Type I interferon does not promote susceptibility to foodborne Listeria monocytogenes

Abstract

Type I IFN (IFNα/β) is thought to enhance growth of the foodborne intracellular pathogen Listeria monocytogenes (Lm) by promoting mechanisms that dampen innate immunity to infection. However, the type I IFN response has been studied primarily using methods that bypass the stomach and, therefore, fail to replicate the natural course of Lm infection. In this study, we compared i.v. and foodborne transmission of Lm in mice lacking the common type I IFN receptor (IFNAR1^−/−). Contrary to what was observed using i.v. infection, IFNAR1^−/− and wild type mice had similar bacterial burdens in the liver and spleen following foodborne infection. Splenocytes from wild type mice infected intravenously produced significantly more IFNβ than those infected by the foodborne route. Consequently, the immunosuppressive effects of type I IFN signaling, which included T cell death, increased IL-10 secretion, and repression of neutrophil recruitment to the spleen, were all observed following i.v., but not foodborne transmission of Lm. Type I IFN was also previously shown to cause a loss of responsiveness to IFNγ through down-regulation of the receptor IFNGR1 on macrophages and dendritic cells. However, we detected a decrease in surface expression of IFNGR1 even in the absence of IFNα/β signaling, suggesting that in vivo, this infection-induced phenotype is not type I IFN-dependent. These results highlight the importance of using the natural route of infection for studies of host-pathogen interactions and suggest that the detrimental effects of IFNα/β signaling on the innate immune response to Lm may be an artifact of the i.v. infection model.

Introduction

Type I IFNs (IFNα/β) are multifunctional cytokines with diverse roles in anti-viral, anti-bacterial, and anti-tumor immunity. Over a dozen IFNα subtypes as well as a single IFNβ subtype have been identified, and all of these bind the common type I IFN receptor (IFNAR), a heterodimer composed of IFNAR1 and IFNAR2 (1, 2). Ligation of IFNAR, which is expressed on a variety of immune cells, triggers Jak-Stat signaling and can affect the expression of a diverse array of downstream genes (1). Secretion of type I IFN is induced by ligation of Toll-like receptors and cytosolic sensors such as DDX41 and stimulator of interferon genes (STING) (3, 4).

Studies using mice with a functionally inactivated type I IFN receptor (IFNAR1^−/−) demonstrated that IFNα/β signaling was essential for immunity to acute viral challenge. Despite otherwise normal immune responses, these animals were unable to restrict replication of vesicular stomatitis virus, Semliki Forest virus, or vaccinia virus after a low-titer challenge (5). Type I IFN can directly limit the intracellular niche for viral replication by inducing expression of cyclin-dependent kinase inhibitors, pro-apoptotic TRAIL and FAS/FASL, and the Mx-1 gene (6, 7). These cytokines also stimulate dendritic cell maturation; upregulate expression of MHC-I, MHC-II and costimulatory molecules; and promote the production of anti-viral antibodies in B cells (8, 9). Although a robust type I IFN response is crucial for clearance of most viruses, it has also been correlated with the development of more severe disease during influenza virus infection (10, 11).

The role of type I IFN in the response to bacterial challenge is complex and appears to depend on the nature of the pathogen. Type I IFN signaling improved disease resistance in mice infected with Group B streptococci, Streptococcus pneumoniae, and E. coli (12). In contrast, type I IFNs have been characterized as detrimental to the host during infection with intracellular bacterial pathogens such as Mycobacterium tuberculosis (13), Francisella tularensis (14), Salmonella enterica (15), and Lm (16–18). Studies using IFNAR1^−/− mice suggest that IFNα/β signaling can affect several key areas of innate immunity during infection with these bacteria. During Lm infection, IFNα/β signaling inhibited macrophage responsiveness to IFNγ and sensitized T cells to apoptotic signals (18, 19). Type I IFN also limited neutrophil recruitment to the spleen during both Francisella and Listeria infection (14). Another study found a higher frequency of TNFα-producing CD11b⁺ cells in the spleens of Lm-infected IFNAR1^−/− mice (16). In each of these cases, mice that lacked type I IFN signaling resulted in infections had either reduced bacterial burdens or more rapid bacterial clearance compared to wild type mice.

Lm is transmitted to humans through the ingestion of contaminated food, but innate immunity to Lm is poorly understood because most studies have used either i.v. or i.p. inoculation, methods that result in robust systemic growth of Lm without encountering the harsh environment of the gastrointestinal tract. Some previous studies have used oral gavage to infect mice; however, this delivery method can result in more rapid spread of Lm to the blood, spleen, and liver with significant variability amongst investigators (20). All of these methods fail to replicate the natural course of infection and therefore, disparate effects on host innate immunity may be observed.

In this study, we compared foodborne transmission of Lm to i.v. infection to determine if type I IFN secretion was detrimental to the host when the bacteria were introduced by the natural route. We hypothesized that i.v. infection would trigger robust, rapid IFNβ secretion because a bolus of bacteria reached the spleen within minutes of inoculation. Conversely, during foodborne infection, Lm would asynchronously arrive at the spleen over the course of two to three days. Therefore, it was expected that foodborne transmission of Lm would not trigger a robust IFNα/β response, and that the corresponding host-detrimental effects would not be observed. As predicted, we found that mice lacking the common type I IFN receptor were not more resistant to foodborne Lm infection, and that foodborne infection triggered significantly less production of IFNβ than i.v. infection. The data presented here suggest that most of the detrimental effects attributed to type I IFN during Lm infection may actually be an artifact of the i.v. infection model.

Materials and Methods

Bacteria

Lm SD2000, an EGDe derivative expressing InlA^m, was used for all experiments in this study (21). Expression of the modified InlA^m surface protein allows the bacteria to efficiently bind murine E-cadherin, promoting invasion of the intestinal epithelium (22). Intestinally passaged Lm were grown to early stationary phase in Brain Heart Infusion (BHI) broth (Difco) shaking at 37° C (for i.v. infection) or standing at 30° C (for foodborne infection) and then aliquots were prepared and frozen at −80° C until use as described previously (23).

Mice

C57BL/6J (stock #000664) and IFNγ^−/− (stock #002287) mice were purchased from The Jackson Laboratory (Bar Harbor, ME). The IFNγ^−/− mice were crossed with Thy1.1/luciferase-expressing C57BL/6 mice (24) (originally obtained from Dr. Robert Negrin, Stanford University) to obtain homozygous Thy1.1^+/+IFNγ^−/− progeny. IFNAR1^−/− mice were obtained from Dr. Jayakrishna Ambati (University of Kentucky). Mice were housed in a specific-pathogen free facility with a 14-hour light cycle (7 am–9 pm) and a 10-hour dark cycle (9 pm–7 am). Both male and female animals were used in all experiments and the animals were between 6–12 weeks of age at the time of infection. All procedures were approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Kentucky.

Lm infection

An aliquot was thawed on ice and cultured for 1.5 h in BHI broth shaking at 37° C (i.v.) or standing at 30° C (foodborne). For i.v. infection, Lm were washed once, suspended in sterile PBS, and diluted to the appropriate concentration. A total volume of 200 μL was aseptically injected into the lateral tail vein. For foodborne infection, Lm were washed twice and suspended in 5 μL of salted butter (Kroger) plus sterile PBS (3:2 ratio) and then added to a 2–3 mm piece of white bread (Kroger). Mice were fasted for 16–24 hours prior to ingestion of the Lm-contaminated bread and housed on raised wire flooring as described previously (25, 26). Uninfected mice received a 200 μL injection of sterile PBS via the lateral tail vein. For the challenge experiment shown in Fig. 1, spleens and livers were harvested aseptically and homogenized (Fisher PowerGen 1000, 60% power) in sterile water for 30 seconds. Dilutions were prepared in sterile water and plated on BHI agar (Difco). Colonies were counted after 24 h incubation at 37°C.

FIGURE 1.

Foodborne Lm infection of IFNAR1^−/− and wild type (WT) C57BL/6J mice results in similar burdens in the liver and spleen. Mice were infected i.v. with 9.0 × 10⁴ CFU or fed 10⁹ CFU of Lm SD2000. Data from multiple experiments was pooled for a total of 8–10 mice per group except for the 5 dpi time point, which includes 3–6 mice per group. Mean values +/− SD are shown.

Isolation of splenocytes

Spleens were injected with 100 U of type IV collagenase (Worthington) in a total volume of 1 mL of HBSS (Life Technologies stock # 14025). Spleens were minced, additional collagenase was added for a final concentration of 200 U/mL, and the samples were incubated for 30 min at 37°C in 7% CO₂. The digested tissue was pushed through a sterile screen (# 80 mesh), filtered, and red blood cells were lysed using an ammonium chloride buffer. For the experiments depicted in Fig. 3A only, single-cell suspensions were obtained by mechanical dissociation of spleens through mesh screens without collagenase treatment. For CFU determination, a portion of the splenocyte single cell suspension was removed, diluted in sterile water, and plated on BHI agar.

FIGURE 3.

IFNα/β signaling does not promote T cell loss or IL-10 induction during foodborne listeriosis. C57BL/6J (WT) and IFNAR1^−/− mice were infected i.v. with 5–6 × 10⁴ CFU or fed 10⁹ CFU of Lm SD2000. Splenocytes were harvested at the indicated time points, and the total number of T cells per spleen (A) and the concentration of IL-10 present in clarified spleen homogenates (B) was determined. Pooled data from multiple experiments are shown. For panel A, values for each mouse are shown and horizontal lines indicate means for each group. For panel B, mean values +/− SD are shown; n=6–7 mice per group except WT NI, where n=4. NI, not infected.

ELISA

For detection of type I IFN, lymphocytes were depleted from splenocyte suspensions using APC-conjugated anti-B220 (RA3-6B2) and anti-TCRβ (H57-597) antibodies (eBioscience) and IMag anti-APC magnetic beads (BD Biosciences). This protocol resulted in greater than 90% depletion of both B cells and T cells for all samples. APC-enriched splenocytes were cultured for 24 h at 37° C in 7% CO₂ at a density of 1.0 × 10⁶ cells/200 μL in 96-well flat bottom plates in media [RPMI 1640 (Life Technologies stock # 21870) supplemented with 10% FBS (Gemini), 2.5 mM L-glutamine (Sigma), 10mM HEPES (Life Technologies), 0.1 mM β-mercaptoethanol (Sigma) and 12.5 μg/mL gentamicin]. Cultured cells were centrifuged at 300 × g for 8 minutes and the supernatants were harvested and stored at −80° C. Cytokine concentrations were determined using the Verikine Mouse IFNβ ELISA kit (PBL Assay Science, Piscataway, NJ) and the Mouse IFNα Platinum ELISA kit (eBioscience).

For IL-10 detection, spleens were harvested aseptically and placed into 2 mL of ice-cold PBS, homogenized for 30 seconds (Fisher PowerGen 1000, 60% power), split into aliquots, and stored at −80° C. An aliquot was thawed on ice and centrifuged at 14,000 × g for 10 min, and the supernatant was collected. ELISA was performed using anti-mouse IL-10 capture antibody (JES5-16E3), biotin-conjugated anti-mouse IL-10 detection antibody (JES5-2A5), and mouse IL-10 standard (eBioscience).

Flow cytometry

Cells were stained using fluorescently-conjugated antibodies specific for the following molecules: CD64 (X-54-517.1), purchased from Biolegend; Ly6C (AL-21) and Ly6G (1A8), purchased from BD Biosciences; F4/80 (BM8), CD11c (HL3), CD11b (M1/70), CD19 (D3), B220 (RA3-6B2), CD3 (17A2), and streptavidin, purchased from eBioscience. Biotin-conjugated IFNGR1/CD119 (2E2; Biolegend) was detected with PE Cy5-conjugated streptavidin (eBioscience) Dendritic cells were defined as CD11c^hiF4/80^−/lo, macrophages as CD11c^−/lo and F4/80^hi or CD64^hi; B cells as CD3⁻B220⁺CD19⁺; T cells as CD3⁺CD19⁻ or, for Fig. 3, B220⁻CD3⁺TCRβ⁺; and neutrophils as Ly6G^hiLy6C^intCD11b⁺. Fluorescence was measured using a LSRII Flow Cytometer (BD) and analysis was performed using FlowJo V.10 (Treestar).

Statistical analysis

Statistical analysis was performed using Prism for Macintosh Version 6.0f (GraphPad). Significance was determined using unpaired T-test unless otherwise noted. Values <0.05 were considered significant and are indicated as follows: *, P<0.05; **, P<0.01; ***, P<0.001; ****, P<0.0001.

Results

IFNα/β receptor deficiency is not beneficial during foodborne Lm infection

After i.v. inoculation of Lm, IFNα/β receptor-deficient (IFNAR1^−/−) mice resist high titer growth in the spleen and clear the infection more quickly than wild type mice (16, 17). Intravenously-injected Lm are quickly filtered from the blood by phagocytes in the spleen and liver, producing a rapid-onset infection in these organs (27). However, i.v.-infected IFNAR1^−/− and wild type mice do not show a significant difference in Lm burdens until 2 days post-infection (dpi), with maximal differences observed at 3 dpi (16). To find out if type I IFN signaling also promoted the growth of Lm after oral transmission, we used a natural feeding model of Lm infection. Groups of IFNAR1^−/− and wild type C57BL/6 mice were fed a sublethal dose of Lm or infected intravenously, and bacterial loads in the spleen and liver were compared. In agreement with previous reports, the livers from IFNAR1^−/− mice had more than 1000-fold fewer Lm compared to wild type mice three days after i.v. infection (Fig. 1). In the spleen, IFNAR1-deficient mice had 160-fold less Lm than wild type mice. In contrast, three days after foodborne infection there was no significant difference in bacterial burdens in the liver, and the spleens of IFNAR1^−/− mice had only 20-fold fewer Lm than wild type mice (Fig. 1). These results are similar to a recently published report by Kernbauer et al. who showed that i.g. infection of IFNAR1^−/− and wild type mice with Lm LO28 resulted in equivalent bacterial burdens in the spleen and liver 3 dpi (28). Even when foodborne infection was allowed to proceed for up to 5 days, there remained no significant difference between CFU counts in the liver, and differences in the spleen were 5-fold or less. Notably, foodborne infection did not result in the death of any mice during the 5 day observation period. Thus, while the lack of type I IFN signaling was clearly beneficial after i.v. Lm infection, it did not alter the progression of foodborne listeriosis.

I.v. infection induces more robust IFNβ secretion than foodborne infection

In contrast to the rapid nature of i.v. infection, ingested Lm must first survive passage through the stomach, invade the intestinal epithelium, and then gain access to the circulation via the lymphatics before reaching the spleen. Depending on the size of the inoculum, this process typically requires 24 to 48 hours after ingestion of contaminated food (29). With this in mind, we hypothesized that the delayed and presumably asynchronous exit of Lm from the gastrointestinal tract would trigger a less robust IFNα/β response than a bolus of organisms that arrived in the spleen by i.v. inoculation. Consequently, a phenotype influenced by type I IFN signaling would be more easily observed during i.v. infection than during foodborne infection.

To test this idea, splenocytes were harvested from mice 24 hours after i.v. or foodborne infection (Fig. 2A), and lymphocytes were depleted to increase the concentration of macrophages and dendritic cells in the samples. The cells were cultured overnight without further stimulation, allowing for ex vivo accumulation of secreted cytokines, and ELISA was used to quantify IFNβ. As shown in Fig. 2C, splenocytes from i.v.-infected mice produced more IFNβ than did the splenocytes from mice infected by the foodborne route. However, a considerable disparity existed in the bacterial burdens of these two groups at this time point. The i.v. infected mice averaged 10⁶ CFU in the spleen, while the orally challenged mice had less than 200 CFU per spleen (Fig. 2B).

FIGURE 2.

Robust IFNβ secretion is induced by i.v., but not foodborne Lm infection. Splenocytes were harvested from infected C57BL/6J mice and a portion was used to determine total CFU burden. B and T cells were depleted from the remaining splenocytes, and the cells were cultured overnight to allow for ex vivo accumulation of secreted IFNα/β. (A), (D), Graphical depictions of infection strategies used are shown. (A) Mice were inoculated i.v. with 5 × 10⁴ CFU or fed 10⁹ CFU of Lm SD2000 and spleens were harvested 24 h later. In panel (D), a lower i.v. dose was used (1 × 10³ CFU) and spleens were harvested 24 h after i.v. and 72 h after foodborne infection. (B), (E), Total CFU burdens per spleen are shown. (C), (F), (G), IFNβ or IFNα ELISA data are shown. Representative values from one of three (IFNβ) or two (IFNα) separate experiments using 2–3 mice per group are shown. Horizontal lines indicate mean values for each group; dashed lines indicate limits of detection. NI, not infected.

To find out if the bacterial burden or the route of transmission was the primary factor influencing IFNβ secretion, we next established an infection model that resulted in colonization of the spleen for similar periods of time with comparable bacterial loads. To accomplish this, the i.v. dose was lowered to approximately 10³ CFU and spleens were harvested 24 h after injection (Fig. 2D). Foodborne infection, however, proceeded for 72 hours to give the bacteria time to exit the G.I. tract and arrive in the spleen. This approach resulted in bacterial burdens of approximately 10⁴ CFU in all mice, and in both cases the bacteria had colonized the spleen for approximately 24 hours (Fig. 2E). Splenocytes were then harvested and cultured overnight to assess IFNβ production. Despite the similarity in bacterial burdens, splenocytes from i.v.-infected mice still produced significantly more IFNβ than cells from mice that had ingested Lm (Fig. 2F). Because the murine common type I IFN receptor binds thirteen IFNα subtypes in addition to a single IFNβ (1, 2), it was possible that IFNα secretion could also be influencing the susceptibility of wild type mice to i.v. Lm infection. To test this, the splenocyte culture supernatants were also assayed for IFNα; however, little or none was detected (Fig. 2G). Thus, as predicted, i.v. infection resulted in a more robust type I IFN response than foodborne infection, even when comparable bacterial burdens were present in the spleen. Furthermore, these results suggested that enhancement of bacterial growth by the mechanisms previously attributed to IFNα/β signaling might not occur during foodborne infection due to decreased production of type I IFN.

T cell depletion in the spleen is a consequence of i.v., but not foodborne infection

One consequence of i.v. Lm infection that has been linked to type I IFN signaling is the extensive loss of splenic lymphocytes that occurs within the first few days after infection. In wild type mice inoculated intravenously with 0.1 LD₅₀ of Lm, large numbers of splenic T cells upregulated CD69 and underwent apoptosis, with cell death peaking by 3 dpi (30). When type I IFN signaling was absent, substantially less T cell depletion was observed in the spleen (18). Since oral transmission of Lm did not induce robust IFNβ secretion, we hypothesized that there would be little T cell loss in the spleen during foodborne listeriosis. To examine the extent of T cell death, groups of wild type and IFNAR1^−/− mice were challenged either i.v. or orally and the total number of TCRβ⁺ cells in the spleen 3 dpi was compared to the cell counts of uninfected mice. In agreement with previous work, i.v. infection induced an IFNAR1-dependent loss of more than 50% of the TCRβ⁺ cells in the spleen (Fig. 3A). Following foodborne infection, however, there was no significant T cell depletion in either wild type or IFNAR1^−/− mice.

The presence of apoptotic cells and cellular debris can trigger the scavenger receptor CD36 on macrophages, causing the cells to shift from a pro-inflammatory to a regulatory state with concomitant production of IL-10 (31, 32). Because significant T cell loss was only observed during i.v. infection, we predicted that i.v., but not foodborne Lm infection would result in increased IL-10 production. To investigate this, spleens from infected wild type and IFNAR1^−/− mice were homogenized and the amount of IL-10 present was measured directly ex vivo. As expected, spleens from wild type mice contained significantly more IL-10 than spleens from IFNAR1^−/− mice three days after i.v. infection (Fig. 3B). In contrast, splenic IL-10 production did not increase significantly above the concentrations detected in uninfected mice either three or four days after foodborne infection. Together, these results suggested that the IFNβ response induced by foodborne infection was not substantial enough to trigger T cell loss and subsequent increased IL-10 secretion in the spleen.

IFNα/β signaling does not affect neutrophil recruitment following foodborne Lm infection

Henry et al. previously showed that IFNAR1-deficient mice recruited significantly higher numbers of neutrophils to the spleen following intranasal Francisella infection compared to wild type mice (14). They proposed that a robust IFNα/β response limited the early influx of neutrophils by preventing γδ T cell expansion and production of IL-17, thereby reducing the ability of wild type mice to eliminate bacteria in the early stages of infection. Based on this observation, we hypothesized that the modest IFNβ response triggered by foodborne Lm infection would not alter neutrophil recruitment to the spleen, while robust secretion of IFNβ following i.v. infection would limit this influx.

To test this, the total number of neutrophils (Ly6G^hiLy6C^intCD11b⁺) in the spleens of wild type and IFNAR1-deficient mice was determined by flow cytometry. Spleens were harvested 48 h after i.v. infection and 72 h after foodborne infection. As shown in Fig. 4, i.v. infection of IFNAR1-deficient mice resulted in a greater neutrophil influx to the spleen compared with wild type mice. Three days after foodborne infection, however, there was no difference in the number of neutrophils in the spleens of IFNAR1^−/− and wild type mice. Thus, neutrophil recruitment to the spleen was strongly enhanced by the loss of type I IFN signaling during i.v. infection, but not during foodborne infection. These data again suggested that the type I IFN response induced during foodborne listeriosis was not robust enough to alter the innate immune response necessary for early clearance of Lm.

FIGURE 4.

A lack of Type I IFN signaling enhances neutrophil recruitment to the spleen during i.v., but not foodborne infection. Mice were infected i.v. with 1 × 10⁴ CFU or fed 10⁹ CFU of Lm SD2000 and splenocytes were harvested at either 2 dpi (i.v.) or 3 dpi (food). A) The total number of neutrophils (PMN) per spleen is shown. B) Total Lm burdens per spleen are shown; dashed line indicates limit of detection. Data from multiple experiments were pooled; mean values are indicated by grey bars (A) or horizontal lines (B).

IFNGR1 expression decreases on splenic macrophages and dendritic cells during both i.v. and foodborne infection

IFNβ was also shown to negatively regulate transcription of the receptor for interferon gamma (IFNGR1), thereby limiting the ability of macrophages to respond to the presence of IFNγ (19, 33). This would presumably result in less killing of intracellular bacteria and provide a more hospitable replicative niche for the growth of Lm in vivo. Based on the differential IFNβ secretion we detected, we hypothesized that decreased surface IFNGR1 expression would be observed shortly after i.v. infection, but not following the ingestion of Lm-contaminated food.

To test this, splenocytes were harvested from mice 24 hours after i.v. infection, and the mean fluorescence intensity (MFI) of IFNGR1 on the surface of macrophages, dendritic cells, B cells, and T cells was determined directly ex vivo (Fig. 5A). Although i.v. infection with 10³ CFU resulted in robust IFNβ secretion (Fig. 2B), little to no decrease in IFNGR1 expression was observed in any of the four cell types examined (Fig. 5B). However, in agreement with previous studies, when we increased the inoculum to 10⁴ CFU, IFNGR1 expression on CD11c^hiF4/80^−/lo dendritic cells and F4/80^hiCD11c^−/lo macrophages decreased to levels that were approximately 50% of that found on uninfected cells. In contrast, B cells showed little to no change in IFNGR1 levels during high titer i.v. infection, and T cells showed an increase above uninfected levels (Fig. 5B). These results suggested that dose-dependent decreases in IFNGR1 expression occurred primarily on myeloid-derived antigen-presenting cells and that a bacterial burden of at least 10⁵ CFU in the spleen (Fig. 5C) was required to observe this effect following i.v. challenge. At 24 hours after foodborne infection, little to no decrease in IFNGR1 MFI was seen on any of the four cell types tested (Fig. 5D). This observation was not surprising, since few mice had detectable bacterial loads in the spleen 24 hpi (Fig. 5E). By 48 hpi, however, IFNGR1 MFI had decreased by approximately 50% on dendritic cells and macrophages, despite the fact that bacterial burdens in the spleen averaged only 10² CFU at this time point. By 72 hpi, the MFI of IFNGR1 on dendritic cells and macrophages had decreased to levels similar to those seen during high dose i.v. infection. However, as shown in Fig. 5D, at this time point the mice had only ~10⁴ CFU in the spleen, a bacterial burden that did not trigger changes in IFNGR1 MFI during i.v. infection (Fig. 5C). Thus, down regulation of IFNGR1 expression during foodborne infection was dependent on bacterial load in the spleen, but occurred at a significantly lower threshold than during i.v. infection.

FIGURE 5.

Lm infection results in decreased IFNGR1 expression on macrophages and dendritic cells. Mice were infected i.v. with the indicated doses and splenocytes were harvested 24 hpi (B, C) or mice were fed 10⁹ CFU, with splenocytes harvested at the indicated time points (D, E). A representative histogram depicting the shift in IFNGR1 expression on dendritic cells 24 h after i.v. infection is shown in panel (A). Mean fluorescence intensity values (MFI) for IFNGR1 were normalized to the MFI for uninfected splenocytes, which is represented by the dotted line at 1.0 (B, D). DC, dendritic cell; MΦ, macrophage. Significance was determined using one-sample t-test; asterisks indicate mean values significantly different from a hypothetical mean of 1.0. (C, E) Total Lm CFU per spleen was determined at indicated time points; dashed line indicates limit of detection.

Infection-induced decreases in IFNGR1 expression are not dependent on Type I IFN signaling

The changes in IFNGR1 levels observed during foodborne infection in the absence of robust IFNβ secretion hinted that surface expression of this receptor might not be strictly dependent on the presence of a type I IFN signal. To directly assess the role of IFNα/β signaling in down regulation of IFNGR1, groups of wild type and IFNAR1^−/− mice were given a high-titer dose of Lm intravenously or fed 10⁹ CFU of Lm and IFNGR1 levels were assessed directly ex vivo by flow cytometry. As shown in Fig. 6A, IFNGR1 expression on dendritic cells and macrophages decreased by approximately one-half in IFNAR1^−/− mice, equivalent to or in some cases greater than the change seen on cells from wild type mice. For both groups of mice, B cell IFNGR1 expression underwent little change, while expression of this receptor on T cells increased. Thus, infection-induced down-regulation of IFNGR1 was not dependent on type I interferon signaling.

FIGURE 6.

Infection-induced decreases in IFNGR1 expression are not type I IFN-dependent. A) IFNAR1^−/− and C57BL/6J (WT) mice were i.v. infected with 4.0 × 10³ CFU or fed 10⁹ CFU Lm SD2000 and spleens were harvested at 24 (i.v.) or 72 (food) hpi. Mean fluorescence intensity values (MFI) for IFNGR1 were normalized to the MFI for uninfected splenocytes, which is represented by the dotted line at 1.0. DC, dendritic cell; MΦ, macrophage. Mean values +/− SD are shown; data was pooled from multiple experiments, n= 4–7 mice per group. (B, C) IFNγ^−/− and C57BL/6J (WT) mice were i.v. infected with 1 × 10⁴ CFU of Lm. Spleens were harvested 24 hours later and IFNGR1 MFI (B) and total CFU burden (C) was determined. Data from two experiments was pooled; horizontal lines indicate mean values.

A variety of cells rapidly produce IFNγ during Lm infection (34–36) and the cytokine can be detected in the blood within 24 to 48 hours of either i.v (16, 17) or foodborne (unpublished observation, Bou Ghanem and D’Orazio) transmission. Therefore, it was possible that ligand binding could trigger internalization of IFNGR1 and contribute to the observed decrease in surface expression during the timeframe we tested. If this was the case, then infection-induced decreases in IFNGR1 expression should be greater in wild type mice than in IFNγ-deficient mice (IFNγ^−/−). To test this, groups of mice were infected i.v. and IFNGR1 expression was assessed 24 hours later. As shown in Fig. 6B, dendritic cells and macrophages from IFNγ^−/− mice showed significantly less down-regulation of IFNGR1 than did cells from wild type mice. Although IFNγ^−/− mice are more susceptible to Lm infection over time, there was no significant difference in Lm burdens between IFNγ^−/− and wild type mice 24 hpi (Fig. 6C). Thus, the difference in IFNGR1 expression was not the result of a disparity in bacterial loads. These results indicated that multiple mechanisms could contribute to decreased surface expression of IFNGR1 during infection, and suggested that this phenotype did not contribute significantly to the enhanced growth of Lm in IFNAR1-deficient mice.

Discussion

Intravenous infection of mice results in a highly reproducible model of systemic listeriosis; however, this method fails to replicate the natural course of infection, which originates in the gastrointestinal tract following the consumption of Lm-contaminated foods. Previous studies using i.v. inoculation have suggested that secretion of type I IFN during the early stages of listeriosis may promote growth of Lm (16, 17). In this study, we demonstrated that foodborne infection of mice did not elicit a robust type I IFN response in the spleen. Accordingly, the host-detrimental effects caused by IFNα/β signaling were not observed when Lm were transmitted by the oral route. The results presented here highlight the necessity of using physiologically relevant infection models to address questions regarding the host immune response to infection.

Intravenous inoculation is widely used for studies of immunity to Lm because it produces a robust infection in mice, with up to 90% of the initial inoculum seeding the spleen and liver within 15 minutes (27). Lm are thought to be removed from the blood by phagocytes surrounding the marginal zone of the spleen. These cells transport Lm to the periarteriolar lymphoid sheath, where increases in bacterial burden can be detected as early as 6 hpi (37). In contrast to the rapid nature of i.v. infection, orally acquired Lm typically require 24–48 hours to exit the gastrointestinal tract and begin colonizing the spleen (29, 38, 39). To reach the spleen, Lm must survive in the harsh environment of the stomach, invade the intestinal epithelium, disseminate to the mesenteric lymph nodes, and then gain access to the circulation. The specific timing of these events is unclear, but because of the bottlenecks involved in this process small numbers of Lm are likely to exit the intestine in waves, rather than as a bolus. It is interesting to note that i.p. inoculation, which also involves an indirect route of spread to the spleen, does trigger IFNα/β-dependent effects that promote Lm growth (18, 28). We speculate that dissemination of Lm from the peritoneum to the spleen occurs more rapidly than spread from the intestinal lamina propria, presumably because there are fewer bottlenecks to overcome.

Our results differ from that of Kernbauer et al., who recently concluded that type I IFN signaling promotes resistance to oral Lm infection because they found that approximately 30% of IFNAR1^−/− mice died within 5 days after intragastric (i.g.) inoculation (28). There are two key differences between our studies. First, Kernbauer et al. used two separate i.g. injections (200 μl of bicarbonate, and then 200 μl of bacteria) to orally infect mice. I.g. infection is a more physically traumatic method than natural feeding and may promote a pathway of direct bloodstream invasion in a user-dependent manner (20). However, such physical trauma would be expected to result in an infection that mimicked i.v. inoculation, and neither our group, nor O’Connell et al. saw increased death of i.v.-infected IFNAR1^−/− mice compared to wildtype animals (17). Thus, the more likely explanation for the unique result in the Kernbauer et al. study is that they used Lm LO28, a strain that overexpresses the multidrug efflux pumps MdrM and MdrT due to a spontaneous deletion in the TetR repressor (40–42). This results in increased secretion of c-di-AMP, greater IRF3 signaling, and hyper-induction of type I IFN compared to either the commonly used laboratory strains EGDe or 10403s or other clinical isolates of Lm.

Myeloid-derived cells rapidly initiate IFNβ production after exposure to Lm. In vitro infection of bone marrow-derived macrophages and dendritic cells resulted in a clear induction of IFNβ mRNA as early as 4–6 hpi (17, 43), and intracellular cytokine staining of Lm-pulsed dendritic cells verified that IFNβ protein was secreted 6 hpi (44). In this study, we manipulated the dose and duration of Lm infection to yield similar bacterial burdens in the spleen after oral and i.v. challenge, but found that i.v. inoculation still induced significantly more secretion of IFNβ from splenocytes than foodborne infection. The simplest interpretation of these results is that the impetus for robust IFNβ secretion is a sudden exposure to a large bolus of bacteria, which infection by the foodborne route would not provide. There are very few published reports that have measured IFNβ levels by ELISA during bacterial infection. However, the amount of IFNβ secreted by splenocytes from i.v.-infected mice in this study was similar to that observed in mouse serum after cecal ligation and puncture (45) and from primary mouse lung fibroblasts following in vitro Chlamydia trachomatis infection (46). We were unable to detect increased IFNα in the spleen during i.v. or foodborne infection, despite the fact that IFNα subtypes outnumber IFNβ by a large ratio (2). Notably, previous measurements of IFNα concentrations in either serum or spleen ranged from 30–50 pg/mL of 24 hours after i.v. Lm infection (47, 48). This may indicate that IFNα is produced by cell types that we did not include in high numbers in our splenocyte cultures.

Robust IFNβ secretion following i.v. inoculation of Lm triggered significant T cell loss, IL-10 secretion, and a dampening of neutrophil recruitment in the C57BL/6 mice used in this study. These findings are consistent with previous studies that suggested type I IFN was detrimental to the host during Lm infection (14, 16–18). None of these IFNβ-dependent effects were observed during foodborne infection. Surprisingly, down-regulation of IFNGR1, a phenotype that was previously linked to the induction of type I IFN (19, 33), occurred during both i.v. and foodborne infection. Our findings support those of Rayamajhi et al. (19) who found that IFNGR1 down-regulation was primarily observed in myeloid-derived cells. They showed that a soluble factor was responsible for suppression of IFNGR1 expression during in vivo Lm infection and used in vitro treatment of bone-marrow derived macrophages to demonstrate that IFNβ could induce this phenotype. Our findings, however, indicate that IFNα/β signaling is not absolutely required for IFNGR1 down-regulation to occur in vivo and suggest that a variety of other signal inputs may influence surface expression of this receptor. For example, IFNγ is quickly produced by a variety of cells in the spleen during Lm infection (34–36), and rapid endocytosis of IFNGR1 due to ligand binding is likely to occur. Non-ligand based interactions could also promote internalization of IFNGR1, as has been shown to occur on T cells following TCR engagement (49). Prior in vitro work has shown that IFNα/β may act as an antagonist for IFNGR1 (50); however, it is unclear whether in vivo infection would induce the concentrations necessary to achieve these effects.

Type I IFN has traditionally been thought of as an immunostimulatory agent critical for inducing an anti-viral response, but it is also commonly used to treat autoimmune diseases such as relapsing-remitting multiple sclerosis (51). Thus, the actions of type I interferons are context-dependent and may be either pro-inflammatory or anti-inflammatory (52, 53). The timing of IFNα/β secretion, and the concentration present in any given tissue, are likely to be primary factors in determining the downstream effects of a type I IFN response. Future studies of microbial pathogens and the type I IFN response to infection should take into account the impact that route of transmission may have on both the induction and the effects of this uniquely multifunctional family of cytokines.

Abbreviations used in this text

IFNAR1

type I interferon receptor alpha chain

IFNGR1

interferon gamma receptor alpha chain

Lm

Listeria monocytogenes