|Home | About | Journals | Submit | Contact Us | Français|
Dendritic cells (DCs), monocyte and/or macrophages initiate host-protective immune responses to intracellular pathogens in part through interleukin-12 (IL-12) production, although the relative contribution of tissue resident versus recruited cells has been unclear. Here we showed that after intraperitoneal infection with Toxoplasma gondii cysts, resident mononuclear phagocytes are replaced by circulating monocytes that differentiate in situ into inflammatory DCs (moDCs) and F4/80+ macrophages. Importantly, NK cell-derived interferon-γ (IFN-γ) was required for both the loss of resident mononuclear phagocytes and the local differentiation of monocytes into macrophages and moDCs. This newly generated moDC population and not the resident DCs (or macrophages) served as the major source of IL-12 at the site of infection. Thus, NK cell-derived IFN-γ is important in both regulating inflammatory cell dynamics and in driving the local differentiation of monocytes into the cells required for initiating the immune response to an important intracellular pathogen.
Tissue resident dendritic cells (DCs) respond to pathogens and migrate to lymphoid organs where they present antigen to T cells (Banchereau and Steinman, 1998; Heath and Carbone, 2009). Committed DC precursors (Pre-cDCs) migrate from the bone marrow (BM) and give rise to conventional DCs (CD8α+ and CD8α− cDCs) in lymphoid tissue as well as to other DC subsets in non-lymphoid tissues (Bogunovic et al., 2009; del Hoyo et al., 2002; Liu et al., 2009; Naik et al., 2007; Varol et al., 2009). Monocytes also differentiate into DCs both in vitro and in vivo (Bogunovic et al., 2009; Cheong et al., 2010; Jakubzick et al., 2008; Randolph et al., 1999; Sallusto and Lanzavecchia, 1994; Varol et al., 2009)
Murine CX3CR1loCCR2+Ly6Chi inflammatory monocytes are recruited to sites of inflammation where they differentiate into inflammatory macrophages, DCs (moDCs) and into TNF and inducible nitric oxide synthase (iNOS)-producing DCs (TipDCs) (Geissmann et al., 2003; Gordon and Taylor, 2005; Leon et al.; Nakano et al., 2009; Serbina et al., 2003). The ability of DCs to initiate immune functions may depend on upstream signals regulating their differentiation and migration. The nature of these critical inflammatory signals and how they function in physiological settings to promote DC differentiation remain poorly understood.
The induction of T helper 1 (Th1) cell responses against intracellular pathogens is highly dependent on class specific signals delivered by DCs. In particular, IL-12, a cytokine produced by DCs and other myeloid cells, is a driver of Th1 cell development (Trinchieri, 2003). DCs produce IL-12 largely in response to signaling by pattern recognition receptors through NF-κB dependent pathways. We showed that the production of IL-12 and the transcription of the genes that encode both IL-12p35 and p40 are markedly enhanced in the presence of IFN-γ (Ma et al., 1996). Indeed, IFN-γ has been proposed as a pre-requisite signal for IL-12 synthesis (Abdi et al., 2006).
Resistance to the protozoan pathogen Toxoplasma gondii is critically dependent on IL-12 and for this reason the T. gondii mouse infection model has become a powerful tool for studying the role of the cytokine in innate resistance. Macrophages, neutrophils, DCs and inflammatory monocytes are all capable of producing IL-12p40 in response to stimulation with T. gondii products (Bliss et al., 1999; Dunay et al., 2008; Reis e Sousa et al., 1997). Interestingly, splenic CD8α+ DCs stimulated with either live T. gondii tachyzoites or a soluble parasite extract (STAg) rapidly secrete IL-12 and this response does not require priming by IFN-γ, in contrast to macrophages that fail to make IL-12 in the absence of IFN-γ exposure (Reis e Sousa et al., 1997).
Both CD11c+ DCs and inflammatory monocytes play key roles in host resistance to T. gondii (Liu et al., 2006; Robben et al., 2005). CD11c+ cell depletion or deletion of Myd88 on CD11c+ cells provided evidence that the IL-12 needed for control of infection is derived from DCs (Hou et al., 2011; Liu et al., 2006). These findings raise the possibility that the inflammatory monocytes required for host resistance are precursors of IL-12 producing DCs. Thus, the origin of the IL-12 producing cells at the site of infection and the relative contribution of tissue resident and newly recruited cells following parasite invasion are important issues.
Here we investigated the fate of tissue-resident populations and the origin of IL-12 secreting cells at the infection site as well as the nature of inflammatory signals involved in their differentiation. We show that tissue-resident mononuclear phagocytes fail to produce IL-12 in response to pathogen invasion and instead are replaced by circulating Ly6C+ monocytes that differentiate into both inflammatory macrophages and IL-12 producing DCs. Importantly, we demonstrate that NK cell-derived IFN-γ is critical for both monocyte differentiation at the site of infection as well as for the loss of the resident cell populations. These findings reveal a previously unappreciated requirement for IFN-γ in regulating the cellular dynamics and inducing the functional differentiation of monocytes into cells appropriate for initiating the immune response to parasite invasion.
To define changes in the composition of the cellular infiltrate following parasite challenge, we utilized the well-established intraperitoneal (i.p.) infection model, which provides a defined and sterile site of infection and allows us to readily distinguish resident from immigrant cell populations. IL-12p40YFP (Yet40) reporter mice were infected i.p. with 20 cysts of the ME49 avirulent strain of T. gondii, or in some experiments, ME49 engineered to express the red fluorescent protein (ME49RFP) and peritoneal exudate cell (PEC) composition followed for 72h. We observed a marked change in the composition of peritoneal infiltrates as infection progressed (Figure 1A and S1A). F4/80hiCD11bhiMHC-II− resident peritoneal macrophages gradually disappeared and were replaced by Ly6C+ cells that expressed low to intermediate amounts of F4/80 and intermediate to high amounts of MHC-II (Figure S1A). These Ly6C+ cells displayed the phenotypic characteristics of activated inflammatory macrophages (F4/80intMHC-II+CD11b+CD11c−CD68+Ly6G−7/4+; data not shown) and closely resembled the myeloid population shown in previous studies to be recruited to the site of T. gondii infection (Dunay et al., 2008; Robben et al., 2005; Rosas et al., 2010).
Neutrophils (defined as Ly6GhiLy6Clo), NK cells (NK1.1+TCRβ−CD11cloMHC-II−) and T cells (TCRβ+NK1.1−) were also recruited to the peritoneal cavity by 72 h post-infection (Figure 1A). CD11c+MHC-II+ cells were present in both naïve and infected mice, however these cells appeared to express higher MHC-II and decreased CD11c after infection (Figure 1B). Gating on total ME49RFP+ infected cells showed that within each cell type only a fraction harbored parasites, consistent with the ability of T. gondii to infect virtually any nucleated cell type (Figure S1B).
Expression of CD11c and high levels of MHC-II are commonly used to define DCs, however CD11c can also be expressed on other cell types. To determine if the peritoneal CD11c+MHC-II+ cells observed here display the morphological and functional properties of DCs, we flow cytometry-sorted the different PEC populations in naïve and infected animals and examined their characteristics. Two populations were sorted from naïve PEC utilizing the following gating criteria: cells previously defined as resident peritoneal macrophages (lymphocyte-lineage negative (lin−) F4/80hiMHC-II−; Gate#1), and cells with DC markers (lin−CD11c+MHC-II+; Gate#2). Four populations were sorted from 72 h infected PEC: cells with DC markers (lin− Ly6G−CD11c+MHC-II+; Gate#3); two populations gated based on the Ly6C expression (lin−Ly6G−F4/80intLy6Cint; Gate#4) and (lin−Ly6G−F4/80int-loLy6Chi; Gate#5); and IL-12p40YFP+ cells gated on total live cells (Gate#6) (Figure 1A). As depicted in Figure 1C, Giemsa staining showed that the F4/80hiMHC-II− cells displayed a characteristic macrophage morphology, while CD11c+MHC-II+ cells from either naïve or infected animals were smaller in size, irregularly shaped and exhibited the cytosolic projections typical of DCs. The F4/80intLy6Cint cells represented a fairly heterogeneous population, in which cells with monocyte or macrophage-like morphology as well as some DC-like cells were observed. The F4/80int-loLy6Chi population was more homogeneous and exhibited morphological characteristics of highly activated monocytes or macrophages with heavily vacuolated cytosol. Finally, YFP+ cells displayed the morphological appearance of DCs.
We next tested the ability of the purified populations to stimulate T cell proliferation in a mixed leukocyte reaction (MLR), a hallmark of DC function. Only CD11c+MHC-II+ from naïve or infected animals and YFP+ cells induced a strong MLR, and although F4/80int-loLy6Chi and F4/80intLy6Cint cells also expressed MHC-II molecules (Figure 1A), these cells were very weak or incapable of inducing T cell proliferation (Figure 1D). Therefore, we concluded that based on their phenotypic, morphological and functional characteristics, the CD11c+MHC-II+ cells detected in PEC represent DCs.
We also followed these alterations in cellular composition into the late chronic phase of infection (two months post-parasite challenge). While the total number of cells continued to increase during the first 7 days, there was an unexpected drop in cellularity during the second week of infection when parasites are cleared from the site. This cell loss persisted for the length of the experiment (Figure 1E left panel) and normal cellular homeostasis was never restored including the return of the original F4/80hi macrophage population. An exception was the T lymphocytes which after reaching a peak on day 14 assumed a steady state level (Figure 1E right panels and S1A).
Substantial IL-12p40YFP expression relative to naïve animals was first detected in PEC at 72 h post-infection but not at earlier time points (Figure 1A and S1C). The onset of cytokine production correlated with increased parasite burden as evidenced in the frequency of ME49RFP+ cells (Figure S1C), however, the IL-12p40YFP+ cells were RFP− indicating that direct infection is not required for IL-12 production (Figure S1C).
Phenotypic analysis of YFP+ cells at 72 h revealed that IL-12p40 producing cells expressed CD11c and high MHC-II but only low F4/80 and Ly6C (Figure 2A). Although CD11c+MHC-II+ cells were already present in naïve PEC, no significant increase in CD11c+MHC-II+YFP+ cells could be detected during the initial 48 h assay period indicating that these resident cells fail to respond to infection with the production of IL-12p40 during this time (Figure 2B). The CD11c+MHC-II+YFP+ DCs appearing in PEC 72 h post-infection were largely CD11b+CD8α−, and only low frequencies of CD11bloCD8α+YFP+ and CD103+CD8α−YFP+ were detected (Figure 2C–D).
Unexpectedly, not only were YFP+CD8α+ DCs present in low frequency in PEC, they were also virtually absent in the spleen at 6, 24, 48 and 72 h post-infection (Figure S2A and not shown). The latter observation contrasts with STAg-induced IL-12p40 production, which is exclusively mediated by splenic CD8α+ DCs and peaks at 4 – 6 h post-injection (Reis e Sousa et al., 1997). Indeed, when mice were infected i.p. with tachyzoites of the same ME49 T. gondii strain used for cyst injection, IL-12p40YFP+CD8α+ DCs could be detected in spleen as early as 6 h post-injection (data not shown), indicating that tachyzoite in contrast to cyst infection triggers a response resembling that elicited by STAg challenge.
To further assess the role of CD8α+ DCs in initiating peritoneal IL-12p40 responses, mice were injected before and after infection with an anti-CD8α MAb that effectively depletes splenic CD8α+ DCs (Bedoui et al., 2009). This treatment efficiently deleted CD8α+CD205+ DCs in both spleen and PEC 3 d post-infection (Figure 2E middle and Figure S2B). While anti-CD8α treatment completely abolished STAg-induced IL-12p40 production in spleen (Figure S2C), it failed to affect IL-12p40 responses at the peritoneal infection site (Figure 2E). Thus, following i.p. infection with T. gondii cysts IL-12p40 is produced in situ primarily by CD11b+CD8α− DCs, and CD8α+ DCs are neither a major source of the cytokine nor play a key role in the initiation of the response.
To test whether the local IL-12 response by CD11b+CD8α− DCs requires IFN-γ priming, yet40 animals were injected with anti-IFN-γ neutralizing antibody or isotype control on the day prior to infection and PEC were harvested 72 h post-parasite challenge. In a parallel set of experiments, Ifngr1−/− yet40 mice generated by crossing yet40 with Ifngr1−/− animals were also infected. We first performed a phenotypic analysis on the PEC in the anti-IFN-γ-treated and Ifngr1−/− mice at 72 h post-infection (Figure 3A and Figure S3A; animals that received isotype control showed no difference compared to the un-injected infected yet40 mice presented in Figure 1A (data not shown). Major differences in the overall composition of the different PEC populations were observed compared to PEC obtained from infected IFN-γ sufficient mice. Both the Ifngr1−/− and anti-IFN-γ-treated mice failed to show the major loss in the peritoneal F4/80hi cells previously observed in infected control animals (Figure 3A and S3A compare to Figure 1A and S1A), ruling out a possible dilution of resident macrophages due to recruitment of neutrophils and monocytes. In addition, the Ly6Chi monocytes recruited to the peritoneal cavity in the former animals displayed a unique F4/80− phenotype, and the Ly6CintF4/80int cells previously documented in infected control mice were no longer detected (Figure 3A and S3A). Morphologically the F4/80hi and Ly6ChiF4/80− populations resembled resident macrophages and resting monocytes respectively and as expected neither population displayed the APC activity exhibited by CD11c+MHC-II+ cells (Figure 3B–C).
Nevertheless, no major alterations were detected in the frequencies of CD11c+MHC-II+ cells in IFN-γ-deficient versus control mice. The total purified CD11c+MHC-II+ population in the in vivo absence of IFN-γ displayed unaltered morphology and APC function (MLR response) compared to wild type animals (Figure 3B–C). However, relative to YFP+ or bulk YFP− CD11c+MHC-II+ populations from infected IFN-γ sufficient mice, these cells failed to express Ly6C and expressed higher CD11c similar to DCs obtained from naïve animals (Figure 3D). Moreover, the frequency of CD11c+MHC-II+YFP+ cells was significantly reduced in both anti-IFN-γ-treated yet40 and Ifngr1−/− yet40 mice compared to infected yet40 control mice (Figure 3E–F). That this reduction in YFP signal reflects decrease in IL-12 protein secretion was confirmed by showing significant decreases of both IL-12p70 heterodimer and IL-12p40 chain in the peritoneal lavage measured by ELISA (Figure 3G–H). Consistent with previous studies, the splenic CD8α+ IL-12p40 response to STAg in Ifngr1−/− yet40 or anti-IFN-γ-treated yet40 mice was comparable to that observed in wild type animals (Figure S3B and data not shown). Thus, IFN-γ plays a major role in shaping inflammatory infiltrates and IFN-γ-priming is required for peritoneal CD11b+CD8α− DCs to produce IL-12 in response to T. gondii infection
NK cells play a major role in resistance to T. gondii infection as an early source of IFN-γ (Denkers et al., 1993; Goldszmid et al., 2007; Hunter et al., 1994). To determine whether NK cells serve as the source of IFN-γ required for IL-12p40 production by peritoneal CD11b+CD8α− DCs, yet40 mice were injected with anti-AsialoGM1 to deplete NK cells prior to infection and frequencies of IL-12p40YFP+ DCs were assessed in PEC 72 h after parasite challenge. Anti-AsialoGM1 treatment significantly decreased the frequency of IL-12p40YFP+ DCs (Figure 3I–J) to levels comparable with that observed in anti-IFN-γ treated or Ifngr1−/− animals (Figure 3A–B). Similar results were observed when NK cells were depleted using anti-NK1.1 (Figure S3C). To rule out an effect of IFN-γ on the recruitment of NK cells to the site of infection, purified NK cells from Rag1−/− congenic mice were injected i.p. into anti-IFN-γ-treated yet40 animals on the day of infection and IL-12 production was measured in PEC 72 h later. As shown in Figures 3H–J, injection of NK cells directly into the site of infection did not restore IL-12 production in anti-IFN-γ-treated animals. The above observations establish NK cells as the major source of the IFN-γ that promotes DCs IL-12 production during initial infection.
To address the fate of the resident cell populations and to test whether peritoneal resident DCs were the source of IL-12p40, we adoptively transferred total PEC from naïve CD45.2+ yet40 animals i.p. into uninfected or infected CD45.1+ congenic hosts and evaluated frequencies of IL-12p40YFP+ cells at different times thereafter. A reduced number of CD45.2+ transferred cells was found 72 h post-infection compared to uninfected controls (p = 0.0016; Figure 4A). The F4/80hi and CD11c+MHC-II+ cell populations largely disappeared when transferred into infected compared to naïve hosts and no YFP+ cells were detected in either group (Figure 4B). Almost all CD45.2+ grafted cells found in PEC from infected mice were MCH-II+CD19+CD11blo B1 lymphocytes (Figure S4A). Donor-derived macrophages and DCs were not detected in draining lymph nodes or spleens of infected animals and the CD45.2+ graft-derived cells migrating into those tissues were mostly CD19+ with some TCRβ+ cells (Figure S4B and data not shown).
The kinetics of disappearance of transferred yet40 peritoneal macrophages and DCs in infected animals paralleled the gradual changes in PEC composition and disappearance of the corresponding resident populations as the infection progressed (Figure 4C and Figure S1A). Importantly, F4/80int-loMHC-II+ cells of donor origin were never detected in PEC from infected recipients (Figure 4B–C), indicating that those cells exclusively represent newly recruited cells rather than phenotypic variants of activated resident macrophages. This conclusion was strengthened by data obtained in Ccr2−/− mice that fail to recruit inflammatory monocytes to the site of T. gondii infection (Dunay et al., 2008; Robben et al., 2005). In these mice F4/80hi cells disappeared from PEC by 72 h post-infection but in contrast to Ccr2+/+ mice, were not replaced by F4/80int-loMHC-II+ cells (Figure 4D). In addition, PEC from infected Ccr2−/− mice also showed reduced numbers of CD11c+MHC-II+ cells (Figure 4D) but similar numbers of recruited neutrophils compared to wild type animals. Thus, resident macrophages or DCs do not appear to contribute to IL-12 production but rather are replaced by recruited inflammatory monocytes that acquire DC characteristics and are the main source of IL-12p40.
The infection of animals treated with anti-IFN-γ or animals lacking Ifngr1 provided clear evidence that in situ IL-12 production requires IFN-γ-priming. In addition, PEC transfer experiments indicated that IL-12p40YFP+ cells are not derived from resident cells. To determine whether IL-12p40YFP+ cells originate from circulating monocytes and IFN-γ is necessary for their initial recruitment or later for their ability to produce IL-12, we utilized a monocyte adoptive transfer approach. BM monocytes from CD45.2+ yet40 naïve donors were sorted as lin−CD115+Ly6ChiCD11bhi and 1.8 – 2 × 106 purified cells were injected i.v. into CD45.1+ hosts that were either uninfected or 24 h post-infection, and treated or not with anti-IFN-γ. The transferred marrow monocytes were YFP−CD11c−MHC-II−c-kit−F4/80loFlt3lo (data not shown). Seventy-two hours after cell transfer, PEC were harvested and analyzed by flow cytometry for the presence of CD45.2+ donor-derived cells. Grafted CD45.2+ cells were found in the peritoneal cavity of intact hosts at steady state; however, in the absence of an inflammatory stimulus, IL-12p40YFP+ cells were not detected (Figure 5A–C). Consistent with their preferential homing to inflamed tissues (Geissmann et al., 2003; Varol et al., 2007), recruitment of grafted monocytes was significantly enhanced by infection, and even more so in infected mice receiving anti-IFN-γ treatment (Figure 5A–B). The latter observation could be explained by enhanced inflammation induced by the higher parasite burden observed in the absence of IFN-γ. Despite the elevated numbers of recruited graft-derived monocytes in anti-IFN-γ-treated infected hosts, frequencies of IL-12p40YFP+ cells were significantly lower compared to infected control recipients (Figure 5C).
Phenotypic analysis of grafted cells revealed that in PEC of infected animals the majority of CD45.2+ cells up-regulated MHC-II+, whereas a large proportion of donor-derived cells remained MHC-II− in uninfected and in anti-IFN-γ-treated infected recipients (Figure 5E). Moreover, a marked difference was observed in F4/80 expression. The majority of grafted cells in PEC of infected mice expressed intermediate levels of F4/80, in contrast to naïve and anti-IFN-γ-treated animals in which donor cells showed markedly lower F4/80 expression (Figure 5D–E). Together these data indicate that IFN-γ is dispensable for the recruitment of monocytes to the site of inflammation but is necessary for their phenotypic and functional maturation into inflammatory macrophages and IL-12p40 producing DCs.
IL-12 production by DCs in response to T. gondii has been shown to be MyD88-dependent (Scanga et al., 2002). To determine if MyD88-mediated signaling is also needed for the phenotypic and functional maturation of moDCs and to establish if the IFN-γ requirement is cell-intrinsic, we transferred monocytes from naïve yet40, Ifngr1−/− yet40 or Myd88−/− yet40 (generated by crossing yet40 with Myd88−/− mice) animals into CD45.1+ hosts one day post-T. gondii infection and evaluated the ability of monocyte-derived cells to produce IL-12p40 72 h later. No differences were observed in the number of grafted monocytes that entered the infection site regardless of their donor origin (Figure 6A). However, IL-12p40 production was abolished when donor monocytes were obtained from Myd88−/− yet40 or Ifngr1−/− yet40 (Figure 6B). Nevertheless, in direct contrast to the cells from Ifngr1−/− yet40, the majority of graft-derived cells from Myd88−/− yet40 donors up-regulated MHC-II molecules and expressed F4/80 at intermediate/low levels similar to cells derived from grafted WT monocytes (Figure 6C). The above findings demonstrated a cell-intrinsic requirement for IFN-γ in the up-regulation of MHC-II and F4/80 and expression of IL-12p40 by monocyte-derived cells and showed that, in contrast, MyD88 signaling in cooperation with IFN-γ, is necessary for IL-12p40 production but not for the phenotypic changes of these cells.
To confirm the role of IFN-γ in inflammatory cell maturation and to further study the different PEC populations and their BM monocyte precursors, we analyzed the expression of 122 genes associated with mononuclear phagocyte cell lineage development and functions (Table S1) utilizing NanoString technology (Geiss et al., 2008). PEC from naïve or infected mice treated or not with anti-IFN-γ were sorted and BM monocytes from naïve mice were purified. Samples from three independent experiments were analyzed for gene expression. Non-supervised hierarchical clustering and principal component analysis (PCA) were performed. Biological replicates within the same sort gate were found to cluster tightly and no overlaps among the different sorted populations was observed, confirming that each gate indeed represents a distinct population (Figure 7A–B). Comparison of individual populations showed that the F4/80hi cells found in naïve and anti-IFN-γ-treated infected animals clustered together but were distant from all other cell populations studied arguing that they represent a distinct population. This finding is consistent with our previous observation that resident peritoneal macrophages are not precursors of the F4/80int-lo cells found in the infected control animals (Figure 7A–B).
As expected, F4/80intLy6Cint and F4/80int-loLy6Chi cells from infected mice clustered together and with the F4/80−Ly6Chi cells found in anti-IFN-γ-treated infected animals (Figure 7A–C). Although gene expression profiles of the BM monocytes and the Ly6C+ PEC displayed several differences, the Pearson coefficient correlation analysis among all populations indicated that Ly6ChiF4/80− PEC from anti-IFN-γ-treated infected mice had the highest correlation with the monocytes followed by the F4/80int-loLy6Chi and F4/80intLy6Cint cells from infected mice (data not shown), suggesting that these cells represent different differentiation/maturation stages of inflammatory monocytes.
CD11chiMHC-IIhi DCs obtained from naïve and anti-IFN-γ-treated infected animals clustered together but separately from the IL-12 positive and negative DCs from infected mice, which were closely related to each other (Figure 7A–B and 7D). As expected because of their monocytic origin, both DCs populations from infected mice expressed Ly6c1 (Ly6C), Csf1r (CD115), and Emr1 (F4/80) transcripts, albeit at a lower level compared to that of the monocyte or inflammatory macrophages populations, and consistent with flow cytometry analysis, also displayed lower Itgax (CD11c) transcript accumulation than DCs from naïve or anti-IFN-γ-treated infected animals (Figure S6). The IL-12 producing subset of DCs expressed decreased Ciita (MHC-II transactivator), increased Ccr7, Cd86, Cd70 and increased levels of Irf8 (Figure S6), indicative of a more mature phenotype. Together these data support our hypothesis that the DCs in infected animals are not resident, but instead derive from newly recruited monocyte precursors and require IFN-γ for their phenotypic and functional maturation.
NK cells are an innate source of IFN-γ and drive optimal Th1 cell responses in vivo (Martin-Fontecha et al., 2004; Scharton and Scott, 1993). We showed that animals depleted of NK cells or with a defect in NK cell licensing resulting in reduced IFN-γ production displayed impaired Th1 cell responses and succumbed to T. gondii infection (Goldszmid et al., 2007). IL-12 plays a major role in amplifying the NK cell IFN-γ response to T. gondii (Gazzinelli et al., 1993; Hunter et al., 1994), though other co-factors (e.g. TNF and IL-1) and the NK cell activating receptor NKG2D may also be involved (Goldszmid et al., 2007; Guan et al., 2007; Hunter et al., 1995; Sher et al., 1993). Here we show that IL-12 production by DCs at the site of infection requires NK cell-produced IFN-γ that unexpectedly also regulates the differentiation of circulating monocyte precursors into DCs.
These findings contrast with the IL-12 response of splenic DCs to T. gondii tachyzoites or STAg that is mediated exclusively by the CD8α+ DC subset and does not require IFN-γ priming. The enhanced susceptibility of Batf3 deficient mice (which lack both CD8α+ and CD103+ DCs) to T. gondii tachyzoite infection suggest that CD8α+ DCs may represent a source of IL-12 in systemic infection (Mashayekhi et al., 2011). This hypothesis is consistent with our previous observation that IFN-γ−/− mice that are unable to control infection and have an extremely high systemic parasite burden develop an unimpaired IFN-γ-independent IL-12 response (Scharton-Kersten et al., 1996) likely due to early systemic activation of CD8α+ DCs. However, the results with antibody depletion of CD8α+ DCs presented here indicate that the CD8α+ DC subset does not play a major role in the IL-12 response at the site of infection. Thus, there is a fundamental dichotomy in the role of IFN-γ in the regulation of IL-12 production by the CD8α+ and CD8α− DC subsets, a property that may stem from the constitutive expression of Irf8, a transcriptional factor important for expression of Il12 genes, in the former cells (Aliberti et al., 2003).
We found that following T. gondii infection the resident DC population in the peritoneum is replaced by de novo generated moDCs, which in turn become the major source of IL-12. This process of cellular exchange and differentiation is regulated by IFN-γ signaling. It is likely that, the IL-12+ moDCs s recruited to the site of T. gondii infection stimulate local T cell expansion and IFN-γ production as described in other systems (Iijima et al., 2011; Wakim et al., 2008).
The importance of inflammatory monocytes in host resistance to T. gondii infection was first revealed by Sibley and colleagues (Dunay et al., 2008; Robben et al., 2005) who showed that Ccr2−/− mice failed to recruit inflammatory monocytes to the site of infection and were unable to control parasite replication. In those studies, the inflammatory monocytes were characterized as F4/80+Ly6ChiCD11c− and expressed iNOS, TNF and IL-12. Because they lacked CD11c expression, the authors argued that the monocytes involved failed to differentiate into DCs. The discrepancy with our data identifying the IL-12 producing cells in early T. gondii infection as moDCs can be explained by the fact that in the previous studies the cells analyzed were gated as Ly6ChiF4/80+, a phenotype that as shown here corresponds to monocyte-derived inflammatory macrophages. The moDCs described in our study express Ly6C and F4/80 at significantly lower levels compared to inflammatory macrophages and thus would have been excluded from the gate. Thus, our data confirm that inflammatory monocytes give rise to macrophages that control early parasite replication but further demonstrate that they are precursors of IL-12 producing DCs.
The data from both our PEC transfer and Ccr2−/− mice infection experiments indicate that resident macrophages do not undergo maturation, but instead decreased in number as a consequence of infection and are fully replaced by newly recruited cells. Indeed, we also showed that when monocytes differentiate into macrophages, IFN-γ in addition to inducing MCH-II is responsible for the early up-regulation of F4/80 expression and later decline as the infection progresses. While our findings attribute a major role to immigrant cells in the initiation of the IL-12 response to T. gondii, resident cells must also contribute in signaling the recruitment of inflammatory monocytes precursors. In that regard, Robben et al showed that resident macrophages produce a major CCR2 ligand, CCL2 (MCP1) in response to T. gondii (Robben et al., 2005) and we also observed increased levels of Ccl2 transcripts in F4/80hi cells following infection of anti-IFN-γ treated animals (data not shown).
Although we did not detect PEC-transferred macrophages or DCs in the draining LN or spleen, it remains to be determined whether these cells die in situ or migrate to other sites such as the omentum-associated lymphoid tissue. Regardless of the mechanism, the results with anti-IFN-γ treated and Ifngr1−/− mice indicates that their loss is dependent on IFN-γ. Indeed, the gene expression analysis clearly showed that the F4/80hi population found in anti-IFN-γ treated infected mice correspond to resident macrophages and that the CD11chiMHC-IIhiLy6C− cells likely represent peritoneal resident DCs.
The major conclusion of the present study is that following T. gondii infection resident mononuclear phagocytes are exchanged for newly recruited inflammatory macrophages and IL-12 producing moDCs; IFN-γ regulates both this cellular exchange as well as the monocyte differentiation process. Interestingly, we recently observed a similar requirement for IFN-γ in the differentiation of monocytes into IL-12-producing DCs in the lungs of mice infected with Mycobacterium tuberculosis (Mayer-Barber and Sher unpublished data). Thus, the cellular exchange process regulated by IFN-γ may be of particular importance in host defense against intracellular pathogens where unprimed tissue resident cells may be incapable of Th1 response initiation.
Unlike the regulation of the Th1 response to intracellular pathogens, in helminth-induced Th2 inflammation, IL-4 promotes in situ proliferation and alternative activation of tissue resident macrophages (Jenkins et al., 2011). Thus, the archetypal Th1 and Th2 cytokines IFN-γ and IL-4 condition the inflammatory microenvironment as an initial step in the induction of appropriate innate and adaptive immune responses. Targeting this step in the initiation of local immune responses may be a powerful approach for manipulating their outcome and may be of particular use in cancer, where reprogramming the tumor microenvironment represents an important strategy for immunotherapy.
C57BL/6 (CD45.2) and BALB/c mice were purchased from Taconic Farms (Germantown, NY). B6.SJL CD45a (Ly5a)/Nai (CD45.1) and B6.SJL-Rag1−/− were provided by Taconic Farms from the NIAID Animal Supply Contract. Ifngr1-deficient (Ifngr1−/−) and Ccr2-deficient (Ccr2−/−) mice were purchased from the Jackson Laboratory (Bar Harbor, ME). C57BL/6 IL-12p40eYFP (yet40) (Reinhardt et al., 2006) mice were kindly provided by R. Locksley (University of California, San Francisco, CA). Myd88 deficient animals (Myd88−/−) were originally obtained from S. Akira (Osaka University, Osaka, Japan). All mice were maintained in a NIAID, NIH (Bethesda, MD) animal care facility under specific pathogen-free conditions and treated in accordance with the regulations and guidelines of the Animal Care and Use Committee of the National Institutes of Health. Age (8–12 wk) and sex matched animals were used in all experiments.
Cysts of the avirulent ME-49 strain of T. gondii were obtained from the brains of chronically infected C57BL/6 mice. For experimental infection, animals received 20 ME-49 or ME-49 expressing the red fluorescent protein (Oldenhove et al., 2009) (ME-49RFP; kindly provided by M. Grigg, NIAID, NIH) cysts in a volume of 0.5 ml via the i.p. route. Soluble tachyzoite antigen (STAg) was prepared as previously described (Grunvald et al., 1996).
PEC from naïve, infected or anti-IFN-γ treated infected yet40 animals were obtained by peritoneal lavage with 10 ml of ice-cold PBS/2 mM EDTA washed once and incubated with the appropriate antibody cocktail. Two, three and four way sorting were performed using a FACSAria instrument (BD Bioscience). Purified populations were used for MLR assay, cytospin or NanoString analysis. For phenotypic analysis live/dead cells were identified using Live Dead Fixable Blue Dead Cell stain kit (Invitrogen). Data was acquired using an LSR II flow cytometer (BD Bioscience) and analyzed with FlowJo software (Tree Star, Ashland, OR).
Peritoneal lavage was performed with 3 ml of ice-cold PBS, cells were spun and supernatants stored at −40°C. The amount of IL-12/23p40 or IL-12p70 was measured by ELISA using BD Bioscience or R&D Systems standard sandwich ELISA kits respectively.
To deplete NK cells mice were injected with 50 μl of anti-AsialoGM1 (Wako Pure Chemical, Richmond, VA) or with 0.5 mg of anti-NK1.1 (clone PK136) intravenously (i.v.) on day −1 and +3 relative to infection. NK cell depletion was evaluated by flow cytometry analysis. >90% depletion of DX5+NK1.1+CD3− in blood and PEC and >75% depletion in spleen was observed. To deplete CD8α+ DCs mice were injected i.v. with anti-CD8 (clone 2.43) every two days starting one day prior to infection. CD8α+ DCs depletion was assessed by flow cytometry using anti-CD11c, I-Ab, CD8α, CD205, CD11b and CD24. In a different set of experiments mice were injected i.v. with 1 mg of anti-IFN-γ mAb (clone XMG-6) on day −1 and +3 relative to infection. Control mice received similar amounts of normal rabbit serum or the corresponding isotype controls.
BM was obtained from yet40, Myd88−/− yet40 or Ifngr1−/− yet40 mice and CD115+ cells were MACS-enriched (Miltenyi Biotec Inc) according to the manufacturer’s instructions. CD115-enriched cells were then stained with the corresponding mAb cocktail and monocytes were flow cytometry-sorted as lineage−CD115+Ly6ChiCD11bhi using a FACSVantage or FACSAria instrument (BD Biosciences). Monocyte purity post-sort was > 98%. Between 1.8 and 2 × 106 purified monocytes were injected i.v. into infected, anti-IFN-γ-treated infected or uninfected CD45.1+ congenic recipients. In a different set of experiments freshly isolated PEC obtained from naïve yet40 animals by peritoneal lavage were injected i.p. into infected, anti-IFN-γ-treated infected or uninfected congenic hosts. In all cell-transfer experiments donor cells were gated as CD45.2+CD45.1− and non-grafted animals were used as staining controls to ensure accurate gating. For NK cell transfer experiments, NK cells were purified from uninfected B6.SJL-RAG1−/− mice by negative selection using the NK cell isolation kit followed by positive selection with anti-DX5 MACS microbeads (Miltenyi). 2 × 106 NK cells were administered i.p. into anti-IFN-γ-treated mice on day of infection. Transferred NK cells were detected in PEC of anti-IFN-γ-treated infected mice as CD45.1+CD45.2−NK1.1+CD3− cells.
Purified BM monocytes and PEC populations obtained as described above were lysed in RLT buffer (Quiagen) and subjected to NanoString nCounter analysis system as previously described (Geiss et al., 2008). For gene expression analysis, data was normalized to the 5% least variable genes. Normalized data was then Log2 transformed and genes differentially expressed between groups were identified with one-way ANOVA model using Method of Moments (Eisenhart, 1947) followed by correction for False Discovery (Storey and Tibshirani, 2003) (Partek 6.6 software, Partek Inc., St.Louis, MO). Genes for comparisons mentioned in the text were selected if q<0.001. Heatmaps were created using average-linkage clustering of selected genes and samples. Principal component analysis was done using all genes and samples in the array. Pearson correlation between samples was done using all the genes in the dataset with Partek software.
The statistical significance of differences was determined using an unpaired, two-tailed Student’s t test. The Welch’s correction was applied when analysis of variances (F test) determined that significant differences existed. (Prism 5, GraphPad Software Inc.). Differences were considered significant when p < 0.05.
We are grateful to the NIAD Flow Cytometry core facility and in particular Thomas Moyer for performing sorting. We are also indebted to Dr Richard Locksley for kindly providing the Yet40 mice and to Dr Michael Grigg for the RFP expressing parasites. We also gratefully acknowledge Drs. Daniel McVicar, Katrin Mayer-Barber and Dan Barber for critical reading of the manuscript. This research was supported by the Intramural Research Program of the NIAID and NCI, NIH.
The authors have no conflicting financial interests.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.