Activated NKT Cells Can Condition Different Splenic Dendritic Cell Subsets To Respond More Effectively to TLR Engagement and Enhance Cross-Priming Free

Phenotypically distinct subsets of dendritic cells (DCs) play specialized roles in promoting different aspects of adaptive immunity. For example, studies in mice showed that splenic DCs expressing CD8α-chain homodimers (CD8α⁺ DCs) play a major role in stimulating CD8⁺ T cells (1–3), whereas CD8α⁻ DCs play a greater role in the control of CD4⁺ T cell responses (4). It remains unclear how restricted these roles are, particularly because the nature and timing of activatory stimuli that can alter DC function are likely to change with exposure to different pathogens or even change during the course of a single infection.

TLRs are a family of innate receptors that sense microbial products and trigger DC maturation and cytokine production, thereby providing important cues required to induce adaptive responses (5). Each TLR is triggered by a different set of microbial compounds and, in turn, is coupled to a distinct signal transduction pathway, so that the capacity of DCs to mediate adaptive responses is effectively fine-tuned by the type of microbial insult encountered. Although DCs express the broadest range of TLRs, there are differences in expression among DC subsets, including between CD8α⁺ DCs and CD8α⁻ DCs (6), suggesting that an additional level of fine-tuning is related to the type of DC that can be stimulated by a given microbial product. Adding to this complexity is the high likelihood that microbes express many molecular structures that can act as agonists for different TLRs. All of these factors ultimately help to determine the nature of the adaptive response induced.

The size and quality of an adaptive immune response can also be influenced by signals provided to DCs by other leukocytes in the local lymphoid environment. Cells with innate effector responses, such as NKT cells, can be particularly influential in this regard. The spleen contains a large population of NKT cells distributed throughout the parenchyma in the steady-state (7). Type I NKT cells expressing an invariant TCR Vα14 chain coupled with a limited TCR Vβ-chain repertoire can recognize bacterial glycolipids (8–10) or upregulated expression of endogenous glycolipids (11) in the context of the MHC class I–like molecule CD1d on APCs. Once stimulated, NKT cells consolidate in the marginal zone (MZ) and from this location can provoke a powerful cascade of molecular and cellular interactions that can influence effector outcomes throughout the spleen (7, 12). Among the important cues provided by NKT cells are CD40-mediated signals that license APCs to engage in cross-priming of Ag-specific CD8⁺ T cells (13).

It is known that NKT cell–mediated signals can be integrated with TLR-mediated signals to further enhance the capacity to cross-prime CD8⁺ T cells (14). However, it remains unclear whether integration occurs at the level of a specific DC subset or whether a range of APCs can be invoked to participate in cross-priming under the stimulatory conditions created by NKT cell activation. To address this, we investigated CD8⁺ T cell responses to circulating Ag in the presence of the synthetic NKT cell ligand α-galactosylceramide (α-GalCer), together with agonists for TLRs that are differentially expressed among the different splenic DC subsets. Using this strategy, we show that potent CD8⁺ T cell responses are dependent on early activity of a subpopulation of CD8α⁺ DCs expressing langerin (CD207). However, with appropriate temporal phasing of activation stimuli, CD8α⁻ DCs that have been exposed to activated NKT cells can become conditioned to respond more effectively to subsequent TLR ligation. Significantly, under these conditions, CD8α⁻ DCs can contribute to improve the size and functionality of the induced CD8⁺ T cell response. These results highlight the contribution of temporal phasing of stimuli in influencing adaptive immune responses, information that may be useful in designing more effective vaccines.

C57BL/6J mice were obtained from The Jackson Laboratory (Bar Harbor, ME) and from the Animal Resource Centre (Canning Vale, WA, Australia). OT-I mice, which express a transgenic TCR specific for the H-2K^b–binding peptide of OVA (OVA_257–264) (15), were crossed in-house with the CD45.1-congenic strain B6.SJL-Ptprc^aPep3^b/BoyJArc (The Jackson Laboratory) to allow for identification of adoptively transferred OT-I cells in CD45.2 recipients. Also used were lang-EGFP and lang-DTREGFP knock-in mice, which express the human diphtheria toxin (DT) receptor and/or enhanced GFP (EGFP) under the control of the langerin promoter (16), and siglecH-DTR mice (17) which express the human DT receptor under the control of the siglecH promoter. The latter were crossed in-house with C57BL/6J mice. All mice were bred and housed by the Biomedical Research Unit at the Malaghan Institute of Medical Research. Animal protocols were approved by the Victoria University Ethics Committee (AEC 2009R13M; AEC 2012R29M).

Unless otherwise stated, Lang-DTREGFP mice were depleted of langerin⁺ DCs with i.p. treatment of 350 ng DT (Sigma-Aldrich, Auckland, New Zealand) 1 d prior to Ag and stimuli administration, resulting in high depletion efficiency, as previously described (18), SiglecH-DTR mice were depleted of plasmacytoid DCs (pDCs) with DT at days 1 and 2 prior to Ag administration (Supplemental Fig. 1). Splenectomy was performed as previously described (19). A sham control was used, wherein an incision was made in the left flank, and the wound was closed without removing the spleen.

Single-cell suspensions were prepared from spleens and lymph nodes of OT-I × B6.SJL-Ptprc^aPep3^b/BoyJArc mice, and CD8⁺ cells were enriched using FlowComp Dynabeads (Invitrogen, Auckland, New Zealand). For assessing immune responses in vivo, 1 × 10⁴ enriched CD8⁺ cells were transferred into CD45.2⁺ recipient mice 1 d prior to i.v. administration of 200 μg OVA (Hyglos, Bernried, Germany), with or without i.v. administration of activatory stimuli at the times indicated in the text. Stimuli used were 20 μg N-palmitoyl-(R)-Cys-(S)-Ser(S)-Lys4-OH, trihydrochloride (Pam3Cys; EMC Microcollections, Tübingen, Germany), 6.67 μg resiquimod (PharmaTech Labs, Shanghai, China), 33.3 μg polyinosinic-polycytidylic acid (Poly I:C; InvivoGen, San Diego, CA), and 200 ng the invariant CD1d-binding NKT cell ligand α-GalCer, which was manufactured by a modified Gervay-Hague glycosylation procedure (20). Briefly, the glycosyl iodide donor was prepared from per-TMS–protected galactose by reaction with iodotrimethylsilane. The acceptor was synthesized from commercially available phytosphingosine by azido transfer; selective protection of the primary hydroxyl with tri-isopropylsilyl chloride, followed by benzylation of the secondary hydroxyls afforded the fully protected intermediate. Removal of the primary hydroxyl silyl–protecting group was carried out with acid to afford the acceptor that was subsequently glycosylated under the reported conditions (20). Routine deprotection and installation of the cerotic acid afforded α-GalCer. Spectroscopic data were identical to those published previously (21). Solubilization of α-GalCer was achieved by freeze-drying in the presence of sucrose, l-histidine, and Tween 20, as described previously for the solubilization of α-GalCer (22). The solubilized α-GalCer was diluted in water for i.v. administration. For assessment of immune responses, blood was collected from the lateral tail vein, and RBCs were lysed with RBC lysis solution (QIAGEN, Valencia, CA) before Ab staining and flow cytometry.

Animals that had been administered OVA, with or without activatory stimuli, received 1 × 10⁶ CFSE-labeled lymph node cells from OT-I × B6.SJL-Ptprc^aPep3^b/BoyJArc mice to assess induction of proliferation. Labeling was conducted at room temperature using 1 μM CFSE (Invitrogen) for 7 min, followed by quenching with 1:1 volume FCS (Sigma-Aldrich) and washing with PBS (Invitrogen). Blood was collected at the times indicated for flow cytometry.

All Ab-staining steps were performed at 4°C. Nonspecific FcR-mediated Ab staining was blocked by incubation for 10 min with anti-CD16/32 Ab (2.4G2; prepared in-house from hybridoma supernatant). Flow cytometry was performed on a BD Biosciences FACSCalibur or LSR II SORP, and data were analyzed using FlowJo software (TreeStar, Ashland, OR). The following Abs were used: anti-CD8 (53-6.7), anti-Vα2 (B20.1), anti–IL-12p40 (C15.6), anti-CD40 (3/23), anti-B220 (RA3-6B2) (all from BD Biosciences, San Jose, CA) and anti-CD45.1 (A20) and anti-CD11c (N418) (both from BioLegend, San Diego, CA). The viability dyes used were DAPI (Invitrogen) and propidium iodide (BD Biosciences).

Mice were treated with various stimuli i.v. and 2 h later were administered 250 μg Brefeldin A (Sigma-Aldrich) i.v. to inhibit cytokine release by cells. Spleens were extracted 4 h later, incubated for 30 min at 37°C with Liberase TL and DNase I (both from Roche, Auckland, New Zealand) and 2 μM Monensin (Sigma-Aldrich), and processed through a 70-μm filter. RBCs were lysed, and CD11c⁺ cells were enriched by AutoMACS using CD11c magnetic beads (both from Miltenyi Biotec Australia, Sydney, NSW, Australia). Cells were stained with surface Abs and LIVE/DEAD Fixable Blue viability dye and fixed and permeabilized with the FIX & PERM Cell Permeabilization kit (both from Invitrogen). Cells were then incubated with anti–IL-12p40 and analyzed by flow cytometry.

Mice were left naive or treated i.v. with indicated stimuli. Serum was collected 6 h later and analyzed for levels of IFN-α using a FlowCytomix kit, following the manufacturer’s instructions, and assessed using FlowCytomix Pro 3.0 software (both from eBioscience, San Diego, CA).

The T cell lymphoma cell line E.G7-OVA, which expresses OVA protein (23), was cultured in IMDM supplemented with 5% FBS, 100 U/ml penicillin, 100 μg/ml streptomycin, 50 μM 2-ME, and 500 μg/ml G418 (all from Invitrogen), strained through a 70-μm filter, and resuspended in PBS for s.c. injection. Groups of naive C57BL/6 mice (n = 5) were challenged with 5 × 10⁵ cells in the flank. On day 5, when tumors were palpable, the mice were subjected to one round of i.v. therapy with 200 μg OVA plus stimuli indicated in text or with PBS. Tumor growth was monitored every 1–2 d, with tumor size calculated as the product of the two bisecting diameters. Measurements were stopped for each group when the first mouse developed a tumor > 200 mm².

All data were graphed and statistical analyses were performed using GraphPad Prism 5 software (GraphPad, La Jolla, CA). One-way ANOVA with the Tukey posttest was used as indicated.

The MZ of the spleen is an important location for initiation of adaptive immunity because it contains a vascular bed, the marginal sinus, where the arterial blood supply can be screened for particulate or cell-based Ags (24). A subset of CD8α⁺ DCs in the MZ that expresses langerin (CD207) can acquire Ags from circulating apoptotic cells for the purpose of inducing cross-tolerance of Ag-specific CD8⁺ T cells in the steady-state (25, 26). Significantly, when Ag enters the MZ in close temporal association with activatory stimuli, these same DCs can cross-prime Ag-specific CD8⁺ T cell responses with cytotoxic function (27). Thus, i.v. injection of soluble protein Ag together with the NKT cell ligand α-GalCer results in effective cross-priming of Ag-specific CD8⁺ T cell responses (28, 29), which can be enhanced using a combination of α-GalCer with selected TLR ligands (14). To explore the relationship between these DCs and other splenic APCs in inducing CD8⁺ T cell responses, we first established and assessed a model to evaluate responses to OVA and α-GalCer with TLR ligands, focusing initially on the TLR2 ligand Pam3Cys. To enable effective immune monitoring by flow cytometry, hosts received a small cohort of transgenic OVA_257–264-specific CD8⁺ T cells from OT-I mice before the Ag, with or without activatory stimuli, was injected. Assessment in the blood 7 d later showed a significant response to the injected Ag and stimuli, which was greatest when both α-GalCer and Pam3Cys were provided (Fig. 1A). This response was ablated in splenectomized mice, confirming an important role for splenic APCs in priming the T cell response (Fig. 1B). Because langerin⁺ CD8α⁺ DCs are located primarily in the MZ, they could be expected to be early recipients of the injected circulating Ag. To investigate the timeframe in which these cells play a key role in stimulating CD8⁺ T cell responses, lang-DTREGFP recipients expressing the DT receptor from the langerin promoter were depleted of langerin-expressing cells by administering DT at different times relative to the administration of OVA, α-GalCer, and Pam3Cys (Fig. 1C). Depletion of langerin-expressing cells before Ag administration reduced the stimulation of CD8⁺ T cell responses, with significantly less accumulation of transgenic T cells in the blood after 7 d. When DT was administered 2 h after Ag, the response was similarly reduced. However, when DT was administered after 4 h, the response was not affected. We showed previously that langerin⁺ CD8α⁺ DCs are depleted in the spleen within 4 h of DT administration (18). Taking this into account, we conclude that langerin⁺ CD8α⁺ DCs were required for only 6–8 h after Ag administration.

To determine whether the combination of TLR-mediated signals and NKT cell–derived signals that drive enhanced cross-priming were focused on langerin⁺ CD8α⁺ DCs specifically or might involve other APCs, we exploited the differential expression of TLRs by different subsets of DCs. Although CD8α⁺ DCs express both TLR2 and TLR3, they do not express TLR7. In contrast, CD8α⁻ DCs express TLR2 and TLR7, but not TLR3 (6). When OVA and α-GalCer were administered in combination with ligands for these different receptors (Fig. 1D), OVA_257–264-specific CD8⁺ T cell responses were significantly enhanced when ligands for TLR2 or TLR3 were used (Pam3Cys or Poly I:C, respectively) but not when a ligand for TLR7 was used (resiquimod). Therefore, even when stimulated directly, CD8α⁻ DCs could not contribute significantly to the response. These experiments were conducted in lang-DTREGFP mice, so it was possible to examine the effect of depleting langerin⁺ cells before the Ag was administered. The induction of OVA_257–264-specific CD8⁺ T cell responses was reduced considerably in the absence of langerin⁺ cells when each of the different TLR ligands was used (Fig. 1D). These data suggest that cross-priming is enhanced when integration of TLR-mediated signals and NKT cell–derived signals involves langerin⁺ CD8α⁺ DCs.

It remained possible that APCs other than langerin⁺ CD8α⁺ DCs could acquire and cross-present injected Ag without providing the stimulatory capacity to cross-prime CD8⁺ T cells; these APCs may become important contributors to cross-priming if appropriately timed stimuli are provided. Therefore, we investigated the role of langerin⁺ and langerin⁻ APCs in a model system that differentiates between the ability to cross-present Ag and the ability to cross-prime a sustained CD8⁺ T cell response. Assessing proliferation of CFSE-labeled OVA_257–264-specific CD8⁺ T cells after transfer into Ag-bearing mice is known to be a very sensitive measure of Ag presentation in vivo (30). Early T cell proliferation (within 3 d) after administration of OVA is dependent on the presence of the H-2K^b–binding epitope in the host, suggesting that the protein has been processed and cross-presented. However, proliferation and accumulation of OVA_257–264-specific CD8⁺ T cells are only sustained over longer periods if the Ag has been cross-presented in a stimulatory fashion (i.e., if the T cells have been cross-primed). Therefore, we used this strategy to assess responses to the administration of OVA and α-GalCer with Pam3Cys, using lang-DTREGFP hosts, so that the impact of depleting langerin⁺ cells could be observed. We showed previously, using similar assays, that langerin⁺ CD8α⁺ cells are critical for cross-priming at the time of Ag administration (27). In this study, we examined whether langerin⁺ CD8α⁺ cells can sustain this priming capacity over a longer period, either as the key presenting cells themselves or through capacity to influence other APCs. Therefore, the CFSE-labeled OVA_257–264-specific CD8⁺ T cells were transferred 24 h after Ag administration into animals, with or without prior depletion of langerin-expressing cells (Fig. 2A, 2B). Proliferation of CFSE-labeled cells was observed 2 d later in all animals, with depletion of langerin⁺ CD8α⁺ cells before Ag administration having only a limited negative impact on this response. Therefore, APCs other than langerin⁺ cells had acquired OVA and were able to cross-present at this time. However, when the T cell response was examined at 5 d after transfer, significant accumulation of OVA_257–264-specific CD8⁺ T cells was only seen when langerin⁺ CD8α⁺ cells had been present at the time of Ag administration (Fig. 2C). Therefore, even when the langerin⁻ APCs come into contact with specific T cells 24 h after Ag administration, langerin⁺ CD8α⁺ cells were still the primary cells capable of cross-priming CD8⁺ T cell responses.

These results were surprising in light of previous observations that the bulk of langerin⁺ CD8α⁺ cells in the spleen die within 24 h of administration of α-GalCer through an activation-induced cell mechanism (31), which also was observed in this study (data not shown). This could either mean that a small population of langerin⁺ CD8α⁺ DCs with cross-priming activity actually survives initial activation in the presence of the injected stimuli or that initial activation of langerin⁺ CD8α⁺ DCs at the time of Ag administration modifies the local environment, enabling other APCs to contribute to cross-priming. To test this, we added another group to the analysis: any remaining langerin⁺ CD8α⁺ DCs were depleted with DT 12 h after Ag administration, and the CFSE-labeled T cells were transferred an additional 12 h later. In this situation, cross-presentation was still observed 2 d after T cell transfer, but there was not the profound accumulation of OVA_257–264-specific CD8⁺ T cells at day 5 seen when langerin⁺ CD8α⁺ cells were present throughout (Fig. 2). Therefore, although our previous experiments showed that the most significant cross-priming activity occurred within the first six h of Ag administration, a population of langerin⁺ CD8α⁺ DCs with enduring cross-priming capacity was still present 24 h after Ag administration. Therefore, although langerin⁻ APCs in the host may be capable of cross-presentation, langerin⁺ CD8α⁺ DCs were the principal inducers of cross-priming, and a small number of these cells retained the ability to prime naive CD8⁺ T cells for ≥24 h after the Ag entered the spleen.

Given that langerin⁺ and langerin⁻ APCs were capable of cross-presenting Ag for ≥ 24 h in the previous experiment, we investigated whether the different APCs could be stimulated to contribute to cross-priming through an appropriate temporal “phasing” of stimuli. In this regard, the cooperative activity of α-GalCer–mediated NKT cell activation and direct TLR ligation was explored by providing each of the ligands at different times in relation to OVA administration. Initial experiments were conducted with Pam3Cys administered at different time points relative to OVA plus α-GalCer, measuring the accumulation of transferred OVA_257–264-specific CD8⁺ T cells in the blood 7 d after OVA was administered. Introducing Pam3Cys before OVA plus α-GalCer had a significant negative impact on cross-priming, with administration as little as 3 h before OVA and α-GalCer effectively negating the adjuvant effect of NKT cell activation (Fig. 3A). This may reflect an inability of the TLR-stimulated APC to acquire and process Ag, which was reported previously for systemic TLR administration (32). As expected, significant stimulation of OVA_257–264-specific CD8⁺ T cells was seen when Pam3Cys was administered simultaneously with OVA and α-GalCer. When Pam3Cys was administered up to 6 h after OVA and α-GalCer, a significant cytokine storm was induced that severely incapacitated the animals, requiring them to be euthanized. Serum cytokine analysis indicated a sharp peak in TNF-α, IL-12p70, and IFN-γ (data not shown). However, animals administered Pam3Cys between 12 and 24 h after OVA and α-GalCer exhibited CD8⁺ T cell responses that were superior to those in animals given OVA plus α-GalCer alone (Fig. 3B). This cooperative activity between the temporally separated stimuli was lost when Pam3Cys was delayed an additional 24 h. These results indicate that APCs that have been exposed to activated NKT cells are particularly sensitive to further stimulation via TLR2 for 24–48 h.

Importantly, because most APCs are CD1d⁺, this conditioning may occur in many cell types. It was possible that conditioning rendered cells other than langerin⁺ CD8α⁺ DCs capable of participating in cross-priming, such as those langerin⁻ cells capable of cross-presentation highlighted earlier. Therefore, the phasing of TLR stimulation after NKT cell stimulation was examined in more detail, with different TLR ligands used to determine whether CD8α⁺ or CD8α⁻ DCs could be induced to participate in enhancement of the CD8⁺ T cell response. By conducting these experiments in lang-DTREGFP mice and assessing the effect of depleting langerin-expressing cells after injection of OVA and α-GalCer, but before TLR administration, it also was possible to determine whether any enhancement of cross-priming could be attributed to the activity of langerin⁺ or langerin⁻ APCs. When langerin⁺ DCs were present at the time of injection of OVA and α-GalCer, delayed administration of ligands for TLR2 or TLR3 enhanced CD8⁺ T cell responses. In fact, delayed administration of Pam3Cys was often able to enhance the CD8⁺ T cell response further than was concomitant administration of the stimuli (Fig. 3C). However, this enhancement over responses to OVA and α-GalCer alone was considerably reduced, although not entirely ablated, when langerin⁺ CD8α⁺ DCs were removed before delayed administration of the Pam3Cys, suggesting that the TLR2 ligand had the most impact on these DCs (Fig. 3C). Similarly, the enhancement seen with delayed administration of the TLR3 ligand Poly I:C also was reduced when langerin⁺ CD8α⁺ DCs were depleted. In both of these situations, we conclude that the delayed TLR stimulation was primarily able to enhance CD8⁺ T cell responses by acting directly on langerin⁺ CD8α⁺ DCs that survived initial stimulation by NKT cells. Interestingly, delayed administration of the TLR7 ligand resiquimod, which should not be able to target langerin⁺ CD8α⁺ DCs directly, also resulted in an increased CD8⁺ T cell response. This response was not affected by depletion of the langerin-expressing cells, further implying involvement of other APCs. Given that we showed previously that contemporaneous administration of OVA and resiquimod was unable to induce significant cross-priming (Fig. 1D), this enhanced response with delayed administration of the TLR7 ligand likely reflected an altered or conditioned phenotype in TLR7-expressing APCs as a consequence of the earlier stimulation of NKT cells.

The significant cytokine storm observed in response to TLR ligation after α-GalCer suggested that activated NKT cells were providing signals that lowered the threshold for TLR-induced cytokine production. To determine which cells may have been influenced in this manner, we assessed the production of IL-12p40 by splenic DCs as an early readout of TLR ligation. For this purpose, we assessed splenocytes from lang-EGFP mice, which express EGFP from the langerin promoter (16), in combination with Abs to CD8α and CD11c. The treated mice were in three groups: those that were administered α-GalCer alone, those that were administered α-GalCer in combination with the TLR7 ligand resiquimod, and those that were administered α-GalCer followed by resiquimod 24 h later. Analysis was conducted 6 h after the last stimulus was injected (Fig. 4). As previously reported, administration of α-GalCer alone induced the most IL-12p40 production by langerin⁺ CD8α⁺ DCs (27), and only limited production was seen from langerin⁻ CD8⁺ DCs. Interestingly, contemporaneous administration of α-GalCer and resiquimod inhibited the production of IL-12p40 by langerin⁺ CD8α⁺ DCs, suggesting an inhibitory effect mediated by local TLR7 stimulation. However, when resiquimod was administered after α-GalCer, all of the DC populations assessed were able to produce significant quantities of IL-12p40 (Fig. 4A, 4B). Thus, the NKT cells that responded to α-GalCer administration were able to provide a conditioned environment that enabled a more effective TLR7-induced cytokine response. Because many of the DCs that were able to produce IL-12p40 in response to resiquimod did not express TLR7, at least some of this conditioning must have been via a released factor from the TLR7-stimulated APCs.

It was possible that NKT cell–mediated stimulation of langerin⁺ CD8α⁺ DCs provided a signal that, in turn, conditioned other APCs to produce more IL-12p40 in response to resiquimod. To test this, lang-DTREGFP mice were administered α-GalCer and delayed resiquimod, with langerin-expressing cells depleted before administration of both stimuli or between the two stimuli (Fig. 4C). Regardless of the timing of depletion, when resiquimod was delivered after α-GalCer, there was a trend toward increased IL-12p40 release, either by the few CD8α⁺ DCs that remained after α-GalCer treatment or by CD8α⁻ DCs (that were weak producers when only α-GalCer was provided). Therefore, the NKT cell–mediated conditioning of langerin⁻ APCs observed was independent of langerin⁺ CD8α⁺ DCs and may have occurred by direct interaction with NKT cells.

It was shown that NKT cells can interact via OX40/OX40L with pDCs, which then contribute to the activation of conventional DCs for T cell priming (33, 34). Because pDCs are a major producer of type I IFN (35), we assessed whether this cytokine contributed to the NKT cell–mediated conditioning observed in the previous experiments. Significantly enhanced levels of IFN-α were detected in the serum of mice injected with α-GalCer prior to TLR7 stimulation (Fig. 4D). To determine whether pDCs were the source of this IFN-α, a similar experiment was conducted in siglecH-DTR mice that were depleted of pDCs by administration of DT (Fig. 4E). When depletion was performed before either stimulus was administered, IFN-α release was ablated, indicating that pDCs were indeed the prime source of this cytokine.

It is known that IFN-α stimulation of conventional DCs can improve IL-12 production for T cell priming (36). Therefore, it was possible that the improved IL-12p40 responses seen when resiquimod was administered after α-GalCer were a consequence of exposure to IFN-α from NKT cell–conditioned pDCs. To test this, pDCs were depleted from siglecH-DTR mice before the administration of α-GalCer and delayed resiquimod (Fig. 4F, 4G). Interestingly, CD8α⁺ DCs were unable to enhance the production of IL-12p40 with delayed resiquimod when pDCs were not present (Fig. 4F). In contrast, enhanced IL-12p40 by CD8α⁻ DCs was independent of pDCs (Fig. 4G), presumably reflecting an ability to be stimulated directly via TLR7, with NKT cell–mediated conditioning facilitating this response.

Although NKT cell activation conditions langerin⁻ APCs to enhance CD8⁺ T cell responses with delayed TLR7 stimulation, it was unclear whether, under these conditions, cross-priming could actually become fully independent of langerin⁺ CD8α⁺ DCs. Therefore, cross-priming was examined in lang-DTREGFP mice depleted of langerin-expressing cells before Ag and α-GalCer were administered, leaving only langerin⁻ APCs to cross-present Ag and to be exposed to subsequent resiquimod administration after NKT cell conditioning (Fig. 4H). No significant induction of CD8⁺ T cells was observed in the absence of langerin-expressing cells in these experiments. Therefore, conditioned langerin⁻ APCs were only able to contribute to the priming of CD8⁺ T cells after cross-priming was initiated by langerin⁺ CD8α⁺ DCs.

To investigate the functional relevance of the T cells cross-primed in the context of NKT cell activation and delayed TLR7 stimulation, thereby exploiting both CD8α⁺ and CD8α⁻ DCs, we explored antitumor responses against established s.c. E.G7-OVA T cell lymphoma, which expresses OVA protein as a model of a tumor-associated Ag. To reduce the toxicity associated with delayed TLR stimulation, we combined OVA with a 100-fold lower dose of α-GalCer (2 ng/mouse), which still gave rise to the accumulation of significant levels of IFN-α in the serum once resiquimod was administered (Fig. 5A). Reduced toxic responses were indeed observed, with an average of 60% of animals unaffected. These animals showed superior antitumor responses to animals that received OVA concurrently with α-GalCer and resiquimod or OVA with either stimulus alone (Fig. 5B).

This study shows that CD8⁺ T cell responses generated to blood-borne protein Ags are primarily induced by langerin⁺ CD8α⁺ DCs in the spleen. However, depending on the stimuli provided in association with the Ag, it is possible for other DCs to participate in this response. In this context, the rapid activation of NKT cells can provide a ready source of stimuli that can condition langerin⁻ APCs to support an improved T cell burst. However, involvement of these langerin⁻ APCs was dependent on subsequent exposure to additional activatory stimuli, which, in our experiments, was provided by administration of the TLR7 ligand resiquimod. A number of consequences of NKT cell–mediated conditioning were highlighted in our study. Splenic pDCs were sensitized to make copious quantities of IFN-α in response to delayed stimulation via TLR7, and this was associated with enhanced production of IL-12p40 by CD8α⁺ DCs, including langerin⁻ and langerin⁺ cells. Additionally, NKT cell–mediated conditioning permitted CD8α⁻ DCs to enhance production of IL-12p40 in response to direct TLR stimulation. A therapeutic benefit of such NKT cell–mediated conditioning on antitumor CD8⁺ T cell responses was observed in a model of s.c. lymphoma. Overall, these studies highlight the capacity to shape an adaptive response by the coordinated activity of different DC subsets within the spleen through temporally phased activities of accessory cells, such as NKT cells, together with stimulation via pattern recognition receptors such as TLRs.

The sensitivity of mammalian organisms to bacterial insult is under constant modification due to daily contact with commensals and pathogens. Therefore, the host immune system is forced into an important balancing act between action and inaction, which is potentially resolved by requiring integration of multiple stimuli via innate pathogen recognition systems before significant immunity is unleashed. In this study, we show that the rapid activity of the large local network of NKT cells in splenic tissue is sufficient to induce DCs to stimulate T cells and that this activity can be enhanced significantly through timely TLR ligation. It was shown recently that mammalian APCs express low levels of α-linked glycosylceramides similar to the potent NKT cell agonist α-GalCer used in our studies. Although expression is tightly regulated, it is possible that increased expression of these compounds in times of stress may cause stimulation of NKT cells, creating an opportunity for the integration of stimuli observed in our studies. Alternatively, pathogenic species may supply the NKT cell agonists themselves. Bacterial glycolipids generally induce weaker NKT cell activity (37, 38), so the integration of multiple stimuli may be particularly important to drive potent adaptive responses to infection. Interestingly, our results suggest that NKT cell activation should be concomitant with or precede TLR engagement to enhance CTL responses. Using stimulation with the TLR2 agonist Pam3Cys as an example, when TLR2 stimulation was initiated before Ag was delivered, cross-priming was almost completely ablated, which likely reflects the downregulation in Ag uptake and processing that accompanies maturation of APCs (39, 40). If Ag was delivered at the time of TLR2 stimulation alone, an adjuvant effect of the TLR stimulus was observed, and a cross-primed response was induced. Significantly, the CD8⁺ T cell response induced in the presence of both TLR2 stimulation and NKT cell activation was superior in size (Fig. 1), duration, cytokine production, and cytotoxic activity (data not shown) to that induced in the presence of either stimulus alone. When NKT cell activation preceded TLR2 engagement, a potent cytokine response was observed, suggesting that signals from the NKT cells condition the APC to respond more vigorously to TLR ligation. Although this sequencing of stimuli was generally associated with an increase in cross-priming, there was a limitation; toxic levels of proinflammatory cytokines were released when TLR ligation was provided within 24 h of NKT cell activation that were detrimental to the adaptive response. This overtly toxic cytokine storm was avoided by using lower doses of NKT cell agonist or by delaying TLR2 simulation as long as 24 h after administration of Ag and NKT cell agonist. A longer delay of 48 h provided no adjuvant effect, indicating a temporal boundary to the conditioning of APCs by NKT cells. These data suggest that NKT cell–mediated conditioning provides a window in which lymphoid-resident APCs are particularly sensitive to pathogen-associated pattern recognition, thereby effectively providing a mechanism to sensitize the host to low pathogen levels.

The downstream effects of NKT cell activation were shown to alter immune function and activation throughout the spleen (41, 42). We initially focused analysis on the impact on the specialized function of CD8α⁺ DCs in cross-priming. This population can be subdivided on the basis of expression of langerin and CX₃CR1. The CX₃CR1⁺ CD8α⁺ DC subset does not express langerin and does not exhibit an enhanced capacity for cross-priming (43). CX₃CR1⁻ CD8α⁺ DCs can be divided into langerin⁺ and langerin⁻ populations, both of which express DEC205, Clec9A, and high basal levels of CD86. We reported previously that langerin⁺ cells are primarily required for cross-priming (shown again in this study) and showed more recently that the langerin⁺ CX₃CR1⁻ CD8α⁺ subset has a superior capacity for acquiring cellular material and producing IL-12 (44). Furthermore, following purification and adoptive transfer into new hosts, langerin⁻ CX₃CR1⁻ CD8α⁺ cells survive longer and upregulate expression of langerin, suggesting differentiation of a precursor population to mature cells with the classical features typically attributed to the CD8α⁺ population as a whole. Therefore, the availability of lang-DTREGFP mice conveniently provided a model to selectively deplete the most important cells involved in cross-priming and allowed us to examine the role of other APCs in their absence. Our studies suggest that langerin⁺ CD8α⁺ DCs are crucial early in the induction of cross-priming in the presence of α-GalCer and TLR ligation (within 6 h), which may reflect their key position in the MZ to screen Ags in the blood. However, a feature of NKT cell–mediated activation of langerin⁺ CD8α⁺ DCs is their susceptibility to activation-induced cell death, with as much as 70% of the population disappearing from the spleen within 6 h of administration of α-GalCer at the dose used in our studies. Therefore, it was surprising that administration of TLR agonists for TLR2 and TLR3 18 h later could still enhance cross-priming by acting on these DCs (Fig. 3). It is possible that langerin⁺ CD8α⁺ DCs that survive NKT cell activation do so because of survival factors provided within specific niches within the lymphoid architecture or that these cells may have received survival signals through interactions with Ag-specific T cells (45, 46). Regardless, the crucial role of langerin⁺ CD8α⁺ DCs in cross-priming is not simply because they are the only cells capable of cross-presentation, because Ag was effectively processed and presented in their absence (Fig. 2). Factors such as location, heightened capacity for providing costimulatory molecules, and release of appropriate chemokines and cytokines may be more relevant in this regard.

As already noted, langerin⁻ APCs were shown to be capable of cross-presenting Ag and that, when conditioned through the activity of NKT cells, they could be stimulated by delayed administration of the TLR7 ligand resiquimod to enhance cross-priming. However, it remains unclear whether these two observations are related; the ability to cross-present does not necessarily mean that this function was necessary for their role in enhancing cross-priming. In fact, the increased T cell burst size may reflect exposure to enhanced levels of stimulatory mediators released by the conditioned APCs in response to TLR ligation. In this context, we showed that conditioning by activated NKT cells permits pDCs to release large quantities of IFN-α in response to delayed resiquimod. Although this cytokine can be suppressive to immature APCs in some systems (36), in general its effect on CD8⁺ T cell responses was reported as stimulatory, notably by stimulating enhanced DC maturation and cross-presentation (47, 48), although it can also act directly on CD8⁺ T cells themselves to allow clonal expansion and memory formation (49). Our studies show that NKT cell–conditioned pDCs can provide large quantities of IFN-α, which was associated with improved conventional DC function, as indicated by enhanced IL-12p40 release. Whether it was this process that ultimately drove the enhanced CD8⁺ T cell responses observed or a direct effect of this cytokine (or others) on the T cells themselves remains unanswered.

The observation that activation of NKT cells can condition APCs to manufacture more cytokines in response to TLR ligands is consistent with a previous study conducted on splenocytes in vitro (50). Our results are significant in that they show that this conditioning can apply to many APCs and has an impact on induced T cell responses. Like the results presented in this article, other investigators showed that CD8α⁻ DCs are capable of cross-presentation (51–53) and that activation of pDCs can enhance CD8⁺ T cell responses by improving conventional DC functions (33). Collectively, these observations may have implications for vaccine design. At present, there is much interest in targeting Ags to specific DC subsets with specialized function. For cross-priming CD8⁺ T cell responses, attention has been focused on introducing Ags to CD8α⁺ DCs, or equivalent cells in humans, through targeting cell surface receptors, such as CD205 (54–57). Other investigators are conjugating Ags directly to TLR ligands that target these same cells (58–60). However, we showed that through exposure to different stimuli, including from NKT cells, different DC subsets can become involved. Therefore, the need to specifically target a given subset may not be clear-cut. It is possible that vaccines could be designed to take advantage of conditioned APCs, such as through inclusion of α-GalCer as an immune adjuvant, which, together with slow release from a depot of TLR ligands, could drive the type of enhanced T cell responses that we observed in this study. It is also possible that other innate-like T cells could be similarly exploited, such as mucosal-associated invariant T cells, which could potentially be triggered with vitamin B metabolites (61), or Vγ9Vδ2 T cells, which respond to bisphosphonates in humans (62). However, the longer-term functional significance of the enhanced CD8⁺ T cell responses seen with conditioning by innate-like T cells requires further investigation before this approach should be adopted.

We thank the personnel of the Biomedical Research Unit of the Malaghan Institute of Medical Research for animal husbandry, and Kylie Price for flow cytometry support.

The authors have no financial conflicts of interest.