Memory CD8⁺ T cells participate in innate immune responses

We next hypothesized that Lm-induced memory CD8⁺ T cells contribute to innate immune responses in immunized mice. We thus compared the activation of OT-I or endogenous LLO_91–99-H2-K^d memory CD8⁺ T cells to that of innate NK lymphocytes in mice that received primary immunization with WT Lm and were challenged with Lm lacking the expression of either Ova_257–264 or LLO_91–99 cognate epitopes (Figures 2A and S2A). While NK cells rapidly expressed IFN-γ and GrB, the proportion of memory CD8⁺ T cells expressing these functions was markedly higher than NK cells (factor of 1.5–2). Because antigen was not required, we tested whether memory CD8⁺ T cells also expressed these effector functions in mice challenged with other non-OVA expressing microbial pathogens, including the murine Cytomegalovirus (MCMV), the extracellular bacterium Streptococcus pneumoniae (S. pneumoniae) and the murine parasite Plasmodium berghei (P. berghei). Mice (B6 and B6-K^d) respectively grafted with naïve OT-I or WP20 T cells (bearing another monoclonal TCR recognizing Lm-derived p60_449–457 presented by H-2K^d) were immunized with WT Lm-Ova and challenged with each of the pathogens (Figures 2B–C and S2B–C). In all infections, and in WT BALB/c mice, the proportion and numbers of activated (IFN-γ⁺, GrB⁺) memory CD8⁺ T cells was either equivalent or higher than that of NK cells. Furthermore, activated memory CD8⁺ T cells generally expressed higher amounts (MFI) of these functions compared to activated NK cells (up to a factor of 2). The proportion of IFN-γ⁺ cells in infected organs, e.g., spleen and liver (MCMV, P. berghei) or lung (S. pneumoniae), of unimmunized versus Lm-immunized mice also showed that CD8⁺ T cells account for at least 37% of IFN-γ⁺ cells (Figures 2D–E; S2B–C and not shown). Thus in immunized animals only, CD8⁺ T cells which include Lm-specific memory cells, participate in IFN-γ⁺ innate immune responses similarly to NK cells. The transfer of purified OT-I memory cells into naïve recipient animals that were subsequently challenged with non-Ova expressing WT Lm conferred modest (factor of 3–5) but significant bacteriocidal immunity (Figure 2F).

Open in a separate window

Figure 2

Memory CD8⁺ T cells contribute to innate immune responses and protection

WT B6 mice transferred with OT-I GFP⁺ cells were immunized with WT Lm-Ova (A–F) or left unimmunized (D, E). Animals were challenged 6 wks later with WT Lm (A, F), MCMV (B, D) or Streptococcus pneumoniae (C, E). In (G, H) mice were immunized with the lymphochorionmeningitis virus (LCMV). Sixty days later, animals were challenged with WT Lm (G) or LCMV-specific H2-D^b/GP_33–41 and H2-D^b-NP_396–404 memory CD8⁺ T cells were purified and transferred into naïve B6 mice subsequently challenged with WT Lm (H). Spleens (A, B, D, G) and lungs (C, E) were harvested at indicated times, incubated with Golgi Plug and stained for CD8, CD3, LCMV tetramers and NKp46 (NK cell) cell surface markers and intracellular IFN-γ and GrB. In (A–C, G), data show the relative proportion (%) and fold increase in MFI (activated to resting cells) of IFN-γ⁺ or GrB⁺ cells amongst OT-I (black), NK (red) cells, and LCMV tet⁺ cells after gating on each cell type. Representative dot-plots and compilation of 2–3 independent replicate experiments with each dot featuring one individual mouse are shown (n=4–13). In (D, E) are shown representative dot-plots of % IFN-γ⁺ cells in unimmunized versus WT Lm-immunized mice challenged with either the murine Cytomegalovirus, MCMV (spleen) or S. pneumoniae (lung). The relative proportion of NK, CD8⁺ T and other cells amongst IFN-γ⁺ cells is shown in the pie-charts with p-values. In (F, H), flow-purified OT-I memory (~2×10⁴) (F) or LCMV-specific (~10⁵) (H) cells were transferred or not into naïve WT mice subsequently challenged with WT Lm (F:5×10⁴, H:2,000). CFUs in spleens were determined by plating 48 hrs later. Data pool 2–4 independent replicate experiments (n=4–8) with p-values.

Finally, to generalize our findings to memory CD8⁺ T cells induced by other immunizations, we infected mice with a distinct microbial pathogen, the lymphochoriomeningitis virus (LCMV Armstrong), and challenged them with WT Lm 60 days later (Figure 2G and S2D). Similarly to Lm-specific memory CD8⁺ T cells, endogenous LCMV-specific memory CD8⁺ T cells recognizing 2 distinct epitopes derived from the LCMV glycoprotein (GP_33–41) and nucleoprotein (NP_396–404) produced IFN-γ (~17%) and GrB (~45%) and upregulated activation markers (CD69, CD25, not shown), therefore extending these results to memory CD8⁺ T cells induced by acute viral infection. It is also noticeable that the proportion of memory and NK cells undergoing IFN-γ and GrB expression was equivalent (Figure S2D). Importantly and in line with these observations, naïve recipient mice adoptively transferred with LCMV-specific memory CD8⁺ T cells and subsequently challenged with WT Lm exhibited lower bacterial burden (factor of ~3) as compared to untransferred mice (Figure 2H). Thus pathogen-induced memory CD8⁺ T cells can participate to innate immunity and immunological protection.

Triggering of inflammasome and IFN-I pathways promote memory CD8⁺ T cell differentiation into effectors

We next investigated the molecular mechanisms regulating the activation of memory CD8⁺ T cells. Since activation was observed across infections with various microbial pathogens, we hypothesized that common cytokine signals are likely to control these processes. In line with others studies (Berg et al., 2003; Kambayashi et al., 2003; Kohlmeier et al., 2010; Liu et al., 2002; Yajima et al., 2005; Zhang et al., 1998), addition of combinations of IL-12 and IL-18, IFN-I or IL-15, induced the activation of memory CD8⁺ T cells in vitro with no TCR cross-linking requirement (Figure 3A). Incubation of antigen-specific memory CD8⁺ T cells (OT-I and endogenous) with IL-12 and IL-18 induced their secretion of IFN-γ, while incubation with IL-15 or IFN-I led to their expression of GrB, consistent with the idea that a combination of these cytokines is required for IFN-γ and GrB expression in vivo. Secretion of these mediators involves non-overlapping “danger” pathways; production of bioactive IL-18 from its pro-IL-18 precursor requires inflammasome-caspase-1 proteolytic cleavage (Kuida et al., 1995), while the trans-presentation of IL-15 (Burkett et al., 2004) upon IFN-I stimulation requires IRF-3 and IRF-7 transcription factors (Beuneu et al., 2011; Lucas et al., 2007). To investigate which danger pathways control memory CD8⁺ T cell activation in vivo, we adoptively transferred memory CD8⁺ T cells into WT chimera mice reconstituted with either Casp1^−/− or Irf3^−/− bone-marrow cells or into Il15^−/− mice and monitored their activation after WT Lm infection (Figure 3B and S3). A significantly lower proportion of OT-I memory cells expressed IFN-γ in Casp1^−/−, Irf3^−/− and Il15^−/− recipient mice compared to WT animals, demonstrating that the inflammasome pathway and IL-15 contributed to IFN-γ production by memory CD8⁺ T cells in vivo. As importantly, upregulation of GrB expression was abrogated in the Irf3^−/− chimera and Il15^−/− recipient mice compared to WT mice. The proportion of GrB⁺ memory cells in both Irf3^−/− and Il15^−/− mice was comparable to unchallenged controls, suggesting that IL-15 regulates the expression of early cytolytic effector functions in memory cells, possibly in an IRF3-dependent manner. Altogether, this suggested that memory CD8⁺ T cells express IFN-γ and cytolytic effector functions in vivo through inflammasome- and IRF-3-pathways independently of cognate antigen recognition.

Open in a separate window

Figure 3

IL-18, IL-15 and IFN-I mediate the reactivation of memory CD8⁺ T cells in vivo

CD8⁺ T cells were purified from WT B6 mice grafted with OT-I GFP⁺ cells and immunized with WT Lm-Ova 6 wks prior. (A) CD8⁺ T cells were cultured with indicated recombinant cytokines. The proportion (%) of OT-I GFP⁺ memory cells expressing IFN-γ/GrB and the fold increase in MFI (activated/resting) are shown. Each dot represents OT-I memory CD8⁺ T cells from a different donor mouse and data pool 3 independent replicate experiments (n=4-7). (B) Il15^−/− or irradiated WT bone-marrow chimera of Casp1^−/− or Irf3^−/− mice were challenged or not (grey) with WT Lm and transferred with purified CD8⁺ T cells either into WT (closed) or knockout (open) mice. At 8 or 24 hrs, OT-I memory cells from spleens were analyzed for expression of IFN-γ and GrB. Data are pool of 3 independent replicate experiments (n=4-11) with p-values. Error bars on graphs represent Mean+/−SEM.

Inflammatory monocytes provide IL-18 and IL-15 activating cytokines to memory CD8⁺ T cells

We next sought to define the cells providing these cytokines in vivo. Sentinel cells of the immune system such as DCs, macrophages, neutrophils and Ly6C⁺ monocytes, a subset of monocytes involved in antimicrobial host defenses (Shi and Pamer, 2011), were potential candidates. By flow-cell sorting, we purified conventional DCs (cDCs), neutrophils and Ly6C⁺ monocytes from mice infected with WT Lm and incubated them with resting memory CD8⁺ T cells in vitro (Figures 4A and S4A). A significant proportion of memory cells secreted IFN-γ (30–40%) in presence of inflammatory monocytes while weaker (<10%) activation was measured with activated cDCs or neutrophils. A similar trend was also observed for expression of GrB. We further analyzed whether these cells could provide IL-15 and IL-18 activating cytokines to the memory cells. Since direct measure of IL-18 or of active caspase-1 inside ex-vivo isolated innate cells was below level of detection, we performed the same experiment as in Figure 4A, except that memory CD8⁺ T cells were either coated or not with an anti-IL-18 receptor (IL-18R) blocking serum prior to incubation with the distinct subsets of in vivo activated myeloid cells (Figure 4B and not shown). Whereas expression of GrB on OT-I memory cells remained unchanged in all groups, secretion of IFN-γ, which was only measurable in presence of activated Ly6C⁺ monocytes (Figure 4A), was inhibited by ~80%. Thus Ly6C⁺ monocytes most likely promoted IFN-γ-secretion from the memory cells through inflammasome-mediated release of IL-18. Next, we monitored the cell-surface expression of IL-15Rα by various myeloid splenic cell subsets that could trans-present bioactive IL-15 (Sandau et al., 2004). Ly6C⁺ monocytes (CD3⁻B220⁻NK1.1⁻CD11b^hi, Figure S4A) upregulated this receptor at far superior amounts than cDCs, macrophages or neutrophils (Figure 4C and S4B). Strong upregulation of IL-15Rα was true not only in mice infected with Lm, but also with MCMV, S. pneumoniae, P. berghei and upon poly(I:C) injection (Figure 4D). Therefore, in all organs targeted by the different microbes and in which Ly6C⁺ monocytes are rapidly recruited (Shi and Pamer, 2011), these cells exhibited a functional advantage in expressing high cell-surface bioactive IL-15Rα-IL-15 complexes over other myeloid cell types.

Open in a separate window

Figure 4

Inflammatory Ly6C⁺ monocytes provide bioactive IL-18 and IL-15 to memory CD8⁺ T cells

(A–B) CD8⁺ T cells, cDCs, neutrophils or Ly6C⁺ monocytes were purified by flow-cell sorting respectively from WT B6 mice that (i) contained OT-I GFP⁺ memory cells or (ii) were naive or immunized with WT Lm 20 hrs earlier. Purified CD8⁺ T cells were incubated with each innate cell subset before staining for cell surface expression of CD8 and intracellular IFN-γ and GrB. The proportion (%) of OT-I GFP⁺ memory cells expressing either marker is shown. (B) CD8⁺ T cells were incubated with anti-IL-18R prior to adding activated Ly6C⁺ monocytes. (A–B) represent 3 independent replicate experiments (n=3–6) with p-values. (C) Spleen cells from WT B6 mice immunized with WT Lm (with 10⁵ or 10⁶ Lm) 24 hrs prior were stained with mAbs against cell surface markers discriminating conventional dendritic cells (cDC, CD11c^hi), neutrophils (CD11b^hi, Ly6C^dim, ly6G⁺) and Ly6C⁺ monocytes (CD11b^hi, Ly6C^hi, T/B/NK cell exclusion), and for expression of IL-15Rα. Black gates shown on dot plots define the different cell populations and histograms overlay IL-15Rα expression levels on the distinct cell subsets in uninfected (grey) versus infected (dark) mice. (D) Relative increase of IL-15Rα expression on resting versus activated DC (red), Ly6C⁺ monocytes (black) and neutrophils (blue) from spleen, lung and liver of B6 WT mice immunized with WT Lm, MCMV, S. pneumoniae, P. berghei or injected with poly(I:C) at indicated times. Data are representative of 2 independent replicate experiments (n=6) with p-values.

Inflammatory monocytes efficiently respond to IFN-I and trans-present bioactive IL-15

We further characterized the origin and fate of activated Ly6C⁺ monocytes. For this, we took advantage of a genetically engineered mouse model that express the CFP fluorescent reporter protein fused to the diphtheria toxin receptor (DTR) under the control of the CCR2 promoter (CCR2-DTR-CFP⁺ (Hohl et al., 2009)). In these mice, Ly6C⁺ monocytes, which also represent the major CCR2⁺ cell subset in vivo (Geissmann et al., 2003; Serbina et al., 2003), could not only be traced (CFP⁺) but also selectively eliminated (DTR⁺) upon diphtheria toxin (DT) injection (Figure S4C). In line with current knowledge (Geissmann et al., 2010), CFP⁺CD11b^hi cells (CCR2⁺Ly6C^hi monocytes) in non-infected animals are clearly distinct from cDCs (CD11c^hiCD11b^lo/+), macrophages that could be extracted from the spleen (defined as F4/80⁺CD11b^int), neutrophils (CD11b^hiLy6C^dim) and plasmacytoid DCs (pDCs) (SiglecH⁺CD11b^int) (Figure 5A). Although CFP⁺ cells remain phenotypically distinct from all of the previous cell populations in mice challenged with Lm or MCMV, they upregulated cell-surface markers that define other myeloid cell types such as CD11c (DC), F4/80 (macrophages) and pDCs (PDCA-1), suggesting that they may differentiate into functional DCs and/or macrophages during microbial-driven activation. In support of this hypothesis, activated CCR2⁺Ly6C⁺ monocytes upregulated cell-surface expression of MHC-II molecules, CD40 and CD86 costimulatory molecules, the C-type lectin CD209b, Sca-1 (that marks IFN-I signaling (Essers et al., 2009)) as well as IL-15Rα in the course of the various infections (Figures 5A, B and S4C, D).

Open in a separate window

Figure 5

IL-15Rα upregulation by activated inflammatory monocytes requires IRF3-secreted IFN-I and IFN-I receptor signaling

(A) Phenotypic analysis of Ly6C⁺ monocytes (black dots), cDCs (red) and neutrophils (blue) overlaid amongst whole splenocytes (grey) in naïve and WT Lm-infected mice. (B) Histograms show expression of indicated cell surface markers on Ly6C⁺ splenic monocytes from uninfected (grey) versus WT Lm-infected (black) mice. Data pool 2 replicate experiments (n=6). (C) Bar graphs show cell surface expression of IL-15Rα on Ly6C⁺ monocytes from the spleen of WT (black), Irf3^−/− (red), and Ifnar1^−/− (hatched) bone-marrow chimera infected with WT Lm for 24 hrs. Histogram shows IL-15Rα expression on Ly6C⁺ monocytes from a WT mixed bone-marrow chimera of WT CD45.1⁺ (black) and Ifnar1^−/− (red) mice. Data represent 3 independent replicate experiments (n=4) with p-values. (D) Histograms are gated on WT CD45.1⁺ or Irf3^−/− Ly6C⁺ monocytes and show IL-15Rα expression on Ly6C⁺ monocytes from irradiated WT mixed bone-marrow chimera of WT CD45.1⁺ (black) and Irf3^−/− (red) bone-marrow cells. Data represent 1 experiment (n=4).

Results in Figure 3B suggested that IFN-I signaling acts by enhancing IL-15 trans-presentation. In line with this hypothesis, Ly6C⁺ monocytes responded to recombinant IFN-I in vivo and induced downstream phosphorylation of STAT1, greater than in cDCs or neutrophils (factor of 1.5–3 (MFI), Figure S5). We next investigated whether expression of IL-15Rα on activated Ly6C⁺ monocytes required IFN-I signaling within the monocytes. For this, we generated mixed chimera mice reconstituted with bone-marrow cells from IFN-I receptor- deficient (Ifnar1^−/−) and WT CD45.1⁺ mice and monitored IL-15Rα expression after WT Lm infection (Figure 5C). Upregulation was severely impaired (factor of 3) in the WT-Ifnar1^−/− chimera, suggesting that signaling through the IFN-I receptor is a major pathway of IL-15 trans-presentation by inflammatory monocytes in vivo. However, Irf3^−/− Ly6C⁺ monocytes could differentiate into IL-15Rα⁺ monocytes in mixed chimera of Irf3^−/− and WT CD45.1⁺ bone-marrow cells (Figure 5D), demonstrating that IFN-I may be secreted by other cell types, which still induced Ly6C⁺ monocytes to express IL-15Rα. Thus, IFN-I directly regulates the ability of inflammatory monocytes to trans-present bioactive IL-15 in vivo, possibly in a paracrine manner.

Inflammatory monocytes drive memory CD8⁺ T and NK cells differentiation into effectors during Lm infection

To prove that inflammatory monocytes regulated pathogen-induced memory CD8⁺ T cell activation in vivo, we treated the CCR2-DTR mice with DT, which eliminated Ly6C⁺ monocytes (Figure S4C), prior to OT-I and endogenous memory CD8⁺ T cell transfer, and before infection with WT Lm (Figure 6A). DT-treated mice had significantly less activated memory CD8⁺ T cells, e.g., expressing IFN-γ (52%) and GrB (34%) compared to untreated control mice (74 and 60%), both in frequencies and numbers at 8 hrs post-challenge infection, further supporting a role for inflammatory CCR2⁺ monocytes in optimal activation of memory CD8⁺ T cells in vivo. Likewise, in Ccr2^−/− mice in which Ly6C⁺ monocytes are mostly retained in the bone-marrow (Serbina and Pamer, 2006), we measured half as many IFN-γ⁺ memory CD8⁺ T cells than in WT control mice (not shown), consistent with our previous observations. In support of this hypothesis, we also found that the transient clusters of OT-I memory cells -some of which were IFN-γ⁺- that we previously described in the red pulp area of infected spleens by 6–12 hrs post challenge infection (Bajenoff et al., 2010) also contained massive numbers of interacting CCR2⁺ inflammatory monocytes (Figure 6B). This was not observed in unchallenged mice containing resting OT-I memory cells.

Open in a separate window

Figure 6

Inflammatory monocytes are essential for antigen-independent reactivation of memory CD8⁺ T and innate NK lymphocytes

WT CCR2-DTR^−/− or CCR2-DTR^+/− littermates were injected with DT prior to infections and transfers with OT-I and endogenous memory CD8⁺ T cells purified from WT B6 mice grafted with OT-I GFP⁺ cells and immunized with WT Lm-Ova 6 wks earlier. At indicated times after WT Lm (10⁵ or 10⁶) challenge, the activation (IFN-γ, GrB) of OT-I (A) and endogenous (C) memory CD8⁺ T cells, and NK cells (D) was monitored in the spleen of mice after staining for the appropriate cell-surface markers (CD8, CD3, NKp46). Grey dots depict resting memory CD8⁺ T cells, each dot representing an individual mouse in 2 independent replicate experiments (n=6-11) with p-values. (B) CCR2-DTR mice were transferred with OT-I-tomato⁺ cells, immunized with WT Lm-Ova and challenged or not (resting) with WT Lm 6 wks later. After 8 hrs, spleens were fixed, sectioned and stained to reveal IFN-γ, CFP (CCR2⁺ monocytes) and tomato (OT-I memory), and analyzed by confocal microscopy. Images are representative of 4 mice in 2 independent replicate experiments. Error bars on graphs represent Mean+/−SEM.

By 24 hrs, the proportion and numbers (not shown) of IFN-γ⁺ OT-I (6.9%) and endogenous (non-OT-I) memory cells in CCR2-DTR-depleted mice was equivalent to that of unchallenged mice (0.2%), and significantly reduced in comparison to non-DT treated mice (60.8%) (Figure 6A, C). At this time, endogenous GrB⁺ memory CD8⁺ T cells (12.4%) were also significantly diminished and at equivalent proportions (5.8%) to that of unchallenged mice (4%) (Figure 6C). However, the frequencies of OT-I memory cells expressing GrB was comparable in both experimental groups, suggesting that other cell types such as cDCs or macrophages (Lucas et al., 2007; Mortier et al., 2009) could compensate for the loss of Ly6C⁺ monocytes and trans-present sufficient levels of bioactive IL-15 to OT-I memory cells (Figure 6A).

Since activation of NK cells depends on IFN-I, IL-15 and IL-18 cytokines (Beuneu et al., 2011; Chaix et al., 2008; Lucas et al., 2007), we reasoned that inflammatory monocytes also regulated NK cell activation. As hypothesized, the proportion of IFN-γ⁺ and GrB⁺ NK cells decreased substantially in DT-treated CCR2-DTR mice challenged with WT Lm respectively at 8 and 24 hrs, and was close to that of unchallenged mice (Figure 6D). Thus, collectively Ly6C⁺ monocytes not only control memory CD8⁺ T cell, but also NK cell differentiation into robust effector cells.

Sustained cytolytic marker expression by memory CD8⁺ T and NK cells requires Ly6C⁺ monocytes and macrophages

Since eliminating Ly6C⁺ monocytes only had a partial and transient effect on the expression of GrB by both categories of lymphocytes (Figure 6A, D), we hypothesized that other myeloid cell types must be involved. Several reports using the CD11c-DTR transgenic mouse model, in which the DTR is expressed under the CD11c promoter, allowing for selective depletion of CD11c⁺ cells (Jung et al., 2002), suggested that cDCs orchestrate the activation of NK and memory CD8⁺ T cells in vivo (Lucas et al., 2007; Mortier et al., 2009; Zammit et al., 2005). To investigate this possibility, we transferred OT-I memory cells 12 hrs post-DT treatment into CD11c-DTR and CCR2-DTR along with WT littermate animals as controls, and monitored the expression of GrB by OT-I memory and NK cells 24 hrs after the challenge infection (Figure S6). GrB was still upregulated on the distinct lymphocytes from DT-treated DTR-expressing animals and was equivalent to that of non-depleted animals. Analysis of the populations eliminated upon DT injection in WT Lm infected and uninfected DTR-expressing mice revealed that in CD11c-DTR mice, cDCs (CD11c^hi CD8α⁺ and CD8α^- subsets) and splenic macrophages (F4/80⁺CD11b⁺) were efficiently eliminated (80 and 96% respectively) whereas in CCR2-DTR mice, depletion mostly targeted Ly6C⁺ monocytes (>95%) (Figure S7 and not shown). This suggested that inflammatory monocytes or cDCs and subsets of splenic macrophages compensated for each other’s respective loss and provided a sufficient amount of trans-presented IL-15 to lymphocytes.

To assess this hypothesis further, we used clodronate liposomes, a well-established non-genetic method that eliminates a wide array of phagocytic cells (Chow et al., 2011). Clodronate liposome injection depleted Ly6C⁺ monocytes (by~90%), splenic macrophages (by~80%), as well as cDCs (by~50%) and neutrophils (~20%) (Figure S7). However, myeloid cells exhibiting high turnover, such as monocytes or neutrophils, underwent rapid replacement and fully restored their homeostasis by 5–7 days ((van Rooijen et al., 1989) and Figure S7). We injected WT and CCR2-DTR B6 mice with clodronate or PBS control liposomes 1 or 6 days prior to transfer of purified OT-I memory cells, DT treatment and infection with WT Lm (Figures 7A and S7). Twenty-four hrs post-challenge infection, upregulation of GrB expression and secretion of IFN-γ by OT-I memory CD8⁺ T cells was fully abrogated in day 1-clodronate liposome-depleted mice compared to untreated control mice (GrB:28 versus 63%; IFN-γ: 4 versus 55%) both in frequencies and numbers (Figure 7A and not shown). Likewise, upregulation of GrB and secretion of IFN-γ by NK cells was prevented (Figure 7B). Thus, as hypothesized, eliminating these subsets of phagocytic cells efficiently abrogated the upregulation of GrB by memory CD8⁺ T and NK cells, suggesting that they are required for sustained expression of bioactive IL-15 (Lucas et al., 2007; Mortier et al., 2009). In WT and CCR2-DTR mice treated with clodronate liposomes 6 days prior, in which Ly6C⁺ monocytes and neutrophils levels were largely restored but splenic macrophages were still absent and the proportion of cDC stayed comparable to day 1-treated mice (Figure S7), memory CD8 and NK lymphocytes still differentiated into effector cells (Figure 7A, B). When clodronate liposome-treated CCR2-DTR mice received DT to further deplete Ly6C⁺ monocytes, lymphocyte activation was fully abrogated. This suggested that inflammatory monocytes and splenic macrophages indeed contribute to sustain memory CD8⁺ T and NK lymphocytes differentiation into cytolytic effector cells. Thus altogether, inflammatory monocytes, and to a lesser extent splenic macrophages, most critically orchestrate the activation of both categories of lymphocytes in vivo.

Open in a separate window

Figure 7

IFN-γ secreting memory CD8⁺ T and NK lymphocytes differentiate upon signals from Ly6C⁺ monocytes while GrB depends upon both Ly6C⁺ monocytes and splenic macrophages

B6 or B6-K^d+/− WT mice were respectively transferred with 500 naïve OT-I-tomato⁺ and WP20-GFP⁺ T cells and immunized with WT Lm-Ova and Lm. Six wks later, purified OT-I or WP20 memory cells were adoptively transferred to CCR2-DTR mice. (A–C) Mice were further injected or not with clodronate liposomes and treated or not with DT prior to WT Lm or MCMV infection. At indicated times after the challenge infection, expression of IFN-γ and/or GrB by splenic OT-I and NK cells was monitored in the different experimental groups. Dot plot are representative FACS profiles after gating on OT-I or NK cells and numbers correspond to the % of IFN-γ⁺ or GrB⁺ cells in the gate shown. Scattered plots represent the pool of 2–3 independent replicate experiments (n=3–9) with p-values. In (D), WP20 memory cells were transferred to CCR2-DTR mice (+/−DT) expressing or not expressing the MHC-I molecule K^d prior to WT Lm infection. Grey dots depict resting memory CD8⁺ T cells. In (E), OT-I memory cells were transferred into CCR2-DTR mice (+/−DT) further challenged with WT ΔActA Lm-Ova and their frequency in the spleen of mice was measured 3 days later. Results represent 2–3 independent replicate experiments (n=2–5) with p-values.

Microbial pathogens, mice infections and measure of protective immunity

Listeria monocytogenes (Lm): Mice were inoculated with Lm 10403s strain, either WT Lm, Lm-Ova, LLO_Ser92 Lm, for primary immunization with 0.1×LD₅₀ Lm (3×10³ or 10⁴ WT Lm and Lm-OVA respectively) and for secondary infections 4–6 wks later with 10⁶ WT Lm or Lm-Ova unless otherwise specified. Streptococcus pneumoniae: 5×10⁵ CFU of the ST3 strain A66.1 (A66) grown in tryptic soy broth to midlog phase was inoculated i.n. to mice in 25 μl/nare. Murine cytomegalovirus: 2×10⁴ PFU/mouse of the Smith strain derived from a salivary gland stock was injected i.v to mice. Plasmodium Berghei (Pb): 5×10⁵ Pb ANKA infected red blood cells from seeder mice were injected i.v. to mice. Lymphocytic Choriomeningitis Virus: 2×10⁵ PFU/mouse of the Armstrong strain were injected i.p. to mice.

Generation of memory CD8⁺ T cells in vivo

Lm: WT B6 mice were adoptively transferred with ~500 OT-I or 1,000 WP20 cells isolated from spleen and lymph nodes of OT-I RAG-1^−/− or WP20 K^d+ RAG-1^−/− expressing GFP and/or Tomato (Td). One day later, mice were immunized i.v. with 10⁴ WT Lm-Ova or Lm, and were used 4–6 wks later when they had developed memory cells. LCMV: WT B6 mice were immunized with 2×10⁵ PFU of virus i.v. and used at least 8 wks after the primary immunization.

In vivo memory T cell and NK cell activation assays

In immunized mice: Mice containing Lm-specific memory CD8⁺ T cells, e. g. immunized 30 days prior, were challenged with WT Lm or Lm-OVA, 50μg poly(I:C), S. Pneumoniae, MCMV, or Pb, and organs analyzed by FACS at the indicated time points in the text/figures/legends. In adoptively transferred mice: CD8⁺ T cells were purified from the spleens of primary immunized mice and 3–5×10⁶ cells transferred into mice that were either infected with WT Lm 30 min before or left uninfected. In MCMV experiments, mice were infected for 24 hrs before transfer. After 8 or 24 hrs, spleens were recovered and OT-I, WP20, or endogenous memory cells activation was monitored. For antigen-specific proliferation, 6×10⁵ purified CD8⁺ T cells were transferred into recipients and proliferation analyzed 72 hrs after ΔActA Lm-Ova infection.

H2-K^d/LLO_91–99 tetramers came from the NIH Tetramer Facility. We are grateful to W. Jacobs Jr. for privileged access to the HHMI Aria II, and thank L. Tesfa (FACS Core). We thank F. Sutterwala (University of Iowa) and R. Flavell (Yale University) for providing Casp1^−/− bone-marrow cells and mice; D. Woodland and J. Kohlmeier (Trudeau Institute) for providing Ifnar1^−/− bone marrow cells; T. Moran (Mount Sinai School of Medecine) for Irf3^−/− mice; and E. Pamer (Memorial Sloan-Kettering Cancer Center) for CCR2-DTR and WP20 mice. We thank C. Biron (Brown University) for providing stocks of MCMV. We thank M. Bajenoff (CIML, France), S. Porcelli (AECOM, NY), L. Chorro, C. Chandrabos, E. Spaulding for helpful discussions. Work was supported by Institutional Funds of the Albert Einstein College of Medicine of Yeshiva University (GL), the National Institute of Health (R21AI095835, GL), the Agence Nationale pour la Recherche (ANR-10-LABX-61, JM), InCa Atip/Avenir (JM), and the fundation Bettencourt-Schueller (JM). Core flow cytometry were supported by the Albert Einstein Cancer Center (NCI Grant 2P30CA013330).