Route of Infection Determines the Impact of Type I Interferons on Innate Immunity to Listeria monocytogenes

Listeria monocytogenes is a food-borne pathogen which causes mild to life threatening disease in humans. Ingestion of contaminated food delivers the pathogen to the gastrointestinal tract, where it crosses the epithelial barrier and spreads to internal organs. Type I interferons (IFN-I) are produced during infection and decrease host resistance after systemic delivery of L. monocytogenes. Here we show that mice benefit from IFN-I production following infection with L. monocytogenes via the gastrointestinal route. Intragastric infection lead to increased lethality of IFN-I receptor chain 1-deficient (Ifnar1−/−) animals and to higher bacterial numbers in liver and spleen. Compared to infection from the peritoneum, bacteria infecting via the intestinal tract localized more often to periportal and pericentral regions of the liver and less frequently to the margins of liver lobes. Vigorous replication of intestine-borne L. monocytogenes in the livers of Ifnar1−/− mice 48 h post infection was accompanied by the formation of large inflammatory infiltrates in this organ and massive death of surrounding hepatocytes. This was not observed in Ifnar1−/− mice after intraperitoneal infection. The inflammatory response to infection is shaped by alterations in splenic cytokine production, particularly IFNγ, which differs after intragastric versus intraperitoneal infection. Taken together, our data suggest that the adverse or beneficial role of a cytokine may vary with the route of infection and that IFN-I are not harmful when infection with L. monocytogenes occurs via the natural route.


Introduction
Listeria monocytogenes (Lm) is a food-borne pathogen causing potentially life threatening disease in immunocompromised humans [1]. Lm's ability to traverse epithelial barriers including the intestinal epithelium, the placental and the blood brain barrier is a prerequisite for systemic dissemination and its sequels such as sepsis, fetal abortion and encephalitis. The natural route of infection is via ingestion of contaminated food and although the course of infection is in most cases very mild, manifested disease has one of the highest mortality rates among food-borne diseases. Lm uses a set of virulence factors to successfully invade hosts. These include internalins that allow attaching to, and subsequently invade various different cell types. Internalins (Inl) A and B are instrumental for the invasion of epithelial cells and hepatocytes [2].
Type I interferons (IFN-I) are cytokines essential for the establishment of innate antiviral immunity. Their role in inflammatory disease or bacterial infections varies between protective or detrimental, depending on the pathogen and conditions of infection [3,4]. Mice lacking the IFN-I receptor show increased resistance to intraperitoneal or intravenous infection with Lm compared to wildtype (Wt) mice [5,6,7]. Therefore, IFN-I are thought to decrease the ability of mice to combat infection. The harmful effects of IFN-I have been assigned to different aspects of the immune response. The death of effector cells such as T-cells and macrophages increases with IFN-I signalling during Lm infection [7,8]. As a consequence of the increased uptake of apoptotic cells interleukin 10 (IL10) is produced which hampers protective immune responses [9]. In line with this, IL102/2 mice are more resistant to Lm infection than Wt mice [10]. In addition, IFN-I may also interfere with IFNc dependent macrophage activation by decreasing cell surface expression of the IFNc receptor [11].
The data describing the adverse effect of IFN-I on the course of Lm infection stem from systemic infection models, mostly intravenous or intraperitoneal infection. Immune reactions to gastrointestinal Lm infections, the natural route of infection, have not been studied intensively due to the limitations of the murine system, namely low infection efficiency and incompatibility of cell surface molecules required for invasion. Specifically, murine Ecadherin on epithelial surfaces interacts poorly with the bacterial Internalin A (InlA), an invasion protein located at the surface of Listeria [12,13].
Here we show for the first time how IFN-I affects infection with Lm via the natural route. Our experiments were carried out with a mouse-adapted Lm strain, which allows robust and efficient infection of mice via intragastric gavage. We clearly demonstrate that the route of infection matters for the impact of IFN-I on innate resistance, as mice benefit from IFN-I after intragastric infection. Differences in splenic cytokine production, inflammatory cell recruitment and hepatotoxicity are suggested to underlie the infection route-dependent impact of IFN-I on the outcome of infection with Lm.

Materials and Methods
Mice, bacteria C57BL/6N (Wt) and Ifnar12/2 (B6.129P2-IfnaR1 tm1 [14]) mice were housed under SPF conditions. Ifnar12/2 mice used in this study were obtained after more than ten backcrosses to pure C57BL/6N. Animal experiments were discussed and approved by the University of Veterinary Medicine Vienna institutional ethics committee and carried out in accordance with protocols approved by the Austrian law (BMWF-68.205/0204-C/GT/2007; BMWF-68.205/0210-II/10b/2009, BMWF-68.205/0243-II/3b/2011). A scoring-system for assessment of distress of the animals used was established before infection experiments were started. On the basis of these guidelines general condition and behavior of the animals was controlled by well-educated and trained staff (participants of FELASA B training courses). Depending on the progress of the disease, animals were monitored every 3-4 hours during the ''dayphase'' (7:00 am to 7:00 pm). In order not to disturb the circadian rhythm of the animals, there was no monitoring after 7:00 pm. Humane endpoint by cervical dislocation was conducted if death of the animals during the following hours was to be expected. Animal husbandry and experimentation was performed under the national laws (Federal Ministry for Science and Research, Vienna, Austria) and ethics committees of the Medical University of Vienna and the University of Veterinary Medicine Vienna and according to the guidelines of FELASA which match that of ARRIVE.
As a prerequisite for constructing the mouse-adapted Listeria monocytogenes LO28InlA S192N/Y369S strain we prepared an InlA knockout in LO28wt using the pMAD vector [15] and the following primers for amplification of the upstream and downstream region of the InlA gene from genomic LO28 DNA : InlA_A  59 CAT GGT CGA CGG CAG TCC GCG ATT TAA TGG  AAG T 39, InlA_B 59 CAT GGG ATC CCC TAA TCT ATC  CGC CTG AAG CGT TGT 39 InlA_C 59 CAT GGG ATC  CGG GAA TTC AGC CAG CAC AAC AAG T 39 and InlA_D  59 CTG CCA TGG AGG TTT AGG TGC AGT TAT CCG  CGT 39. For genomic integration we used the protocol described in [16]. We transformed the LO28InlA knockout strain with the pAUL-A InlA S192N/Y369S -InlB construct kindly provided by WD Schubert (Molecular Host-Pathogen Interactions, Division of Structural Biology, Helmholtz Centre for Infection Research, Braunschweig, Germany [17]) and obtained genomic integration as described in [18]. Here the abbreviation Lm for experiments always refers to LO28InlA*.

Infection of mice, determination of bacterial organ loads
Bacteria were prepared for infection as described previously [19]. For infection, Lm LO28InlA* were washed, diluted to the respective concentration with PBS (Sigma) and injected intraperitoneally (i.p.) or intravenously (i.v.) into 8-to 10-week-old, sex and age matched C57BL/6N (Wt) and Ifnar12/2 (B6.129P2-IfnaR1 tm1 [14]) mice. For intragastric infection mice were infected as described [20]. Briefly, mice were starved over night and orally gavaged with CaCO 3 in 200 ml PBS (50 mg/ml) succeeded by the respective dose of Lm in 200 ml of PBS. The infectious dose was controlled by plating serial dilutions on Oxford agar plates. The survival of mice was monitored for 10 days and data were displayed as Kaplan-Meier plots. For determination of bacterial loads (colony forming units, CFU) of livers, spleens, mesenteric lymph nodes (MLN) and Peyer's patches (PP), mice were sacrificed at the indicated time points, organs isolated and homogenized in PBS. Serial dilutions of the homogenates were plated on BHI plates and incubated at 37uC for 24 h.

RNA isolation, cDNA synthesis and RT-qPCR
Organs of respective animals were isolated and stored at 280uC until further use. For RNA preparation organs were homogenized in RA1 buffer of the NucleoSpin II RNA isolation kit (Macherey and Nagel) and processed according to protocol. cDNA was prepared as described [21]. The RT-qPCRs were run on an Eppendorf cycler. mRNAs were normalized to the GAPDH housekeeping control and values obtained from infected Wt mice set to 1. Data from Ifnar12/2 mice were obtained and the ratio to the value obtained to respective samples from the Wt controls was calculated. Values in the figures for Ifnar12/2mice thus represent the relative induction compared to the value obtained at the same time point from Wt animals. Sequences of primers are listed in table I.

Analysis of IFNc and alanine aminotransferase levels (ALT) in mouse serum
For cytokine analysis mice were bled via the retro-orbital sinuses and serum was prepared and stored at 280uC. IFNc was measured using the FlowCytomix kit (eBioscience) in 25 ml of serum. ALT concentrations were measured in mouse serum using a COBASc11 analyzer (Roche). Cell culture Bone marrow derived cells were isolated and grown as described [19]. Briefly, bone marrow was isolated from femurs of 6-8 week old mice. For differentiation of bone marrow derived macrophages, cells were grown in DMEM (Gibco, Invitrogen) in the presence of 10% FCS (Gibco, Invitrogen) and L-cell derived CSF-1 as described [19]. The cultures contained .99% F4/80+ cells. Dendritic cells (mDC) were obtained by culture of mouse bone marrow cells in DMEM (Gibco, Invitrogen), supplemented with 10% FCS (Gibco, Invitrogen) and X-6310 derived GM-CSF as described [19]. mDC cultures contained virtually no F4/80+ cells and the purity of CD11c+/CD11b+ cells was between 60 and 70%. Cells were infected with Lm at a MOI of 10 for 1 h, extensively washed and resuspended in PBS for injection into mice.

Histology
Mouse organs were fixed with 4% paraformaldehyde over night, paraffin embedded and 3 mm sections were prepared using a microtome. Hematoxyline and eosin staining (H&E) were performed using standard protocols. For TdT-mediated dUTP nick end labelling (TUNEL), sections of liver and spleens were stained as described before [22]. For Gr-1 staining liver sections were blocked for endogenous peroxidase activity in 50% methanol with 3% H 2 O 2 and boiled for 30 min in 10 mM sodium citrate antigen unmasking solution. After cooling down for 30 min, the samples were blocked with 5% normal goat serum to reduce background staining. Primary Gr-1 antibody (BD Pharmingen) was applied overnight at 4uC and before adding AEC+ high sensitivity chromogen substrate (Dako) the sections were incubated with biotinylated rabbit anti-rat IgG (1:250 in PBS) for 30 min at RT following incubation with ABC reagent (Vector) for 30 min at RT. To stain Lm in infected tissue, sections were blocked for endogenous peroxidase activity in 50% methanol with 3% H 2 O 2 and incubated with 500 mg/ml pronase (Roche) for 10 min at 37uC, washed with PBS containing 0,05% Tween (PBS-T) and blocked with 5% normal goat serum for 30 min at RT. Next, primary Listeria antibody (1:100) (Abcam) was applied for 1 h at RT, washed with PBS-T following incubation with HRP rabbit/ mouse polymer (Dako) for 30 min at RT. After washing, Lm was visualized by AEC+ high sensitivity chromogen substrate and the cells counterstained with haematoxylin.

Flow cytometry
For isolation of non-parenchymal cells, mice were sacrificed and the liver was perfused immediately with liver perfusion medium (Gibco, Invitrogen) via the portal vein for 3 min at a speed of 8 ml/min followed by liver digest medium (Gibco, Invitrogen) for 5 min. The liver was removed, minced with scissors in 15 ml icecold DMEM+10%FCS+penicillin/streptomycin and filtered through a 70 mm cell strainer. To collect the non-parenchymal cell-enriched supernatant the suspension was centrifuged for 5 min at 50 g. Next, non-parenchymal cells were harvested by centrifugation for 5 min at 300 g and washed. After red blood cell lysis the cells were blocked with anti-mouse CD16/32 and stained for Ly6G (BD-Pharmingen), CD11b, Ly6C, F4/80, CD3, CD45 and with eFluor fixable viability dye (eBioscience). Flow cytometry was performed using BD FACSAria.

Statistical analysis
Bacterial loads of organs were compared using the Mann-Whitney test and indicate the median. mRNA expression data, serum protein levels and histological data were analysed with the Students t Test and indicate the mean and SEM. For both the GraphPad Prism (Graphpad) was used. Asterisks describe statistical significance as follows: *p#0,05; **p#0,01; ***p#0,001

IFN-I signalling increases resistance to Listeria monocytogenes contracted via the intragastric route
To assess whether the impact of IFN-I on innate immunity to Lm is determined by the infection route, we administered Lm to Wt and Ifnar12/2 mice both by intragastric gavage (i.g.) and intraperitoneal injection (i.p.). These experiments were performed with a mouse-adapted, mutated LO28 strain of Lm, LO28InlA* (see materials and methods). This strain expresses an InlA mutant with increased affinity to murine E-cadherin [17]. Compared to LO28 Wt, intragastric administration of LO28InlA* produced a roughly 5-fold higher bacterial burden in the liver and about 3-fold higher burden in spleen 48 h after infection (Fig. S1). This difference is similar to or larger than that reported for the mouseadapted EGD strain [17,23]. By contrast, infection rates were virtually indistinguishable 48 h after i.p. infection (Fig. S1).
In keeping with previous reports, a drastic difference in the bacterial load was observed in livers and spleens of Wt and Ifnar12/2 mice 72 h after i.p. infection (Figs. 1A, B). When mice were infected i.g., bacterial loads at day three were similar or even slightly elevated in Ifnar12/2 livers and spleens (Figs. 1C, D). To monitor survival we infected Ifnar12/2 and control mice i.g. with a high dose of Lm. IFN-I signalling protected mice from lethal infection with Lm, as Ifnar12/2 mice were the only animals to succumb to infection (Fig. 1E). This is in striking contrast to the results reported after i.p. or intravenous (i.v.) infections [7].
The bacterial burden of Lm-infected animals is routinely assessed at day three after inoculation. However, the organ loads shown in figure  The detrimental or beneficial effect of IFN-I is not determined by the size of the L. monocytogenes inoculum I.p. or i.v. injections of bacteria deliver the entire inoculum directly to internal organs. By contrast, i.g. infection might result in a more gradual release of bacteria from the intestinal tract, hence a lower primary infectious dose for internal organs. To examine whether the amount of the primary inoculum delivered to target organs influences the effect of IFN-I, we infected Ifnar12/2 and control animals with 10 2 Lm i.v. Figures 3A and B show that at this very low dose Ifnar12/2 mice showed an increased ability to prevent Lm replication, much like mice infected with 100-fold more bacteria. Therefore, parenteral delivery of small inocula does not reproduce the IFN-I effect seen after gastrointestinal administration of Lm. Alternatively, the difference between i.p. and i.g. infection might arise from the mode of dissemination. According to previous reports (e.g. [24]), Lm traversing the intestinal epithelium are taken up and spread throughout the host organism via phagocytic cells residing in mucosal lymphoid organs such as macrophages or DC.
We tried to mimic this situation by injecting various amounts of in vitro Lm-infected macrophages or dendritic cells i.v. into Wt and Ifnar12/2 mice. 40-50% of these cells harboured bacteria (data not shown). Figures 3C and 3D demonstrate that intracellular delivery of Lm resulted in enhanced bacterial clearance by Ifnar12/2 compared to Wt mice. While these experimental conditions do not faithfully mimic the host-pathogen interactions in the intestinal tract, they show that intracellular dissemination per se does not alter the impact of IFN-I on innate resistance to Lm. The experiment shown in figure 3C and 3D was performed in an identical manner with infected Ifnar12/2 cells and the outcome was highly similar (data not shown). This suggests that under our experimental conditions the response of the infected cells to IFN-I does not contribute to the overall effect of the cytokines on innate resistance to Lm.

IFN-I do not inhibit invasion of the gut mucosa or of mucosa associated lymphoid tissue
We tested the hypothesis that the beneficial effect of IFN-I on i.g.-infected mice might result from a decreased rate of intestinal invasion. This is suggested by older reports that epithelial cells treated with IFN-I show increased resistance against invasion by enteropathogens [25,26]. To monitor the presence of Lm in intestinal tissue or the gut-associated lymphoid tissue we performed immunohistochemistry on the intestinal mucosa 48 h after i.g. infection. Visualization of Lm with a specific antiserum demonstrated the presence of bacteria in mucosal tissue. Very low numbers of Lm were found in epithelial cells, the vast majority had crossed the epithelial barrier to reside in the underlying mucosa (Fig. 4A). No differences between Wt and Ifnar12/2 mice were noted (Fig. 4A, upper panels)  Absence of a type I IFN response exacerbates inflammatory pathology in livers of mice infected via the gastrointestinal route Analysis of the bacterial organ burden during the course of infection demonstrated a pronounced peak of multiplication of Lm in the livers of Ifnar12/2 mice infected i.g. for 48 h. Mice begin to die shortly after this period. Furthermore, our recent demonstration that liver damage is closely correlated to the lethality of infection [22] suggests that IFN-I reduce the severity of liver pathology in i.g.-infected mice. Livers were therefore subjected to histological, immunohistochemical and flow cytometric examination.
To assess whether the route of infection alters the predominant localization of Lm, livers 48 h after i.g. or i.p. infection were subjected to immunohistochemistry with anti-Listeria serum. The vast majority of bacteria colocalized with inflammatory infiltrates (Fig. 5A). Quantitative evaluation revealed a larger fraction of Lm in periportal or pericentral areas after i.g. infection, whereas Lm in i.p.-infected mice showed increased localization at the margins of liver lobes (Fig. 5A).
In accordance with the bacterial loads determined in figure 2, H&E staining of livers after i.g. infection demonstrated an increased number of small inflammatory infiltrates 24 h after infection in Wt compared to Ifnar12/2 livers (Fig. 5B). By contrast, more and much larger infiltrates were observed in Ifnar12/2 mice compared to their Wt controls both 48 h and 72 h after infection. Livers of Ifnar12/2 mice infected i.p. contained very few large-sized inflammatory infiltrates at 48 h. Consistent with bacterial replication (Fig. 2C) the number and size of infiltrates increased in both Wt and Ifnar12/2 livers after 72 h (Fig. 5B). Closer inspection of the inflammatory infiltrate of Ifnar12/2 mice infected for 48 h (Fig. 5C) and 72 h (Fig. 5D) via the gastrointestinal route showed that they contained a large number of Gr1 + cells (neutrophils and inflammatory monocytes) and that they were the predominant sites of infection (Figs. 5C, D,  right). Strikingly, the infiltrates as well as the surrounding hepatic tissue contained a large number of TUNEL-positive dying or dead cells (Fig. 5 C, D, left), which is consistent with reports that Lm kills infected hepatocytes, macrophages, or dendritic cells [27]. To verify that IFNAR deficiency increases hepatotoxicity after i.g. infection with Lm, serum alanine aminotransferase (ALT) levels were determined. In keeping with our hypotheses, increased serum ALT was found in Ifnar12/2 animals (Fig. 5E).
We proceeded to determine the innate effector cell populations present in the fraction of non-parenchymal liver cells (NPC) 48 h after i.p. or i.g. infection. Ifnar12/2 livers contained a somewhat smaller fraction of neutrophils after i.p. injection of Lm and the difference was larger after i.g. infection (Fig. 6A). Ifnar12/2 livers from i.g.-infected mice also contained a slightly smaller proportion of F4/80+ macrophages that include a large fraction of the resident Kupffer cells (Fig. 6B). Notably, i.g. infection of Ifnar12/2 mice caused a slight but significant reduction of the fraction of inflammatory monocytes compared to Wt, whereas this cell type formed an equal percentage of CD45+ cells in both genotypes following infection through the peritoneum (Fig. 6C). The proportion of CD3+ T cells remained largely unaffected by both route of infection and mouse genotype.
In mice infected through systemic routes, IFN-I strongly enhance the death of splenic T lymphocytes. To assess whether this activity differs after gastrointestinal infection, in situ TUNEL staining of infected splenic tissue was performed 48 h post infection (right panels of Fig. 7) and further compared to staining of serial sections with anti-Lm antibody (left panels of Fig. 7). Consistent with published data [6,22] the absence of Ifnar1 strongly reduced the number of apoptotic cells in extra-follicular areas of the white pulp after i.p. infection (upper right panels of Fig. 7). Spleens after i.g. infection contained less apoptotic cells in correlation with roughly 10-fold less bacteria (lower right panels of    Fig. 7, Fig. 2). In spite of an increase in bacterial numbers relative to Wt controls (lower left panel of Fig. 7, Fig. 2), Ifnar12/2 spleens contained fewer TUNEL+ cells. Thus, the inhibitory activity of IFN-I on splenocyte death appears to be independent of the infection route.

IFN-I accelerate and increase proinflammatory cytokine activity after intragastric Listeria monocytogenes infection
Our analysis of liver inflammation suggests it contributes to, or reflects the different impact of IFN-I on mice infected through enteral or parenteral routes. However, production of cytokines that regulate inflammation and immunity is to a large extent an attribute of leukocytes residing in the blood or in lymphoid organs. Therefore, we determined cytokine mRNA expression in the spleen.
Splenic IFNb mRNA was expressed up to 72 h after i.g. application of Lm. In accordance with expectations, the IFN-I induced gene Mx2 was expressed in infected spleens up to 72 h in Wt, but was strongly reduced in Ifnar12/2 mice (Fig. 8A). Among the pro-inflammatory cytokines and chemokines tested, IL6 and MCP1 were decreased early after infection. Notably, this effect was not observed at a similar time after i.p. infection. Of further interest, early splenic IFNc production was increased in Ifnar12/2 mice compared to Wt after i.p. infection, but not after infection through the gastrointestinal tract. Early TNFa production was reduced in Ifnar12/2 mice infected through the intestinal tract and this was more pronounced after i.p. infection. Compared to Wt mice, immunosuppressive IL10 was elevated in Ifnar12/2 72 h after i.g. infection, but reduced after i.p. infection. The data suggest that splenic cytokine production is an important determinant of the different impact of IFN-I after enteral and parenteral infection routes. Due to the overwhelming importance of IFNc for innate resistance to Lm [28,29], delayed production of this cytokine in i.g.-infected versus i.p.-infected Ifnar12/2 mice is likely a major determinant of the difference in innate resistance. It reflects the detrimental increase of hepatic Lm between 24 h and 48 h after i.g. infection. To determine whether systemic IFNc production corresponds to splenic mRNA, serum IFNc was measured and found to correlate well with the levels of IFNc mRNA in the spleen (Fig. 8B).

Discussion
Lm is a widely studied pathogen. Mouse infection models have been helpful in deciphering innate responses to intracellular bacteria. However, the vast majority of data about immunity to Lm stems from animals infected via the intraperitoneal or intravenous route. Comparably little knowledge has been obtained about infection through the intestinal tract which represents the natural entry route in humans. The incompatibility of the InlA/Ecadherin interaction in murine hosts, required for efficient invasion of epithelial cells, posed an obstacle to murine models of gastrointestinal infection. To overcome this limitation, we followed the approach of Wollert et al. [17] to 'murinize' the InlA gene for improved interaction with mouse E-cadherin and increased invasion of Lm. The Lm strain LO28 was chosen for this approach because it is a particularly potent inducer of IFN-I in vitro and in vivo [30]. E-cadherin is located at the basolateral side of epithelial cells, making it less accessible to bacteria in the gut lumen. Pentecost and colleagues have shown that at sites of epithelial cell turnover E-cadherin becomes accessible to Lm [31]. InlB, another Lm invasin, contributes to the initial uptake into host epithelia [32,33]. Finally, the major virulence factor Listeriolysin O (LLO), a member of the bacterial hemolysin family, promotes the uptake of Lm into hepatocytes [27].
The results of our study clearly show that the route of uptake, hence the initial interactions of Lm with the host, determine the impact of IFN-I on the innate antibacterial response. Strikingly, the impact of IFN-I synthesis changes from being adverse [5,6,7] to being beneficial when i.p. and i.g. routes of infection are compared. Both high-dose and low-dose systemic infections produced an adverse effect of IFN-I, suggesting that the switch to beneficial IFN-I action is not caused by a slow release of Lm from the intestine. Likewise, systemic administration of Lm as a passenger of bone marrow-derived macrophage or DC effector cells did not switch the effect of IFN-I. This suggests that dissemination in infected cells does not per se change the impact of IFN. In spite of this we cannot rule out the possibility that dissemination via a cell residing in the gut-associated lymphoid tissue or in the intestinal mucosa would produce a different outcome.
Mechanisms proposed to explain the detrimental actions of IFN-I signalling after i.p. infection are increased T lymphocyte and macrophage death and unresponsiveness of macrophages to IFNc [34,35]. Following systemic delivery, IFN-I increase production of immune-suppressive IL10 and restrain TNFa producing cells in the spleen, thereby limiting a protective inflammatory response [5,9]. Here we show that the production of critical cytokines is altered in absence of type I IFN, either because they are direct targets of IFN-I signalling or because the lack of IFN-I signalling misbalances complex cascades of events that include proinflammatory cytokine synthesis. Among affected cytokines are the protective TNFa, IL6 and IFNc and the antiinflammatory IL10, in addition to chemokines such as MCP1/ CCL2 that regulate the recruitment of myeloid cells [10,29,36,37,38]. Unlike early TNFa synthesis that requires IFN-I after both i.p. and i.g. infection, the production of IL6 and MCP1 during the initial 24 h of infection is reduced specifically when bacteria enter their host via the intestine. Most importantly, increased IFNc production in absence of type I IFN responsiveness was delayed and less pronounced after i.g. compared to i.p. infection. Thus, reduced synthesis of protective cytokines during the early phase of infection provides a likely explanation for the different impact of IFN-I on mice infected via enteral or parenteral routes. This explanation is in accordance with the timing of bacterial replication, which is strongly accelerated in i.g.-infected Ifnar12/2 mice between 24 h and 48 h. The increased IFNc synthesis at 48 h explains why a significant fraction of Ifnar12/2 mice proceeds to clear bacteria with equal efficiency as their Wt counterparts and survives. We hypothesize that those Ifnar12/2 mice that die from i.g. infection are unable to cope with the damage inflicted by the large Lm burden between 24 h and 48 h. Unexpectedly, intestinal invasion and replication in the intestinal mucosa or the mucosa-associated lymphoid tissue demonstrated little control by IFN-I. In contrast, the liver, most likely the most rapidly infected internal organ, revealed striking differences comparing intragastric or intraperitoneal infection routes, most obviously the vigorous replication of Lm between 24 and 48 h and the correspondingly stronger inflammatory response. Our recent report clearly demonstrated a close correlation between the lethality of Lm infection and the extent of liver damage [22]. It appears likely, therefore, that the hepatic response is a major determinant for the beneficial or adverse effects of IFN-I on Lm infection. Contrasting the spleen, the liver is not itself a site of IFN-I synthesis during Lm infection and it responds poorly to IFN-I [19]. This supports our conclusion that the innate hepatic response is critically influenced by IFN-Iregulated splenic cytokine synthesis. Additionally, tonic signalling by the IFNAR, resulting from low levels of constitutive IFN-I production, plays an important role in priming the innate immune system for appropriately vigorous antimicrobial responses [39]. We cannot exclude the possibility that the lack of such tonic signals in Ifnar12/2 mice contributes to infection-related phenotypes of these animals. However, the lack of tonic IFNAR signalling does provide a straightforward explanation for the different impact of IFN-I after i.g and i.p. infection.
Our data further suggest that in addition to cytokines, the extent of liver inflammation may be influenced by different bacterial entry points after intestinal or intraperitoneal infection routes. Lm arriving from the intestinal tract localizes mostly to periportal or pericentral areas and the inflammatory response is directed to these regions. This suggests that important Lm entry routes are the portal blood stream or, following systemic dissemination, the central veins, with little contribution of peritoneal invasion and direct infection of liver lobes. By contrast, a larger fraction of intraperitoneally administered Lm may choose direct intraperitoneal access to the margins of liver lobules rather than systemic dissemination through blood or lymphatic vessels. Speculatively, the location of replicating Lm may influence the speed with which an inflammatory infiltrate is formed and regulated by blood-borne cytokines and chemokines.
One very obvious cause for liver damage noted in our experiments was the death of cells both within the inflammatory infiltrate and the surrounding hepatic tissue. Cell death may be a direct consequence of intracellular Lm in macrophages, neutro-phils or hepatocytes [40,41,42] or an indirect consequence of inflammation. Liver infiltrates formed more frequently in i.g.infected Ifnar12/2 mice and developed to significantly larger sizes, thus causing a much larger fraction of hepatic tissue to die. Livers of Ifnar1-deficient mice contained a smaller fraction of inflammatory monocytes, important antibacterial effector cells [43], specifically following infection via the gastrointestinal route. This may reflect the reduced synthesis of chemokines and contribute to the enhanced multiplication of Lm in the liver.
Apoptosis of splenic lymphocytes is thought to be an important cause for the adverse consequence of IFN-I production after i.p. infection. While spleens from i.g.-infected mice contained less apoptotic cells compared to i.p.-infected controls, absence of the IFNAR1 reduced the number of dying or dead splenocytes despite a higher bacterial burden. This suggests that the enhancement of splenocyte death by IFN-I is established independently of the route of infection.
Taken together our data suggest that the beneficial effect of IFN-I on the innate response to gastrointestinal infection results from ensuring the rapid upregulation of critical protective cytokines that limit hepatic bacterial replication and inflammation. Diminished restriction of bacterial growth in the liver exacerbates hepatic inflammation and damage. Within the IFN-I-regulated cytokine milieu IFNc, an important determinant of innate immunity, becomes limiting in the early phase of i.g. infection, particularly in absence of other IFN-I-regulated protective proinflammatory cytokines/chemokines. In contrast, i.p.-infected mice benefit from an early boost of IFNc production in absence of an IFN-I response. Therefore, the influence of the entry route on the kinetics of IFNc regulation by IFN-I may function as a key determinant of innate resistance. Interestingly, differences in early cytokine production after i.g. infection of Ifnar12/2 mice were more pronounced in spleen than in the gut epithelium or gutassociated lymphoid tissue. This may reflect the fact that splenic leukocytes are a more productive source of IFN-I and generate a larger influence of IFN-I on splenic cytokine/chemokine production. Alternatively, crossing the intestinal barrier may change the expression of Lm genes and alter some important aspect of the interaction with its host. This and other open questions will be addressed in future studies.