derived DCs (BMDCs)
We then investigated the ability of Lm infection to induce type I IFNs in freshly isolated BMDCs. Although type I IFNs are known to be induced in response to Lm, we were interested in understanding whether delayed production of these important cytokines could affect the generation of very early innate responses. We therefore compared the host response production of type I IFNs using bacteria that modulate these genes through different cellular pathways (cell membrane stimulation versus intracellular stimulation), which result in different type I IFN production kinetics.
We infected wild-type BMDCs with Lm or the Gram-negative bacterium E. coli at an MOI of 20 and measured type I IFN production in the supernatant at different time points. As E. coli LPS signals through the cell surface Toll-like receptor (TLR) TLR4, E. coli is known to induce early production of type I IFNs [22]. As expected, type I IFN production was efficiently induced in BMDCs soon after E. coli infection. Lm infection also induced type I IFN production in BMDCs, but as this requires cytosolic detection by these cells, the kinetics of IFN secretion differed substantially from those triggered by E. coli. During Lm infection, IFNb secretion was first detected at 8 hr p.i., whereas, in E. coli-infected cells, the main peak of IFNb production occurred as early as 4 hr p.i. ( Figure 4A). The difference was even more evident for IFNa: whereas E. coli induced a peak in IFNa production at 2 hr p.i., induction by Lm did not begin until 8 hr p.i., reaching the maximum level at 24 hr p.i. ( Figure 4B). Thus, Lm induced a late, type I IFN response in D1 cells and in BMDCs, which was consistent with our microarray data and predicted by the detection system activated by this microorganism in DCs. The finding that cytosolic Lm detection is responsible for the late production of type I IFNs in DCs compared to the early IFN production caused by cell surface receptor triggering, prompt us to verify whether these different type I IFNs kinetics are also measurable in vivo when mice are challenged with live bacteria. We focused on the production of IFNb as it appears to be the primary type I IFN family member that is induced after infection with Lm. We infected C57BL/6 mice with Lm or E. coli (10 6 CFU) and extracted total RNA at different time points from whole spleens and from CD11c + cells purified by magnetic bead cell sorting. The sorted CD11c + cells were enriched for IFNb production in response to Lm, whereas the CD11c 2 cells showed a much weaker response. IFNb mRNA was detectable in total spleen lysate ( Figure 5A) and in CD11c + cells of Lm-infected mice ( Figure 5B) at only 24 hr p.i. In contrast, in vivo infection with E. coli induced IFNb production at an earlier stage as IFNb gene induction occurred at 4 hr and 8 hr p.i. ( Figure 5C). CD11c + DCs were involved in the production of this cytokine in the spleen of Lm-infected mice. Thus, these data confirm that IFNb production is strongly induced in the CD11c + fraction of spleen cells and that this production is delayed in the mouse model of systemic Lm infection relative to the timing of the IFNb produced in response to E. coli infection.

IFNb Treatment during the First Few Hours after Lm Infection Improves the Survival Outcome of Lethally Infected Mice
We performed a survival assay to determine whether the presence of an early source of IFNb could affect the survival of mice injected with a lethal dose of Lm. We injected mice intravenously with Lm (10 6 CFU), with or without a subsequent injection of IFNb, and then their survival was monitored. Early IFNb treatment significantly prolonged the survival of mice injected with a lethal dose of Lm by up to one week ( Figure 6). Mice injected with Lm alone had a low survival rate after one week (20%), whereas mice injected with Lm and IFNb had a survival rate of 90% at this same time point. These results indicate that without the presence of IFNb during the early phase of Lm infection, mice were unable to mount a rapid innate immune response in vivo, resulting in poorer control of bacterial spread. Thus, in our experimental model, the timing of IFNb production is crucial for controlling Lm lethal infection.
We speculate that Lm does not activate the full innate immune response during the very early phase of infection. As treatment of mice with IFNb increased their survival after infection with a lethal dose of bacteria, IFNs appear to triggered antibacterial activity in vivo and may enhance innate immunity during intracellular microbial infections.
BMDCs Infected with Lm are Severely Impaired in their Capacity to Elicit IFNc Production by NK Cells The observation that delayed induction of type I IFN affected the strength and the quality of the early innate immune response prompted us to study the interaction between Lm-infected DCs and NK cells. Type I IFNs have been primarily associated with antiviral responses and their role in antibacterial immunity has remained unclear [23,24]. Type I IFNs are produced immediately after pathogen infection and profoundly affect the nature of innate and adaptive immune responses [25]. Therefore, we investigated whether the late induction of type I IFN genes has a functional effect on the innate immune response to Lm. Recovery from infection depends on the host's ability to mount effective early innate responses to control the invading pathogens. During Gramnegative bacterial infections, early type I IFN production by DCs (during the first five hours) is essential for the initial induction of antibacterial NK cell function [22]. Activation of NK cells during the innate phase of the immune response has also been shown to be important in controlling the spread of a variety of microbial pathogens [26]. We therefore hypothesized that a delay in type I IFN production by DCs following Lm infection may cause a delay in NK cell activation, markedly affecting the development of an efficient innate response able to control the infection.
We first studied the ability of Lm-infected BMDCs to elicit IFNc production by syngeneic NK cells. BMDCs were activated with Lm or E. coli at a MOI of 20. Syngeneic NK cells were then added to the co-culture after one hour. At 18 hr, culture supernatants were collected and the levels of IFNc in the supernatants were measured by ELISA. Whereas E. coli-infected BMDCs induced IFNc production by syngeneic NK cells as previously described [22], the capacity of Lm-infected BMDCs to induce IFNc production was severely impaired (Figure 7). Induction of IFNc production in NK cells was fully restored by adding exogenous recombinant IFNb to the BMDC-NK cell co-cultures immediately after activating the BMDCs with Lm ( Figure 7).
These data confirm our hypothesis that the inability of Lminfected DCs to produce type I IFNs during the early stages of bacterial infection, at least in vitro, disrupts the activation of NK cells.

NK Cell Activation is Enhanced in Response to IFNb Treatment in vivo
Given that an early source of IFNb was required in vitro to elicit IFNc production by NK cells and that, during in vivo Lm challenge, induced IFNb production was maximal at 24h p.i., we investigated whether delayed IFNb production also affects NK cell activation in vivo. We infected mice with 10 6 CFU of E. coli, as an early IFNb inducer, or with 10 6 CFU of Lm either with or without subsequent injection of rIFNb (38,000 U/mouse). IFNc production in the NK cells of infected mice was assessed after five hours of infection by staining for intracellular IFNc in DX5 + cells. Consistent with our in vitro data, only a small percentage of NK cells (6%) in Lminfected mice were positive for IFNc production ( Figure 8A). Thus, mice infected with Lm were unable to mount a full NK cell response, in terms of IFNc production, during the very early stages of infection. Injection with rIFNb one hour after bacterial challenge resulted in a substantially higher percentage of IFNcpositive NK cells (22%) in Lm-infected mice, which was similar to the percentage found in E. coli-infected mice ( Figure 8A).
We then measured the cytotoxic activity of NK cells from the spleens of mice treated with Lm and rIFNb, with Lm only or with rIFNb only. The capacity of NK cells to kill YAC-1 cell targets was higher (15% lysis) in mice injected with rIFNb one hour after bacterial challenge than in mice injected with Lm (4% lysis) alone or rIFNb (5% lysis) alone ( Figure 8B). To assess whether exogenous rIFNb was associated with improved control of bacterial growth, the bacterial burdens were evaluated in the spleens of Lm-infected mice at 5 hr p.i. At this early time point, the mean bacterial titer in the spleens of Lm-infected mice treated with rIFNb was moderately but significantly lower than the mean titer in the spleens of Lm-infected mice lacking rIFNb treatment (Figure 9). These results suggest that the protective action of rIFNb is only partially due to the reduction of bacterial growth, which implies that additional mechanisms are primarily responsible for the effects of IFNb treatment. Type I IFNs mediate many anti-cellular effects by modulating cell viability and function which ultimately may lead to the induction of disease tolerance instead of disease resistance [27]. In conclusion, our results suggest that the production or presence of early IFNb is important for regulating innate responses during the initial phase of Lm infection.

Discussion
The analysis of bacteria-host cell interactions at the molecular and cellular levels has become a major research area in recent years [28,29]. Host defenses against intracellular pathogens such Lm require the coordinated interactions of a number of innate and adaptive immune system components to clear the infection [1,30,31,32]. Early activation of innate immune cells is important for host survival and bacterial clearance [4,6,11,33,34,35,36] whereas development of adaptive immunity is crucial for longterm protection and for providing sterilizing immunity. However, it is becoming clear that in the context of Lm challenge, innate immunity alone is unable to completely control the infection and the development of adaptive immunity is necessary to fully eliminate the pathogen.
The general requirement of DCs during Lm infection has been studied by using different experimental models but the exact role of DCs in combating infection remains unclear [5,13,37,38,39,40]. A subpopulation of CD11c + cells has been implicated in various stages of Lm immunity, including splenic sequestration [5,41,42], early activation of CD8 + T-cells [43].
Expression profiles have been widely used to determine the molecules involved in complex host-pathogen interactions [44].
Post-infection gene expression has been evaluated at short/ middle time points (at 0 hr and at 128 hr p.i.) [15,19,45] and with end-point assays [45]. Nevertheless, no previous study has addressed the genome-wide DC-specific response during the first 24 hours of Lm infection. Microarray analyses have shown that the IFNb gene and interferon-stimulated genes (ISGs) are among the genes most highly induced by cytosolic Lm in macrophages or in Lm-infected mice [19,46,47].
Typically, bacteria activate type I IFN signaling through TLRdependent mechanisms, which involve recognition of LPS from Gram-negative organisms, or via TLR-independent cytosolic receptors that respond to bacterial nucleic acids either endocytosed or secreted directly into host cells [48,49]. Lm activates cell surface receptor and cytosolic receptor signaling pathways [50]. Specifically, cytosolic bacteria trigger a unique cytokine response that includes production of type I IFNs [11,14,51].
Although several cytokines are important for immunity to Lm infection, type I IFNs appear to be deleterious to the host [51,52,53,54] however, opposing roles have been described [55]. Whereas type I IFNs provide protection against viruses, their role in bacterial infection is less clear [24,56]. In addition, the diverse effects of type I IFNs include increasing sensitivity to Lm-induced cell death [57]; inducing the downregulation of the IFNc receptor gene (IFNGR) and thus rendering the host cells refractory to IFNc, which is crucial to achieving host resistance to Lm [58]; and regulating the expression of chemokines important for leukocyte recruitment [55,59,60]. And indeed, monocytes recruitment to Lm infected spleen is maintained in the absence of either MyD88 or IFNAR signaling. However, deficiency of both pathways impair cell migration and increased rather than decreased susceptibility to infection indicating that type I IFN has both positive and negative effects on the resistance to Lm infection [55].
In this study, we confirmed that IFNb gene expression is strongly induced in DCs in response to Lm infection, and that the peak of this production (24 hr p.i.) is consistent with that described previously [61]. Our data indicated that type I IFNs are not induced by Lm with the same early kinetics observed in virus infections, which cause these cytokines to be produced within a few hours after infection.
This observation is compatible with the cytosolic detection of Lm bacteria by an as yet unknown intracellular receptor. We postulated that intracellular recognition as opposed to recognition by a cell surface receptor could lead to specific pathogen adaptations.
Thus, we sought to determine whether the delay in IFNb production could provide a selective advantage for Lm spreading during the very early phase of an activated innate immune response. We hypothesized that the timing of IFNb production could alter the nascent host innate response, and for this reason, we studied the effect of the lack of IFNb production at 5 hr p.i. To determine the impact of delayed IFNb production in Lm infection, we first performed a survival assay in which the mice were infected with Lm with or without the administration of exogenous IFNb. We demonstrated that at the very early phase of Lm infection, treatment with IFNb as early as 1 hr after bacteria inoculation allowed the mice to survive the infection ( Figure 6). These results suggest that delayed IFNb production early during the initial phase of Lm infection may provide a selective advantage for the bacterium because the innate immune system is not properly activated and/or regulated. Our findings are consistent with those of previous studies showing that cytokine delays indeed affect the outcome of both innate and adaptive immune responses [62,63,64]. We propose that the presence of an early source of exogenous IFNb exert an important immuno-regulatory role in that it enhances cellular recruitment (Tip-DC), activation/ regulation (NK cells) whereas inhibits other cellular types that could cause excessive tissue damage [65,66].
The large amount of IFNb induced by a lethal dose of Lm at 24 h p.i. when the inflammatory responses are fully activated and the bacterial burden has probably exceeded above the threshold level for survival, the effect of IFNb on cellular apoptosis may conceal the beneficial effect of IFNb induction. For this reason, we believe that our data are not in contradiction with previous reports on detrimental effect of type I IFNs because we examined the regulatory role of exogenous IFNb at 5 h p.i. At this early time point of Lm infection the inflammatory response is not yet fully activated and the positive effect of IFNb can be appreciated. And indeed, in these conditions the mice early treated with IFNb are able to survive lethal infection.
We also examined IFNb induction in DCs infected in vitro by other Gram-positive bacteria. We found that, for example, Lactococcus lactis is able to induce IFNb production with a peak at 8 hr p.i. rather than at the 24 hr p.i. time point, as occurs in Lminfected DCs ( Figure S5). This data suggested that IFNb production is differentially modulated in DCs during infections by different Gram-positive bacteria.
Mechanistically, we found that Lm-infected DCs were not able to activate NK cells either in vitro or in vivo, both in terms of inducing IFNc production and in stimulating the lysis of target cells, suggesting that innate immunity mediated by NK cells is unpaired at 5 hr p.i. in our model. NK cell activation could be reconstituted by adding IFNb in the co-cultured system, suggesting that IFNb is indeed required for early NK cell activation. Nevertheless, bacterial counts in the spleen of infected mice were only moderately reduced indicating that other mechanisms were operating in our model. Recent reports suggest that NK cells and a cell population that is CD11b+Ly6G+ exhibit tissue protective properties in the context of innate immunity, a function that could be enhanced by the presence of IFNb [67,68].
Our results imply that IFNb maybe able to induce disease tolerance by inducing tissue protection instead of disease resistance by reducing bacterial burden, a mechanism that has been recently suggested by Medzhitov et al. [27]. The activation of the above mechanism should explain the observed increase in mice survival in the group of IFNb treated animals. However, the exact mechanisms remain unknown and it is the focus of our ongoing studies.
In conclusion, the innate immune response is a complex interplay between cells and soluble factors that, in the right context and environment, together provide a framework to combat infection. However, successful pathogens have evolved mechanisms that subvert these ancient controls, allowing them to counteract the innate immune responses of their hosts.
Here we suggest that a transient impaired of NK cell activation/ regulation may be an additional strategy employed by Lm to avoid innate immune activation and that IFNb can regulate antibacterial activity in the very early phases of acute infection acting as a positive regulatory molecule.

Ethics Statement
All animal experiments were performed using protocols approved by University of Milano-Bicocca Animal Care and Use Committee. All experimental procedures were carried out in

Mice and Reagents
All mice were bred on a C57BL/6 background and animals were housed under pathogen-free conditions. C57BL/6 mice were purchased from Charles River and were maintained in our animal facility at the University of Milano-Bicocca.
Monoclonal antibodies for FACS analysis, cell purification and intracellular staining were purchased from Becton and Dickinson (Franklin Lakes, NJ, USA). Recombinant mouse IFNb and IFNa and the IFNb mouse ELISA kit were obtained from PBL Medical Laboratories (Piscataway, NJ, USA). The BD OptEIA mouse IFNc ELISA kit was used to test IFNc in co-culture supernatants. YAC-1 cells were purchased from ATCC (American Type Culture Collection, Rockville, MD).

Microarray Assay and Analysis
We harvested 10 7 D1 cells in the immature state or after 4 h, 8 h, 12 h or 24 h of stimulation. Total RNA was isolated with Trizol Reagent (Invitrogen, Life Technologies, Karlruhe, Germany) and purified on a Qiagen RNeasy column (Qiagen, Hilden, Germany) to remove small fragments. RNA quality was assessed on an Agilent 2100 Bioanalyzer RNA 6000 Nano LabChip (Agilent Technologies, Palo Alto, CA). Only samples with intact total RNA profiles (retention of both ribosomal bands and the   broad central peak of mRNA) were used for the microarray and quantitative RT-PCR gene expression analyses. In vitro transcription (IVT) products were generated and oligonucleotide array hybridization and scanning were carried out according to the instructions supplied by Affymetrix (Santa Clara, CA). We used 10 to 16 mg of total RNA from each sample and T7-linked oligo-dT primers for first-strand cDNA synthesis. The fragmented biotinylated cDNA (15 mg) was hybridized onto the MOE430A GeneChip (Affymetrix), using the recommended procedures for prehybridization, hybridization, washing and staining with streptavidin-phycoerythrin (SAPE). Array images were analyzed with the RMA algorithm [69]. Samples displaying a signal ratio .3.0 for the b-actin and GAPDH probe sets were considered to be poor-quality targets and were excluded from the dataset. Genespring (Silicon Genetics) microarray analysis software was used for further analysis.
Microarray data were deposited in the ArrayExpress database (E-MEXP-3159 and E-MEXP-3158) and followed MIAME requirements.

RNA Preparation and PCR
RNA was isolated from tissue or cells using TRIzol reagents (Invitrogen) and extracted on an RNeasy mini kit column (RNA isolated from the spleen) or micro kit column (RNA isolated from purified cells) (Qiagen, Valencia, CA). RNA was then quantified and used in a reverse transcriptase (RT) reaction using the high capacity cDNA reverse transcription kit (Applied Biosystem). cDNA was used for PCR amplification.
Relative levels of RNA were determined by quantitative PCR on a real time detection system (7500 real time PCR system, Applied Biosystems) using a TaqMan gene expression assay for IFNb. Gene-specific transcript levels were normalized to PPIA mRNA. Amplification conditions were as follows: 50uC for 2 min, 95uC for 10 min, 40 cycles of 95uC for 15 sec, and 60uC for 1 min.

Bacteria
Listeria monocytogenes (Lm) EGD pNF8 (BUG) was used for infection studies. This strain is Lm EGD strain (BUG600) carrying the pNF8 plasmid, which contains the gene encoding the GFP protein [70]. The replicative plasmid was maintained in the strains by growing them on brain heart infusion (BHI) agar containing 5 mg/ml of erythromycin. Listeria innocua (Li) and Lm were provided by the P. Cossard's laboratory. Escherichia coli (E. coli) DH5a strain (Invitrogen). Bacteria were grown in brain heart infusion broth or LB (Sigma Aldrich, St Louis, MO, USA) to a mid-logarithmic phase and stored in small aliquots of 10% glycerol stocks at -80uC until use. Lm concentration was quantified by plating serial dilutions on BHI agar plates containing 5 mg/ml of erythromycin and counting colonies after growth at 37uC for 24-36 hours.

Infections with Bacteria
For in vitro infections with Lm, bacteria were thawed from glycerol stocks, washed in PBS, diluted into appropriate media and added to cells at a multiplicity of infection (MOI) of 20 or 40, as indicated. The Lm MOI used kept cell mortality between 10-20%. Bacterial numbers were confirmed by plating serial dilutions onto BHI agar plates containing 5 mg/ml of erythromycin. Infection was allowed to proceed for 1 h at which time extracellular bacteria were washed away, and gentamicin-containing medium (final concentration 50 mg/ml) was added to prevent extracellular bacterial growth. After another 60 min, medium was changed to medium containing 10 mg/ml gentamicin.
For intravenous infection of Lm, bacteria were thawed from glycerol stocks, washed and diluted in PBS before injection (200 ml) into the lateral vein of the tail. The infection dose was checked by plating serial dilutions of agar plates containing 5 mg/ ml of erythromycin and counting colonies after growth at 37uC for 24-36 hours. Mice were injected with sub-lethal doses (5610 4 210 5 CFU) or lethal doses (10 6 CFU). Escherichia coli (E. coli) was grown on LB and added to cells at a MOI of 20. Mice were injected with 10 7 E. coli. For determination of bacterial load, mice were killed 5 hours post infection, and spleens were homogenized in 0.2 % Triton-X100 solution. Serial dilutions of homogenates were plated on BHI agar plates containing 5 mg/ml of erythromycin, and colonies counted after growth at 37uC for 24-36 hours.

NK Cell Purification
NK cells were purified from C57BL/6 mice by positive selection from splenocytes. After lysis of red blood cells, splenocytes were stained with biotinylated anti-pan-NK cell (DX5) antibody (10 mg/ml), washed and incubated with streptavidin Microbeads (Miltenyi Biotech, Bergish Gladbach, Germany). Cells were positively selected using MS columns, according to the manu-

Purification of Splenic Dendritic Cells
Dendritic cells were purified from C57BL/6 mice by positive selection from splenocytes. Spleens were treated with collagenase before purification. After red blood cell lysis, splenocytes were incubated with CD11c + Microbeads (Miltenyi Biotec.) and positively selected using MS columns, according to the manufacturer's recommendations. Purified cells were stained with CD11c-PE and anti-MHC class II-PE to check the efficiency of purification. Purified cells were usually 90% CD11c + and 95% MHC class II. Both positive (CD11c + ) and negative fractions (CD11c 2 ) were used for PCR studies.

NK-DCs Co-cultures
Co-culture experiments were performed with NK cells derived from C57BL/6 mice. Wild-type BMDCs were resuspended in IMDM complete medium containing 10% GM-CSF supernatant without antibiotics and plated in 96-well plates (10 5 cells/well). DCs were treated with Lm or E.coli DH5a with an MOI of 20 for 1 hour and then supplemented with IMDM complete medium containing 10% GM-CSF supernatant with gentamycin (50 mg/ ml), penicillin (100 U/ml), and streptomycin (100 mg/ml). Activated BMDCs were cultured with rIFNb where indicated (100 U/ ml). NK cells were added (10 5 cells/well) directly to the culture after 30 minutes and supernatants were tested for IFNc production 18 hours later.

In vivo Activation of NK Cells
Wild-type mice were injected intravenously (iv) with 10 6 Lm with or without subsequent injection of IFNb (38000U). Spleens were removed after 5 hours and analyzed for NK cell activation. For intracellular staining, single cell suspensions were prepared and incubated with brefeldin A (10 mg/ml, Sigma), ionomycin (100 ng/ml, Sigma), and phorbol 12-myristate 13-acetate (50 ng/ ml, Sigma) for 3 hours. Cells were then stained with FITC-labeled anti-IFNc monoclonal antibody. NK cells were identified by staining with PE-labeled monoclonal anti-CD49b (clone HMa2) antibodies. Intracellular staining for IFNc was performed using the BD kit according to the manufacturer recommendations (Becton Dickinson, San Diego, CA, USA). Cells were then analyzed on a FACScan (Becton Dickinson, San Diego, CA, USA). The ability of NK cells to kill YAC-1 target cells was evaluated in a DELFIA cytotoxicity assay, as described below. Mice were injected iv with 10 6 Lm with or without subsequent injection of IFNb (38000U). Spleens were removed after 5 hours and splenocytes were cocultured with YAC-1 cells. Cytotoxicity was measured after three hours using a time-resolved fluorometer (Victor, Perkin Elmer).

Statistical Analysis
All experiments were repeated at least three times. Asterisks(*) in the Figures indicate differences deemed significant (P,0.01) by a two-tailed Student's T test. All error bars in graphs indicate SE for three samples per experimental group.