Identifying the initiating events of anti-Listeria responses using mice with conditional loss of IFNγ receptor subunit 1 (IFNGR1)

Abstract

Whereas IFNγ is required for resolution of Listeria monocytogenes infection, the identities of the IFNγ responsive cells that initiate the process remain unclear. We addressed this question using novel mice with conditional loss of the IFNγ receptor (IFNGR1). Itgax-cre⁺Ifngr1^f/f mice with selective IFNγ unresponsiveness in CD8α⁺ dendritic cells displayed increased susceptibility to infection. This phenotype was due to the inability of IFNγ unresponsive CD8α⁺ DCs to produce the initial burst of IL-12 induced by IFNγ from TNFα-activated NK/NKT cells. The defect in early IL-12 production resulted in increased IL-4 production that established a myeloid cell environment favoring Listeria growth. Neutralization of IL-4 restored Listeria resistance in Itgax-cre⁺Ifngr1^f/f mice. We also found that Itgax-cre⁺Ifngr1^f/f mice survived infection with low dose Listeria due to a second wave of IL-12 produced by Ly6C^hi monocytes. Thus, an IFNγ-driven cascade involving CD8α⁺ DCs and NK/NKT cells induces the rapid production of IL-12 that initiates the anti-Listeria response.

Introduction

Listeria monocytogenes is an opportunistic pathogen that causes significant disease in neonates, the elderly, and immunocompromised individuals (1). Production of IFNγ and cellular responsiveness to this cytokine in the host is crucial for the effective resolution of infection, as originally demonstrated using a neutralizing monoclonal antibody to IFNγ (2) and subsequently using mice lacking genes encoding IFNγ (3); IFNGR1—the major ligand binding chain of the IFNγ receptor (4); or STAT1—the major transcription factor that mediates IFNγ receptor signaling (5). Other studies using SCID mice lacking T and B lymphocytes revealed that NK cells were a major source of IFNγ early in the infection and that the IFNγ produced by NK cells activated microbicidal activity in macrophages thus providing the host with an ability to control the infection until such time that sterilizing adaptive immunity to the organism could develop (6–8). A deeper understanding of this innate protective response to Listeria infection came when the cytokines TNFα and IL-12 were found to play important roles in the induction of IFNγ from NK cells (6–11). This work culminated in defining the feed-forward amplification process that leads to development of innate immunity not only to L. monocytogenes but also to many other intracellular pathogens (12).

However, despite all that is known about the importance of IFNγ in the anti-Listeria response, the identities of the specific cellular targets of IFNγ required for initiation of the response and effective control of the infection remain to be established. An early study used transgenic mice expressing a dominant-negative, truncated form of IFNGR1 in certain myeloid cell populations to show that myeloid cell responsiveness to IFNγ was critical for promoting protective host responses to L. monocytogenes (13). Another study used radiation bone marrow (BM) chimera approaches to demonstrate that IFNγ receptor (IFNγR) expression in the hematopoietic compartment was required for controlling Listeria infection (14). However, since functional IFNγRs are expressed in almost every host cell type (15), it has, until now, not been possible to more precisely identify the key IFNγ responsive cells required to initiate the anti-Listeria response.

Recently, much attention has focused on the role of dendritic cells (DCs) in Listeria infection. DCs are the primary cell type that sense, ingest, and present exogenous antigens from pathogens to initiate the pathogen specific adaptive immune response (16). Within this population, the CD8α⁺/CD103⁺ DC subsets have been shown to play a major role in cross-presenting exogenous antigens to CD8⁺ T cells thereby inducing host protective cytotoxic T cell responses (17, 18). Recent studies using CD11c-DTR mice, in which the diphtheria toxin receptor (DTR) was expressed only in CD11c⁺ cells, revealed that mice depleted of all DCs did not develop Listeria infection in the spleen (19, 20). Furthermore, using Batf3^−/− mice, that selectively lack CD8α⁺/CD103⁺ DCs, a role was demonstrated for these specific DC subsets in establishing Listeria infection in the spleen and liver (21). Together these findings support a scenario in which migratory CD8α⁺ DCs carry L. monocytogenes from their entry point in the splenic marginal zone to the periarteriolar lymphoid sheaths (PALS), where L. monocytogenes then multiply in the ensuing 12–24 hours and establish an active infection (22, 23).

Whereas the aforementioned studies reveal a critical role for CD8α⁺/CD103⁺ DCs in L. monocytogenes transport and initiation of infection, they do not provide insight into the interactions of these cells with other immune cells and cytokines. Although the cross-presenting functions of CD8α⁺/CD103⁺ DCs are known to be influenced by type I interferons (24), little is known about the functional effects of IFNγ on these cells. Thus, we asked whether IFNγ responsiveness in CD8α⁺/CD103⁺ DCs directly influenced their ability to initiate anti-Listeria responses. We therefore generated mice with a floxed Ifngr1 gene (Ifngr1^f/f mice) on a C57BL/6 background and then bred them to either C57BL/6 Vav-icre or Itgax-cre mice to impart IFNγ unresponsiveness either broadly in hematopoietic cells or specifically in the CD8α⁺/CD103⁺ DC subsets, respectively (25, 26). Using these novel mice, we report herein the elucidation of the events that underlie development of the innate immune response to L. monocytogenes and show that IFNγ responsiveness in CD8α⁺/CD103⁺ DCs plays a critical role in initiating this process.

Materials and Methods

Generation of Ifngr1^f/f mice

The TNLOX1-3 targeting vector (27) was used to generate the conditional Ifngr1 targeting construct and electroporated into B6/Blu embryonic stem (ES) cells (28). After selection, ES cells were analyzed for homologous recombination by PCR using a 5′ external primer (f3: 5′-ccagtgtttgcctttggatctg-3′, IDT) and a Neo primer (r3: 5′-gttggctacccgtgatattgctg-3′, IDT). PCR positive clones were verified by Southern blotting following BamHI digestion using 5′ and 3′ external probes (Fig. 1A). One ES cell clone displaying complete integration of the targeting construct was verified by PCR using loxP specific primer sets (f1: 5′-aaacagtaaacccagggctttgtac-3′; r1: 5′-cagcctctgaaattcaaatggc-3′; f2: 5′-gtgacgggagcacctgttac-3′; r2: 5′-gtaagtgcattcatctggccag-3′, IDT). The correctly targeted ES clone was expanded and transiently transfected with pTurbo-Cre (ES cell core, Washington University) for removal of the Neo resistance gene. Neo-sensitive clones were screened using 5′ primers (5′-aaacagtaaacccagggctttgtac-3′, IDT) and two 3′ primers (5′-cagcctctgaaattcaaatggc-3′, 5′-cgtggcactgtagatgtactgtcag-3′, IDT) and confirmed by Southern blotting using the 5′ probe. Neo-deleted sub-clones were injected into eight cell embryos using the laser-assisted microinjection technique (29).

Figure 1. Generation of conditional IFNγ receptor gene targeted mice (Ifngr1^f/f).

(A) Targeting strategy. H1, BamH1; Neo, neomycin resistance gene; TK, herpes simplex virus thymidine kinase gene; open triangle, loxP site. Open gray boxes indicate external southern probe to verify 5′ and 3′ homologous recombination. Small arrowheads indicate pairs of primers used in PCR to confirm either 5′ homologous recombination (f3/r3) or insertion of loxP sites (f1/r1 and f2/r2). (B) Confirmation of germline transmission by examining 3′ end of targeted allele by southern blotting. +, wild allele; f, conditional allele. (C) Verification of correct insertion of both 34 bp-long loxP sites by PCR. (D) Expression of IFNGR1 was assessed by flow cytometry analysis of spleen, peripheral blood leukocytes (PBLs), and thymus from wild type, Ifngr1^f/f, and Ifngr1_WU^−/− mice. Data are representative of three separate experiments. (E) Confirmation of germline transmission of the Ifngr1_WU^−/− allele by 5′ southern blotting. (F) Measurement of phosphorylated STAT1 in Ifngr1_WU^−/− mice after in vitro IFN stimulation (10,000 U/ml IFNα5, 10,000 U/ml IFNβ, and 1,000 U/ml IFNγ) for 10 minutes at 37°C.

Mice

B6 (C57BL/6NTac) mice were obtained from Taconic. Itgax-cre (007567; C57BL/6J-Tg (Itgax-cre,-EGFP)4097Ach/J) mice (25) and Vav-icre (008610; B6.Cg-Tg(Vav1-cre)A2Kio/J) mice (26) were obtained from the Jackson Laboratory. Both Itgax-cre and Vav-icre mice were backcrossed onto the C57BL/6 background using speed congenic approaches (>99% purity, Rheumatic Diseases Core Center, Washington University) and then crossed to Ifngr1^f/f mice. Mice were maintained in a specific pathogen-free facility in accordance with American Association for Laboratory Animal Science guidelines, and all protocols involving laboratory animals were approved by the Washington University Animal Studies Committee.

Infection

All mice were infected with indicated doses of L. monocytogenes (strain: EGD) in pyrogen-free saline. Colony counts were determined as previously described (30). In some experiments, portions of spleens and livers were fixed in 10% formalin, embedded in paraffin, and stained with hematoxylin and eosin (H&E).

Antibody treatment

To neutralize cytokines and/or block receptors, mice were treated i.p. on day -1 (relative to Listeria infection) with the following endotoxin-free preparations of mAbs: 200 μg of 11B11 mAb for IL-4 neutralization (31); 500 μg of SK113AE-4 mAb for IL-18 neutralization (32); 250 μg each of ALF-161, B122, and JAMA-147 mAbs for IL-1 neutralization/blockade (33–35); 250 μg each of 55R-170 and TR75-54 mAbs for TNFR1 and TNFR2 blockade (36); 250 μg of H22 mAb or IFNγ neutralization (37). For in vivo depletion of NK/NKT cells, Ifngr1^f/f mice were treated with 200 μg PK136 mAb (Biolegend) i.p. on days -2 and 0. MP CD8⁺ T cells were depleted by injecting Ifngr1^f/f mice i.p. with 200 μg CXCR3-173 mAb (38) on days -3 and 0.

Flow cytometry

The following monoclonal antibodies were purchased from Biolegend and used as lineage markers: PerCP/Cy5.5 anti-CD3ε (145-2C11), PE or APC anti-NK1.1 (PK136), FITC or PE/Cy7 anti-CD4 (GK1.5), PerCP/Cy5.5 or APC/Cy7 anti-CD8α (53-6.7), FITC or PE anti-CD45R/B220 (RA3-6B2), APC anti-CD317/PDCA-1 (927), FITC or PerCP/Cy5.5 anti-F4/80 (BM8), PerCP/Cy5.5 or PE/Cy7 anti-CD11b (M1/70), PE anti-CD115 (AFS98), APC/Cy7 anti-Ly6G&C/Gr-1 (RB6-8C3), APC/Cy7 anti-Ly6C (HK1.4), PerCP/Cy5.5 anti-Ly6G (1A8), FITC or APC/Cy7 anti-CD11c (N418), PerCp/Cy5.5 anti-CD103 (2E7), APC anti-CD183/CXCR3 (CXCR3-173), FITC anti-CD62L (MEL-14), PE anti-CD44 (IM7), PE anti-CD31 (MEC13.3). PE anti-CD122 (TM-β1) and PE anti-Siglec-F (E50-2440) were purchased from BD Biosciences. APC anti-Dec205 (205yekta) was purchased from eBioscience. Single cell suspensions from various tissues were prepared and stained for IFNGR1 and pSTAT1 with biotinylated anti-IFNGR1 (GR20, BD Bioscience) and Alexa647 anti-pSTAT1 (4a, BD Bioscience) as previously described (24). For intracellular cytokine staining, cells were stained with lineage markers without either in vitro re-stimulation or incubation with intracellular transport blockers. To label dead cells, Fixable Viability Dye eFluor® 450 (eBioscience) was used prior to fixation and permeabilization procedures with BD Cytofix/Cytoperm^™ (BD Biosciences). Cells were subsequently stained with PE anti-IL-12p40 (C15.6, Biolegend), PE/Cy7 anti-IFNγ (XMG1.2, Biolegend), PE/Cy7 anti-TNFα (MP6-XT22, Biolegend), and goat anti-NOS2 (Santa Cruz) followed by PerCP/Cy5.5 donkey anti-goat IgG (Santa Cruz).

Quantitative real-time PCR

RNA was prepared from spleens and cell pellets of sorted cells using an RNeasy Mini/Micro Kit (Quiagen). Purified RNA was reverse-transcribed to cDNA using random hexamers and Superscript III Reverse Transcriptase (Invitrogen). Quantitative real-time PCR reaction was carried out on a 7000 Sequence Detection System (Applied Biosystems) by using Power SyBr Green PCR master mix (Applied Biosystems). Every real-time PCR was normalized by 18S. Specific primers (IDT) for quantitative real-time PCR were designed using Primer Bank (http://pga.mgh.harvard.edu/primerbank/).

Immunohistochemistry

Alexa488 anti-B220 (RA3-6B2), biotin anti-CD3ε (145-2C11), and biotin anti-CD11b (M1/70) were purchased from BD Biosciences. Alexa555-SA, Alexa555 goat anti-Rat IgG (H+L highly Cross-Adsorbed), and Alexa647 goat anti-rabbit IgG (H+L highly Cross-Adsorbed) were purchased from Invitrogen. Listeria O antiserum Poly Type 1 & 4 was purchased from BD Diagnostics. Alexa647 anti-CD11c (N418) was purchased from eBioscience. Biotin anti-Siglec-1(MOMA-1) was purchased from BMA Biomedicals. Rat anti-MARCO was purchased from Serotec.

Fresh spleens were embedded in Tissue-Tec OCT (Fisher) and 7 μm frozen sections were fixed in acetone at 4°C for 5 min. All slides were blocked with CAS Block (Invitrogen) and stained with indicated antibodies diluted in CAS Block. Four-color epifluorescence microscopy was performed with an Olympus BX51 microscope equipped with a SPOT RT CCD camera (Diagnostic Instruments). Monochrome images were pseudo-colored with SPOT RT camera software, and merged with Adobe Photoshop.

Statistical analysis

The statistical analysis was performed using the Mann-Whitney test and Prism software (GraphPad Software). Statistical p values less than or equal to 0.05 were considered to be significant. Error bars indicate standard error of the mean (SEM).

Results

Generation and characterization of C57BL/6 Ifngr1^f/f mice

We generated a conditional knockout allele of Ifngr1 in C57BL/6 ES cells by placing two loxP sites surrounding the third and fourth exons that encode the extracellular domain of IFNGR1 (Fig. 1A). Deletion of this region results in a frame shift mutation that induces expression of only a minimal portion of IFNGR1 that is unable to bind IFNγ (39). Southern blotting and PCR analyses confirmed proper gene targeting (Fig. 1B and 1C). Flow cytometric analyses revealed that Ifngr1^f/f mice were indistinguishable from C57BL/6 mice on the basis of cell surface expression of IFNGR1 (Fig. 1D).

This same targeting also produced sub-clones of ES cells in which the entire floxed region was deleted when they were transfected with pTurbo-cre (Fig. 1A). The fully deleted sub-clones gave rise to C57BL/6 Ifngr1^−/− mice (Fig. 1E) that were named Ifngr1_WU^−/− so as to distinguish them from Ifngr1^−/− mice previously generated on a 129/SvEv background (4). Cells from Ifngr1_WU^−/− mice neither expressed IFNGR1 (Fig. 1D) nor responded to IFNγ in vitro (as detected by STAT1 phosphorylation), but responded normally to Type I Interferons (Fig. 1F).

Generation of C57BL/6 mice with IFNGR1 deficiency either in all hematopoietic cells or selectively in CD8α⁺/CD103⁺ DCs

To delete Ifngr1 in hematopoietic cells, we bred Ifngr1^f/f mice to Vav-icre mice (Vav-icre⁺Ifngr1^f/f) since the latter are known to delete floxed genes in all hematopoietic cells (26). Hematopoietic cells from the resulting Vav-icre⁺Ifngr1^f/f mice neither expressed IFNGR1 (Fig. 2A) nor responded to IFNγ treatment by phosphorylating STAT1 (Fig. 2B). In contrast, CD31⁺CD45⁻ endothelial cells from these mice displayed undiminished levels of IFNGR1 (Fig. 2A) and pSTAT1 after IFNγ stimulation (Fig. 2B), thus demonstrating that IFNGR1 expression in the non-hematopoietic compartment was not affected in Vav-icre⁺Ifngr1^f/f mice.

Figure 2. Vav-icre⁺Ifngr1^f/f mice lack functional IFNGR1 in hematopoietic cells.

(A) Splenic IFNGR1 expression in Vav-icre⁺Ifngr1^f/f mice was measured. (B) After in vitro IFNγ stimulation (1,000 U/ml) for 15 minutes at 37°C, phosphorylated STAT1 in splenocytes from Vav-icre⁺Ifngr1^f/f mice was analyzed. All Data are representative of at least two separate experiments. Gating strategies are depicted in Supplemental Fig. 1A.

Ifngr1^f/f mice were also bred to a particular strain of Itgax-cre mice (25) selected because we had shown previously that they induce a selective deletion of floxed genes in CD8α⁺/CD103⁺ DCs (24). Itgax-cre⁺Ifngr1^f/f mice showed significant reductions in IFNGR1 expression primarily in splenic CD8α⁺ DCs, with partial reduction in splenic CD4⁺ DCs and slight reduction in macrophages (Fig. 3A). Splenic CD8α⁺ DCs and peripheral CD103⁺ DCs are functionally and developmentally related (40). Thus it was not surprising to find that CD103⁺ DCs in the peritoneal cavity and liver exhibited an almost complete deletion of Ifngr1 in Itgax-cre⁺Ifngr1^f/f mice, whereas myeloid CD11b⁺ DCs from these tissues did not (Fig. 3B). Of note, CD8α⁺ DCs from the Itgax-cre⁺Ifngr1^f/f mice displayed significant response defects to IFNγ as evidenced by quantitating STAT1 phosphorylation (Fig. 3C) and CD40 up-regulation (Fig. 3D). We observed the same response defects even using 10 fold higher doses of IFNγ for a longer incubation time (Supplemental Fig. 1C). In contrast, no defect was observed in IFNγ receptor signaling or responsiveness in CD4⁺ DCs, macrophages, and other hematopoietic cells. Thus, Itgax-cre⁺Ifngr1^f/f mice display a highly specific IFNγ unresponsiveness within the CD8α⁺/CD103⁺ DC compartment.

Figure 3. Itgax-cre⁺Ifngr1^f/f mice lack functional IFNGR1 in CD8α⁺/CD103⁺ DCs.

(A) Splenic IFNGR1 expression in Itgax-cre⁺Ifngr1^f/f mice was measured. IFNGR1 levels in the indicated cellular subsets in Itgax-cre⁺Ifngr1^f/f mice compared with Ifngr1^f/f mice are summarized in the associated bar graph of panel. (B) FACS analysis in liver and peritoneal cavity to confirm the lack of IFNGR1 expression in hepatic and peripheral CD103⁺ DCs in Itgax-cre⁺Ifngr1^f/f mice. (C) After in vitro IFNγ stimulation (1,000 U/ml) for 15 minutes at 37°C, phosphorylated STAT1 in splenocytes from Itgax-cre⁺Ifngr1^f/f mice was analyzed. pSTAT1 staining of un-stimulated controls was indistinguishable from that in Ifngr1_WU^−/− mice (Supplemental Fig. 1B). (D) Selective lack of up-regulation of CD40 in splenic CD8α⁺ DCs in Itgax-cre⁺Ifngr1^f/f mice after in vitro IFNγ stimulation (500 U/ml) for 18 hr at 37°C. Splenic CD11c⁺ cells were positively enriched by MACS purification prior to IFNγ stimulation. The wildtype, knockout, and isotype controls for panel (A) and (C) are the same as for Fig. 2 panel (A) and (B) because the flow cytometry was performed at the same time. All Data are representative of at least two separate experiments. Gating strategies are depicted in Supplemental Fig. 1A. *, p ≤ 0.05.

IFNγ responsiveness in hematopoietic cells is required to control Listeria infection

When challenged i.p. with two different doses of L. monocytogenes (1 x10⁵ or 2.5 × 10⁵), both Vav-icre⁺Ifngr1^f/f and Ifngr1_WU^−/− mice were more susceptible to infection than Ifngr1^f/f mice succumbing to infection by day 6 (Fig. 4A, left and right panels). Spleens and livers from Vav-icre⁺Ifngr1^f/f mice and Ifngr1_WU^−/− mice contained 10-fold more L. monocytogenes on day 1 and 100- to 1000-fold more bacteria on day 3 compared to the same organs from Ifngr1^f/f mice (Fig. 4B). These results functionally recapitulate the defect previously noted in Ifngr1^−/− BM chimeras (14) revealing an obligate requirement for IFNγ responsiveness in the hematopoietic compartment for resolution of Listeria infection.

Figure 4. Both Vav-icre⁺Ifngr1^f/f and Itgax-cre⁺Ifngr1^f/f mice display increased susceptibility to Listeria infection.

(A) Mice were infected with 2.5 × 10⁵ (left panel) or 1 × 10⁵ (right panel) L. monocytogenes i.p. and the survival were monitored over time. (B and C) Listeria CFUs in spleen, liver, and peritoneum infected with 10⁵ L. monocytogenes i.p. (D) Mice were infected with 1 × 10⁴ (left panel), 5 × 10³ (middle panel), or 2.5 × 10³ (right panel) L. monocytogenes i.v. and the survival were monitored over time. (E) Listeria CFUs in spleen, liver, and peritoneum at 3 days after i.v. infection with 10³ or 10⁴ L. monocytogenes. Each symbol in Listeria CFUs represents an individual mouse and lines represent the mean Log₁₀CFU. *, p ≤ 0.05; **, p ≤ 0.01; ***, p ≤ 0.001.

IFNγ responsiveness in CD8α⁺/CD103⁺ dendritic cells is required for optimal anti-Listeria responses

When challenged i.p. with an LD50 dose of L. monocytogenes (2.5 × 10⁵), all of the Itgax-cre⁺Ifngr1^f/f mice succumbed to infection (Fig. 4A, left panel). At a sub-lethal dose (1 × 10⁵), Itgax-cre⁺Ifngr1^f/f mice exhibited significantly increased L. monocytogenes burdens in the spleen, liver, and peritoneum compared to Ifngr1^f/f mice during the first 7 days of infection (Fig. 4C). As previously reported (41, 42), the LD50 dose of Listeria administered i.v. (between 5 × 10³ and 1 × 10⁴) is much lower than that for i.p. challenge (2.5 × 10⁵) (Figure 4A and 4D). The defect in Listeria clearance in Itgax-cre⁺Ifngr1^f/f mice was not dependent on the route of infection since Itgax-cre⁺Ifngr1^f/f mice harbored more L. monocytogenes than Ifngr1^f/f mice even when the bacteria were administered i.v. (Fig. 4D and 4E). We failed to detect any bacteria in peritoneum after i.v. challenge. This result indicates that the larger Listeria burdens in the spleen and liver following i.p. infection compared to i.v. infection may indicate an increased unidirectional seeding of bacteria from the peritoneum in Itgax-cre⁺Ifngr1^f/f mice to spleen/liver (Figure 4C and 4E). Histologically, Itgax-cre⁺Ifngr1^f/f mice showed highly increased numbers and sizes of Listeria foci in both spleen and liver compared to Ifngr1^f/f mice (Supplemental Fig. 2A). Eighteen hours after Listeria infection, infectious foci in the PALS were observed in both Itgax-cre⁺Ifngr1^f/f and Ifngr1^f/f mice indicating that the normal migration of Listeria-infected CD8α⁺ DCs had occurred regardless of their ability to respond to IFNγ (Supplemental Fig. 2B). Mice solely expressing cre protein (Vav-icre⁺ or Itgax-cre⁺ mice) exhibited no differences in bacterial burdens when compared to wild type C56BL/6 mice, thus excluding potential influences of the cre protein on Listeria susceptibility (Supplemental Fig. 2C).

Two sets of data revealed that the increased susceptibility to Listeria infection in Itgax-cre⁺Ifngr1^f/f mice was specifically due to IFNγ unresponsiveness in CD8α⁺/CD103⁺ DCs. First, strong expression of EGFP was detected in CD8α⁺ DCs, less in CD4⁺ DCs, and not at all in other cells before and after Listeria infection (Supplemental Fig. 2D). This result rules out the possibility that Listeria infection might result in an infection-mediated deletion of Ifngr1 in other cells by up-regulating expression of the transgenic cre-EGFP bicistronic construct. Second, the observation that CD11c was not expressed in marginal zone macrophages (MZM) and metallophilic MZM (MMM) rules out the same possibility in these cells that trap L. monocytogenes at the onset of infection (43) (Supplemental Fig. 2E).

IFNγ-insensitive CD8α⁺ DCs display defective IL-12 production

CD8α⁺ DCs and monocytes are implicated as the major cellular sources of IL-12 in Listeria infection (44–46). We therefore examined the effects of endogenously produced IFNγ on IL-12 production by splenic CD8α⁺ DCs and monocytes from Ifngr1^f/f and Itgax-cre⁺Ifngr1^f/f mice during the first 24 hours of infection (Fig. 5). The limit of detection for ex vivo intracellular staining of IL-12 is 1 × 10⁵ L. monocytogenes i.v. although we can still detect IL-12 transcripts with as low as 1 × 10⁴ L. monocytogenes (Supplemental Fig. 3A and B). To synchronize bacterial infection and generate enough cytokine-producing cells, mice were infected i.v. with 10⁶ L. monocytogenes. The percentage of splenic CD8α⁺ DCs producing IL-12p40 in Ifngr1^f/f mice increased to 4.2% at 9 hr, reaching a maximum level of 9.4% at 12 hr, then decreasing to 5.9% at 18 hr, and back to baseline at 24 hr. (Fig. 5A and 5B). In contrast, the proportion of IL-12p40 producing CD8α⁺ DCs from Itgax-cre⁺Ifngr1^f/f mice was significantly less (1.9% at 9 hr; 3.6% at 12 hr; and 1.3% at 18 hr). Listeria-infected Ifngr1_WU^−/− mice showed complete abrogation of IL-12 production in CD8α⁺ DCs demonstrating that the low level induction of IL-12 in CD8α⁺ DCs from Itgax-cre⁺Ifngr1^f/f mice was due to incomplete deletion of the Ifngr1 gene (Fig. 5C). Expression of IL-12p35 was also significantly decreased in CD8α⁺ DCs from Listeria-infected Itgax-cre⁺Ifngr1^f/f mice compared to infected Ifngr1^f/f mice while CD8α⁺ DCs from neither mouse produced significant amounts of IL-23p19 (Fig. 5D). Thus, CD8α⁺ DCs required IFNγ responsiveness for IL-12 production.

Figure 5. Early production of IL-12 from CD8α⁺ DCs is significantly decreased in Listeria-infected Itgax-cre⁺Ifngr1^f/f mice.

All mice were infected with 10⁶ L. monocytogenes i.v. (A) Representative flow cytometry plots for IL-12p40 expression from CD8α⁺ DCs and Ly6C^hi monocytes. For each quadrant, gating was based on cells from uninfected controls that were analyzed at every time point (For simplicity, only the uninfected control at 9 hr after infection is shown). (B) Summary of percentages of IL-12p40 positive CD8α⁺ DCs in the spleen during the first 24 hr of infection (n ≥ 4 at each time point). (C) Percentages of IL-12p40 positive CD8α⁺ DCs in the spleen from Ifngr1^f/f mice and Ifngr1_WU^−/− mice at 9 hr of infection. (D) The expression of indicated genes in sorted CD8α⁺ DCs from spleens was determined by qRT-PCR after 9 hr of infection. (E) Summary of percentages of IL-12p40 positive Ly6C^hi monocytes in the spleen during the first 24 hr of infection (n ≥ 4 at each time point). *, p ≤ 0.05; **, p ≤ 0.01.

IL-12p40 production by CD8α⁺ DCs subsided 24 hr after infection. However, at this time, approximately 10% of Ly6C^hi monocytes stained positively for IL-12p40 in both Ifngr1^f/f and Itgax-cre⁺Ifngr1^f/f mice (Fig. 5A and 5E). Approximately 30% of IL-12p40⁺Ly6C^hi monocytes had a phenotype ascribed to TipDCs—identified by their production of TNFα and iNOS (Supplemental Fig. 3C). Thus, CD11c^intCD11b⁺Ly6C^hiLy6G⁻ monocytes displayed a temporally delayed production of IL-12p40 compared to CD8α⁺ DCs.

IFNγ insensitivity in CD8α⁺ DCs leads to a reduction of IFNγ from NK/NKT cells

NK and NKT cells are major sources of IFNγ during the early phases of Listeria infection and IL-12 is known to be required for early IFNγ production by NK cells (6). Therefore we examined the consequences of IFNγ unresponsiveness in CD8α⁺ DCs on the early induction of IFNγ. The percentage of IFNγ⁺ NK/NKT cells in Itgax-cre⁺Ifngr1^f/f mice was significantly decreased compared to that in Ifngr1^f/f mice at both 9 and 12 hr after infection (NK cells: 1.9% vs 4.8% at 9 hr and 27.3% vs 36.5% at 12 hr; NKT cells: 3.5% vs 8.0% at 9 hr and 13.9% vs 21.1% at 12 hr) (Fig. 6A–C). The reduced percentages of IFNγ⁺ NK/NKT cells from Itgax-cre⁺Ifngr1^f/f mice occurred concomitantly with the decreased percentage of IL-12p40⁺CD8α⁺ DCs. However, the levels of IFNγ⁺ NK/NKT cells in Itgax-cre⁺Ifngr1^f/f mice normalized to those found in Ifngr1^f/f mice by 18 to 24 hr when Ly6C^hi inflammatory monocytes produced a second wave of IL-12p40.

Figure 6. Early production of IFNγ from NK and NKT cells is significantly decreased in Listeria-infected Itgax-cre⁺Ifngr1^f/f mice.

All mice were infected with 10⁶ L. monocytogenes i.v. (A) Representative plots for IFNγ from NK, NKT and CD8⁺ T cells during the first 24 hr of infection. Summary of the percentages of IFNγ positive splenic NK, NKT, and CD8⁺ T cells is shown in (B), (C), and (D), respectively (n ≥ 4 at each time point). *, p ≤ 0.05.

At 18 hr post-infection, IFNγ production was observed in CD8⁺ T cells in both Itgax-cre⁺Ifngr1^f/f and Ifngr1^f/f mice (Fig. 6A and 6D). These T cells could represent the memory phenotype (MP) CD8⁺ T cells reported to produce IFNγ during Listeria infection (47). MP CD8⁺ T cells express high levels of CD44 and CD62L and very high levels of CXCR3, and are selectively depleted by our CXCR3 specific mAb (CXCR3-173) (38) (Supplemental Fig. 4A–C). CD4⁺ DCs, plasmacytoid DCs, CD4⁺ T cells, B cells, macrophages, and neutrophils were not significant sources of either IL-12 or IFNγ in the first 24 hr of infection (Supplemental Fig. 3D).

NK/NKT cells provide the initial IFNγ to CD8α⁺ DCs for optimal production of IL-12

We considered the possibility that either NK/NKT cells or MP CD8⁺ T cells might be the initial source of IFNγ that primes CD8α⁺ DCs for the first wave of IL-12. To test this idea, we treated Ifngr1^f/f mice with NK1.1 mAb, CXCR3-173 mAb, or both in combination (Fig. 7A), and assessed whether there was a corresponding decrease in IL-12 production by CD8α⁺ DCs (Fig. 7B). Depletion of NK and NKT cells resulted in a 53% decrease of IL-12p40⁺CD8α⁺ DCs (2.7% anti-NK1.1 mAb treated mice vs 5.7% control mice). In contrast, the elimination of ~80% of MP CD8⁺ T cells did not reduce IL-12p40⁺CD8α⁺ DCs. These results point to a crosstalk between NK/NKT cells and CD8α⁺ DCs mediated by IFNγ and IL-12.

Figure 7. Depletion of NK/NKT cells producing early IFNγ significantly reduces IL-12p40 production from CD8α⁺ DCs.

(A) Representative FACS plots documenting depletion of NK/NKT cells by anti-NK1.1 mAb (top and second row), depletion of MP CD8⁺ T cells by anti-CXCR3 mAb (third row), and IL-12p40 production from CD8α⁺ DCs at 9 hr of infection with 10⁶ L. monocytogenes i.v. (bottom row). (B) Summary of percentages of IL-12p40 positive CD8α⁺ DCs after treatment with depleting mAbs (n ≥ 5). **, p ≤ 0.01.

TNFα induces IFNγ from NK/NKT cells and thereby initiates the reciprocal activation of NK/NKT cells and CD8α⁺ DCs

The cytokines TNFα, IL-18, and IL-1 are known to be involved in initiating anti-Listeria immunity in part by inducing IFNγ production from NK cells (8, 35, 48). We therefore assessed whether any of these cytokines initiated the crosstalk between NK/NKT cells and CD8α⁺ DCs. Ifngr1^f/f mice were treated with mAbs that either neutralize specific cytokines or block their receptors prior to infection (Fig. 8). Dual blockade of TNFR1 and TNFR2 decreased the percentage of IFNγ⁺ NK cells from 6% to 3.2% (47% decrease) (Fig. 8A) and of NKT cells from 11% to 2.8% (75% decrease) (Fig. 8B) which was accompanied by a reduction in IL-12p40⁺CD8α⁺ DCs from 6.6% to 2.6% (61% decrease) (Fig. 8C). As expected, direct neutralization of IFNγ achieved a comparable 64% decrease in IL-12p40⁺CD8α⁺ DCs (Fig. 8C). In contrast, IL-18 neutralization had no effect on IL-12p40 production, although neutralizing IL-18 in combination with blocking TNFR1 and TNFR2 down-regulated IFNγ⁺ NK cells 20% more than dual TNFR-blockade alone (Fig. 8A). Neutralization of IL-1 caused a 24% decrease in IL-12p40⁺CD8α⁺ DCs without altering the percentage of IFNγ⁺ NK/NKT cells (Fig. 8C). Thus, the initial production of IFNγ from NK/NKT cells upon Listeria infection that sets in motion a reciprocal activation of NK/NKT cells and CD8α⁺ DCs is a consequence of a process induced predominantly by TNFα with potential participation of other early arising pro-inflammatory cytokines.

Figure 8. Blockade of TNFα prior to Listeria infection reduces both IL-12 production from CD8α⁺ DCs and IFNγ production from NK/NKT cells.

Ifngr1^f/f mice were pretreated with indicated mAbs (as described in Materials and Methods) prior to infection with 10⁶ L. monocytogenes i.v. Spleens were harvested at 9 hr after infection, and cells analyzed for expression of IL-12p40 and IFNγ by intracellular cytokine staining. Percentages of IFNγ positive NK cells (A), IFNγ positive NKT cells (B), and IL-12p40 positive CD8α⁺ DCs (C) are plotted as bar graphs (n ≥ 4). *, p ≤ 0.05; **, p ≤ 0.01; ***, p ≤ 0.001.

IL-4 production increases the Listeria susceptibility of Itgax-cre⁺Ifngr1^f/f mice by inducing alternatively activated macrophages

Enhanced Listeria susceptibility of Itgax-cre⁺Ifngr1^f/f mice could have resulted from a loss-of-function of the early IL-12/IFNγ amplification cycle or from a gain-of-function process due to inappropriate induction of cytokines that create a Listeria permissive environment. We thus compared systemic cytokine mRNA induction in spleens of infected Itgax-cre⁺Ifngr1^f/f versus Ifngr1^f/f mice (Fig. 9A). As expected, 9 hr after infection, splenocytes from Itgax-cre⁺Ifngr1^f/f mice exhibited a significantly decreased transcription of IL-12p40, IL-12p35, and IFNγ compared to Ifngr1^f/f mice. In contrast, we observed enhanced induction of IL-4 transcripts selectively in Itgax-cre⁺Ifngr1^f/f mice. IL-10 and IL-13 transcripts were induced comparably in both Itgax-cre⁺Ifngr1^f/f and Ifngr1^f/f mice. NKT cells were identified as one source of IL-4 in Itgax-cre⁺Ifngr1^f/f mice (Fig. 9B). We next examined whether increased Listeria susceptibility in Itgax-cre⁺Ifngr1^f/f mice was due to increased IL-4 production. Administration of IL-4-neutralizing mAb (11B11) to Itgax-cre⁺Ifngr1^f/f mice significantly reduced their Listeria burden, bringing bacterial loads down to levels observed in Ifngr1^f/f mice (Fig. 9C). IL-4 neutralization in Ifngr1^f/f mice slightly increased the ability of these mice to resist infection but this increase was not statistically significant. These results demonstrate that the increased bacterial burdens seen in Itgax-cre⁺Ifngr1^f/f mice represented a gain-of-function process resulting from the ectopic expression of IL-4 as a consequence of the absence of early IL-12 from CD8α⁺ DCs.

Figure 9. Neutralization of IL-4 restores Listeria resistance in Itgax-cre⁺Ifngr1^f/f mice.

(A) Spleens were harvested at 9 hr after infection with 10⁶ L. monocytogenes i.v. and analyzed for the expression of indicated genes by qRT-PCR. (B) IL-4 expression in NK and NKT cells sorted from spleens was determined by qRT-PCR after 9 hr of infection with 10⁶ L. monocytogenes i.v. (C) Listeria CFUs in spleen and liver of infected mice at 3 days after infection. Mice were pretreated with 11B11 mAb prior to infection with 10³ L. monocytogenes i.v. Data are a combination of two separate experiments. (D) The expression of indicated genes was analyzed by qRT-PCR at 12 hr after infection with 10⁶ L. monocytogenes i.v. Every qRT-PCR data is represented relative to the expression of 18S (ΔC_t). In order to facilitate visualization, values were transformed as indicated on the y-axis label. *, p ≤ 0.05; **, p ≤ 0.01; ***, p ≤ 0.001.

IL-4 is known to induce alternatively activated (M2) macrophages that do not possess strong bactericidal activity compared to IFNγ activated (M1) macrophages (49). Strikingly, at 12 hr after infection splenocytes from Itgax-cre⁺Ifngr1^f/f mice expressed significantly higher levels of mRNA encoding Arginase 1 (Arg1), Mannose receptor (MR) and the secretory protein Ym1—hallmarks of M2 macrophages—compared to Ifngr1^f/f mice (Fig. 9D). Moreover, expression of inducible nitric oxide synthase (iNOS)—a hallmark of M1 macrophages—was decreased in Itgax-cre⁺Ifngr1^f/f mice compared to Ifngr1^f/f mice. These results thus demonstrate that the lack of IFNγ responsiveness in the CD8α⁺ DC compartment not only compromises the initiation of the anti-Listeria response (i.e., reduces the early induction of IL-12, IFNγ and M1 macrophages) but also allows for expression of IL-4 that effects early polarization of the macrophage compartment to alternatively activated M2 macrophages that lack the critical effector functions needed to destroy L. monocytogenes.

Discussion

Several years of studies have established that the innate anti-Listeria response begins within the first 12 hr after infection (12). This response is characterized by the presence of a number of cytokines—the three studied here—TNFα, IFNγ and IL-12—as well as type 1-interferons, IL-1α/IL-1β, IL-6, IL-10 and others. This early response also involves various cells: tissue resident macrophages, DCs, neutrophils, γδ T cells, NK cells, and innate CD8⁺ T cells. In this study, we used novel mice with a selective deficit of IFNGR1 expression in CD8α⁺ DCs to identify the roles of TNFα, IFNγ, and IL-12 in initiating the critical cytokine and cellular interactions that lead to the effective elimination of Listeria infection.

TNFα, potentially produced by infected splenic marginal zone macrophages (50), sits at the top of the initiating cytokine cascade and induces the first IFNγ from NK/NKT cells that, in turn, induces the early IL-12 from CD8α⁺ DCs. TNFα was originally identified as a major participant with IL-12 in the induction of IFNγ by NK cells although the sequence of events was never defined (7). The transient IFNγ dependent induction of IL-12 from CD8α⁺ DCs represents a key step in forming an amplification loop that enhances IFNγ by NK/NKT cells, depresses the latter’s ability to produce IL-4, and establishes an environment within the myeloid compartment that remains receptive to stimulation of enhanced anti-microbial function. Despite the defect in early IL-12 production within several hours p.i., the biggest difference in spleen colony counts was observed in Itgax-cre⁺Ifngr1^f/f mice at day 3 after Listeria infection. It was reported by others that administration of either rIL-12 or anti-IL-12 Ab into Listeria infected mice produced effects on bacterial counts after day 3 (51). This timing agrees with ours and supports the conclusion that defects in early IL-12 production require a certain time window before they manifest changes in bacterial burdens. The basis of the transient nature of the initial IL-12 production remains unclear. Disappearance of CD8α⁺ DCs has been reported to occur at approximately 18 hr post-infection (21). It is also possible that IL-12 production may be actively inhibited by IL-10, a powerful inhibitor of IL-12 produced by DCs (52). It is noteworthy that there is production of IL-10 in splenocytes in the first 9 hr of infection. Strikingly the cellular source of IL-12 shifts, in a relatively short period of time, from CD8α⁺ DCs to Ly6C^hi inflammatory monocytes, whose vastly greater numbers, induces higher levels of IFNγ from NK/NKT cells.

The role of NK1.1⁺ cells in host defense against Listeria has been seemingly controversial. We and others found that IFNγ produced by NK cells plays an important role during Listeria infection by activating mononuclear phagocytes (6–11). On the other hand, it has been proposed that NK1.1⁺ cells have a detrimental role in listeriosis by studies showing enhanced clearance of Listeria after mAb depletion of NK1.1⁺ cells (53, 54) or in Jα18^−/− mice lacking the majority of NKT cells (55). This apparent discrepancy can be explained by one particularly novel finding made in this study using Itgax-cre⁺Ifngr1^f/f mice showing that the first wave of IFNγ/IL-12 from NK/NKT cells and CD8α⁺ DCs contribute to the clearance of the Listeria infection mainly by controlling IL-4 production from NKT cells (and possibly other cells such as basophils, eosinophils, mast cells, and innate lymphoid type 2 cells (56)). A rapid burst of IL-4 from splenic NKT cells was previously observed in C57BL/6 mice, which peaked at 30 minutes p.i. and disappeared by 3 hr (57) and neutralization of this transient IL-4 resulted in increased L. monocytogenes resistance in C57BL/6 mice (58). Thus, depletion of IL-4 producing NKT cells by anti-NK1.1 mAb treatment or genetic deficiency of NKT cells possibly rendered host more resistant to infection while loss of IFNγ producing NK/NKT cells are rapidly compensated by IFNγ producing MP T cells. A recent study using adoptive transfer of NK and MP T cells into IFNγ^−/− mice has shown that MP T cells co-localize with Listeria and macrophages while NK cells do not thus providing the latter with a spatial advantage in mediating IFNγ-dependent clearance of Listeria (59). Our data support a model wherein early IL-12 production from CD8α⁺ DCs not only promotes IFNγ production from NK/NKT cells but also suppresses IL-4 production from NKT cells. Whether the second wave of IL-12 produced by Ly6C^hi monocytes also participates in down-regulating IL-4 production is not known.

An issue to note is that the interactions between NK cells and DCs are dependent upon both cell contact as well as cytokine production (60, 61). Interestingly, Xu et al. reported that IFNγ production from NK cells is stimulated via triggering of TNFR2 on NK cells via membrane-associated TNFα, but not via soluble TNFα (62). Our results support this concept because whereas we could inhibit early IFNγ and IL-12 production using mAbs that block TNF receptors, we could not inhibit early IFNγ production using our TN3-19.12 mAb (63) that neutralizes soluble- but not membrane-associated forms of TNFα (data not shown). Thus, we suggest that reciprocal activation between NK/NKT and CD8α⁺ DCs is likely to occur in infectious foci containing both cell types via direct cellular contact as well as cytokine secretion (20).

Our studies and those of others thus indicate at least two distinct functions for CD8α⁺ DCs in the innate immune response to L. monocytogenes. First, CD8α⁺ DCs play an obligate role in transporting L. monocytogenes into the PALS and are thus required for establishing a productive infection (19, 21). Second, as shown here, CD8α⁺ DCs play a critical role in producing the first IL-12 that initiates an IFNγ and IL-12 dependent amplification loop. The selective abrogation of IFNγ responsiveness in CD8α⁺ DCs does not impede L. monocytogenes transport, spatial regulation of innate cell clustering or L. monocytogenes proliferation. This result demonstrates that the two critical functions of CD8α⁺ DCs are largely independent of one another.

In sum, this study reveals that NK/NKT cell production of IFNγ and subsequent IL-12 production by CD8α⁺ DCs are critical initiators of the innate response to L. monocytogenes and thus illustrates how genetically homogeneous mice with tissue-selective defects in IFNγ responsiveness help refine our understanding of IFNγ’s physiologic roles in vivo.