NKT cells play a protective role in immune responses against infectious pathogens. However, when the NKT cell response to infection is initiated and terminated is unknown. In this study, we demonstrate that NKT cells become activated, proliferate, and exert their effector function before MHC-restricted T cells during infection with Mycobacterium bovis bacillus Calmette-Guérin in mice. After a cell expansion phase, NKT cells underwent cell death, which contracts their numbers back to baseline. Surprisingly, despite ongoing infection, the remaining NKT cells were profoundly unresponsive to TCR stimulation, while MHC-restricted T cells were vigorously proliferating and producing IFN-γ. Similarly, we show that NKT cells became unresponsive in uninfected mice after receiving a single exposure to a TLR agonist LPS, suggesting that NKT cell unresponsiveness may be a major mechanism of terminating their response in many infectious conditions. This characterization of the NKT cell response in antimicrobial immunity indicates that rapid NKT cell activation contributes to the innate phase of the response to the infectious pathogen, but then, the NKT cell response is shut down by two mechanisms; apoptotic contraction and marked unresponsiveness to TCR stimulation, as a synchronized hand off to MHC-restricted T cells occurs.
Natural killer T cells are a unique subset of T cells that were initially characterized as cells that coexpress αβTCR and NK receptors. Many NKT cells express an invariant Vα14-Jα18 TCR chain with either Vβ8.2, Vβ7, or Vβ2 in mouse, the homologous population of human NKT cells also expresses an invariant TCR with Vα24-Jα18 and Vβ11 referred to as invariant NKT cells. Unlike MHC-restricted T cells, NKT cells recognize glycolipid Ags presented by CD1d, a non-polymorphic MHC class I-like Ag-presenting molecule (1, 2, 3). NKT cells display an activated/memory phenotype and express CD69 and CD44 even in germfree mice. Similarly, human NKT cells from cord blood are CD45RO+CD45RA−CD62L−CD25+, suggesting that NKT cells are already partially activated at steady state. A remarkable hallmark of NKT cell function is their capacity to produce large amounts of cytokines very rapidly upon activation by anti-CD3 or the pharmacological mimic, α-galactosylceramide (αGalCer)3 NKT cells constitutively express mRNAs for cytokines, such as IL-4 and IFN-γ, which may enable them to produce these cytokines within hours of stimulation (4).
NKT cells are activated within 1–2 h after administration of αGalCer into mice and up-regulate surface molecules such as CD69 and CD40L and produce both Th1 and Th2 cytokines (5, 6, 7). NKT cell activation, in turn, activates innate and adaptive immune cells, including NK cells, macrophages, dendritic cells (DC), MHC-restricted T cells, and B cells (1, 2, 3). For instance, the interaction between CD40L on NKT cells and CD40 on DC induces the maturation of DC. The effector cytokines such as IFN-γ produced by activated NKT cells induce NK cell proliferation, IFN-γ production, and cytotoxicity.
NKT cells have been implicated to be involved in various types of immune responses including autoimmunity, antitumor responses, allergy, inflammation, and infection (1, 2, 3). Increasing evidence suggests that NKT cells are activated and play a protective role against a variety of pathogens during infection (8, 9, 10). For instance, the accumulation of NKT cells was observed in lungs infected with Cryptococcus neoformans or in lymphoid tissues following infection with Leishmania major and Mycobacterium bovis bacillus Calmette-Guérin (BCG) (11, 12, 13). IFN-γ production by NKT cells has been described after infection with Salmonella, Sphingomonas, and Ehrlichia (14, 15, 16). Moreover, studies using NKT cell-deficient mice showed increased susceptibility to Streptococcus pneumoniae, Borrelia burgdorferi, and Trypanosoma cruzi infection (17, 18, 19, 20, 21). Reduced clearance of pathogen was reported in CD1d or NKT cell-deficient mice in infection with Pseudomonas aeruginosa, Leishmania donovani, and herpes simplex I/II (22, 23, 24, 25). NKT cell activation by αGalCer inhibited the clinical course of infection with S. pneumoniae, Mycobacterium tuberculosis, CMV, T. cruzi, and Plasmodium yoelii and Plasmodium berghei (17, 26, 27, 28, 29).
However, unlike the situation following αGalCer administration, when and how NKT cells become activated and exert their effector function during a natural course of infection in vivo and the manner by which the response is terminated are largely unknown. In this study, we report that the activation of NKT cells occurred before that of MHC-restricted T cells during infection with M. bovis BCG in mice. After expansion of activated NKT cells, the majority of NKT cells underwent cell death. Surprisingly, despite ongoing infection and an abundance of microorganisms in the tissues, the persisting NKT cells after contraction were profoundly unresponsive to TCR stimulation, at the time that MHC-restricted T cells were actively proliferating and producing IFN-γ. Furthermore, NKT cells became unresponsive in uninfected mice that received microbial TLR agonists like LPS, suggesting that NKT cell unresponsiveness may be a major mechanism of terminating their response in many infectious diseases.
Materials and Methods
Female C57BL/6 mice and thymectomized female C57BL/6 mice were purchased from The Jackson Laboratory. All animal studies were approved by the Dana-Faber Cancer Institute Office for the Protection of Research Subjects. The animals were kept in a specific pathogen-free or a biosafety level 3 animal facility and used at 8–12 wk of age. C57BL/6 mice received 1 × 106 CFU of M. bovis BCG in 0.9% NaCl (100 μl) or 0.9% NaCl (100 μl) i.v. In LPS experiments, C57BL/6 mice were injected with 40 μg of Salmonella LPS in 0.9% NaCl (100 μl) or 0.9% NaCl (100 μl) i.v.
Bacteria and reagents
M. bovis BCG strain Pasteur was provided by B. R. Bloom (Harvard School of Public Health, Boston, MA). A single colony of M. bovis BCG was grown to mid-log phase in Middlebrook 7H9 medium supplemented with albumin-dextrose complex at 37°C and aliquots were frozen at −80°C until used. Salmonella LPS and BrdU were purchased from Sigma-Aldrich.
Liver mononuclear cells were prepared by Percoll density gradient centrifugation. Splenocytes were obtained by pressing the spleen though a 70-μm cell strainer. For both organs, erythrocytes were lysed with RBC lysis buffer. Bone marrow-derived DC (BM-DC) were grown from bone marrow progenitors for 6 days in the presence of GM-CSF and IL-4 (R&D Systems) in complete RPMI 1640 medium (RPMI 1640 supplemented with l-glutamine and penicillin/streptomycin; Life Technologies) containing 10% FBS (HyClone). αGalCer (100 ng/ml) was added to the medium on day 6 and αGalCer-pulsed BM-DC were harvested on day 8.
Flow cytometry and cytokine secretion assays
Cells were incubated with anti-CD16/32 to avoid nonspecific staining and then stained with fluorescence-labeled Abs including anti-CD19-PerCP-Cy5.5 and anti-PE-Cy7, anti-CD3ε-FITC and anti-PerCP-Cy5.5, anti-NK1.1-PE, anti-CD69-PECy7 (BD Biosciences), and CD1d/PBS-57 tetramer conjugated with allophycocyanin (National Institutes of Health tetramer facility). PBS-57 is an analog of αGalCer and CD1d/PBS-57 tetramers have been shown to stain NKT cells comparably to CD1d/αGalCer tetramers. The NKT cell population was identified as CD19-negativeCD3-positive CD1d tetramer-positive cells and the population of non-NKT T cells was defined as CD19-negativeCD3-positive CD1d tetramer- negative cells. Samples from infected animals were fixed with 1% paraformaldehyde and stored at 4°C overnight before analysis. Intracellular staining of active caspase 3 was performed after fixation and permeabilization with Perm/Cytoperm solution (BD Biosciences). In BrdU experiments, both infected and uninfected mice were treated with 0.8 mg/ml BrdU in the drinking water for 7 days before tissue collection, and BrdU staining was performed following the manufacturer’s instructions (BD Biosciences). For cytokine secretion assays, liver and spleen cells were first stained with anti-CD3ε-FITC, anti-CD19-PE-Cy7, and CD1d tetramer-allophycocyanin and used for cytokine secretion assays according to the manufacturer’s instructions (Miltenyi Biotec). Dead cells were excluded by staining with 7-aminoactinomycin D. The samples were analyzed with FACSCanto (BD Biosciences). The FlowJo software was used to analyze the data (Tree Star).
The organs were homogenized in 0.9% NaCl/0.02% Tween 80 with a Mini-Bead Beater 8 (Biospec Products). Serial dilutions of organ homogenates were plated onto 7H11 Mitchinson agar plates (Remel), and colonies were counted after 2 wk of incubation at 37°C.
The Welch’s t test was used for statistical analysis.
NKT cell activation during M. bovis BCG infection
To investigate physiological NKT cell activation during the natural course of mycobacterial infection, we inoculated mice with M. bovis BCG i.v. and detected NKT cells isolated from liver and spleen. As previously reported, NKT cells were defined by double staining with anti-CD3 mAb and CD1d/PBS-57 tetramers, hereafter called CD1d tetramers. In uninfected animals, CD3-positive CD1d tetramer-positive cells were found to express intermediate levels of the early activation marker CD69 (Fig. 1,a and data not shown). CD69 expression increased at day 7 postinfection and increased even further at day 14 (mean fluorescence intensity (MFI), from 657 before infection to 1758 in the liver and from 579 to 1042 in the spleen at day 7; Fig. 1, a and b, and data not shown). It is well known that NKT cells express a series of surface NK receptors, one of which, NK1.1, has been shown to be rapidly down-regulated after activation with αGalCer. Therefore, we also analyzed NK1.1 expression from NKT cells of M. bovis BCG-infected mice. In uninfected mice, 90% of liver and 70% of spleen NKT cells expressed NK1.1, similar to published results. Following infection with M. bovis BCG, NK1.1 expression was markedly down-regulated on NKT cells (Fig. 1,c), consistent with NKT cell activation during M. bovis BCG infection. However, unlike αGalCer-stimulated NKT cells, NKT cells activated by infection in vivo continued to express readily detectable levels of TCR as evidenced by CD3 and CD1d tetramer staining (Fig. 1 a).
NKT and T cell proliferation during M. bovis BCG infection
Next, we analyzed the numbers of NKT cells and T cells (defined as CD3+ and CD1d tetramer-negative cells) in the liver and spleen of mice infected with M. bovis BCG to determine whether they undergo expansion following activation by the infection. The total number of NKT cells both in the liver and spleen at day 7 postinfection was increased (from 1.2 × 106 ± 0.2 × 106 before infection to 6.9 × 106 ± 0.8 × 106 in the liver and from 1.4 × 106 ± 0.2 × 106 to 3.2 × 106 ± 0.2 × 106 in the spleen at day 7; Fig. 2,a), although the proportion of NKT cells among lymphocytes at day 7 postinfection was decreased in the liver (data not shown). However, by day 14 postinfection, a reduction back to near baseline numbers of NKT cells was noted (2.2 × 106 ± 0.3 × 106 in the liver and 1.0 × 106 ± 0.1 × 106 in the spleen; Fig. 2,a). In comparison, the total number of T cells started to increase at day 7 and peaked at day 14 postinfection (from 1.1 × 107 ± 0.1 × 107 before infection to 47.6 × 107 ± 5.1 × 107 in the liver and from 4.0 × 107 ± 0.4 × 107 to 11.9 × 107 ± 0.9 × 107 in the spleen at day 14; Fig. 2 a).
To determine whether the increase in the number of NKT cells following M. bovis BCG infection resulted directly from proliferation of NKT cells, we infected mice with M. bovis BCG i.v. and analyzed BrdU incorporation into NKT cells at days 9 and 21 postinfection. Mice were offered 0.8 mg/ml BrdU in drinking water for 7 days, after which they were sacrificed and tissues were collected. Fifty percent of the liver NKT cells and 36.7% of spleen NKT cells were positive for BrdU from mice infected with M. bovis BCG 9 days before, indicating active proliferation of NKT cells (Fig. 2,b). However, <10% of BrdU- positive cells were observed among NKT cells at day 21. Among non-NKT T cells, BrdU-positive cells were found in the liver (46.0%) and spleen (18.4%) at day 9 postinfection and BrdU-positive cells were increased to 62.5% (liver) and 32.2% (spleen) at day 21 postinfection (Fig. 2 b). Thus, these studies show that NKT cells proliferate during the early phase of infection and that NKT cell proliferation then diminishes by the time that MHC-restricted T cells begin to actively expand. This suggests that NKT cell activation and proliferation occur before the major MHC-restricted T cell activation, but then wanes as the adaptive immune response continues.
NKT cell death during M. bovis BCG infection
We noted that although the total number of NKT cells was increased by day 7 postinfection with M. bovis BCG, it was followed by a marked reduction of NKT cell numbers by day 14 (Fig. 2,a). The proportion of BrdU-positive NKT cells was lower at day 21 than at day 9 postinfection (Fig. 2 b), indicating that, during infection, NKT cells are not actively proliferating at the latter time point. Although these findings support the notion that the number of NKT cells do not increase after day 7 postinfection, it does not explain why the number of NKT cells dropped back toward their starting numbers at day 14 postinfection. In typical infections, as a pathogen replicates, naive T cells become activated and proliferate, some of them differentiating into effector cells. When the pathogen is eliminated, the majority of effector MHC-restricted T cells undergo cell death and only a fraction of the T cells survive or become memory T cells (30). However, in the case of M. bovis BCG infection, the bacterial numbers remain high at the time that NKT cell numbers contract (data not shown). Therefore, we next determined if NKT cells decline from the periphery postexpansion as a result of cell death.
To determine whether NKT cells die postinfection, we stained NKT cells from infected mice for active caspase 3, a marker for cells undergoing apoptosis (31). Postinfection NKT cells were found to have a higher percentage of active caspase 3-positive cells compared with NKT cells from uninfected mice (active caspase 3-positive cells, increased from baseline 1.6 to 9.0% in the liver on day 7 and from 1.5 to 7.9% in the spleen on day 11; Fig. 3,a). Thus, the proliferation of NKT cells during infection is followed by a contraction of NKT cell numbers due to cell death. Fas-FasL signaling is believed to be crucial for activation-induced cell death of MHC-restricted T cells (32, 33). Therefore, we analyzed the surface expression of Fas on NKT cells during infection. Fas was up-regulated on NKT cells following infection, increasing from MFI 225 to 568 by day 9 postinfection in the liver (Fig. 3, b and c, and data not shown). This is consistent with the NKT cells becoming more susceptible to apoptosis postinfection.
In summary, we found that, during the natural course of M. bovis BCG infection, NKT cells are activated and proliferate before the MHC-restricted T cell response. Then, when the MHC-restricted T cells start to actively proliferate, the number of NKT cells contracts as a result of reduced proliferation and cell death.
Effector functions of NKT cells during M. bovis BCG infection
One of the important effector functions of NKT cells is to produce inflammatory cytokines such as IFN-γ, which is known to play an important role in the immune response against intracellular pathogens such as M. tuberculosis and M. bovis BCG (34, 35, 36). We next investigated the ability of NKT cells to produce cytokines following infection with M. bovis BCG. Cytokine secretion assay staining showed IFN-γ- positive NKT cells after day 2 postinfection, and the percentage of IFN-γ-positive NKT cells peaked at day 7 postinfection (increased from 3% before infection to 25% in the liver and from 3.4 to 24% in the spleen at day 7; Fig. 4,a). However, on day 14 postinfection, <5% of NKT cells remained positive for IFN-γ (Fig. 4,a). On the contrary, IFN-γ-positive MHC-restricted T cells persisted in the liver and were still increasing in the spleen even 14 days after infection (Fig. 4 b). Neither NKT cells nor MHC-restricted T cells produced IL-4 during infection (data not shown). Thus, NKT cells show their cytokine effector function during the early phase of M. bovis BCG infection but then lose their capacity to produce IFN-γ at later time points; in contrast, MHC-restricted T cells both in the liver and spleen, actively produce IFN-γ after 14 days postinfection.
We next sought to understand why NKT cells had stopped responding to the infection, even though MHC-restricted T cells were still proliferating and producing IFN-γ while the bacterial burden in the tissues persisted at significant levels (CFU per organ at days 7 and day 14, 2.7 × 105 and 2.6 × 105 for the liver and 7.6 × 104 and 1.5 × 105 for the spleen). NKT cells have previously been reported to become unresponsive after activation by αGalCer (37, 38), thus we set out to investigate whether NKT cells become unresponsive also after physiological activation by M. bovis BCG infection. To investigate the ability of NKT cells to produce cytokines during the course of infection, at different time points postinfection, we administered αGalCer-loaded BM-DC i.v. to determine whether the NKT cells could respond to this potent agonist. In uninfected mice, 60% of liver NKT cells stained positive for IFN-γ and 35% of liver NKT cells stained positive for IL-4 3 h after injection of αGalCer-loaded BM-DC (Fig. 5). In infected mice, the level of IFN-γ and IL-4 secreted by NKT cells from mice at day 7 after infection as a result of αGalCer-loaded BM-DC i.v. injection was found to be similar to that of NKT cells from control mice. Strikingly however, the cytokine production ability of NKT cells from infected mice on days 14, 18, and 21 postinfection was significantly reduced compared with the control NKT cells. In addition, the injection of αGalCer-loaded BM-DC induced CD69 up-regulation on liver NKT cells from uninfected mice (fold increase of MFI, 1.46). However, liver NKT cells at day 14 after M. bovis BCG infection did not up-regulate CD69 upon stimulation by αGalCer-loaded BM-DC (fold increase of MFI, 1.01). These results show that NKT cells become unresponsive during infection by M. bovis BCG when tested by αGalCer stimulation and the unresponsive state of NKT cells was not related to the bacterial burdens in the organs (Fig. 5,b). Thus, during the natural course of M. bovis BCG infection, first NKT cells become activated and produce IFN-γ, but then lose their protective role against the pathogen as their numbers decline as a result of reduced proliferation and cell death, and unexpectedly those NKT cells that persist are unresponsive and have lost their ability to produce cytokines. Although this unresponsiveness showed gradual recovery after 28 days postinfection (Fig. 5,a), the NKT cells in vivo were no longer producing IFN-γ at levels as highly as the beginning of infection, despite the ongoing presence of microorganisms. For example, there was a lower proportion of IFN-γ-producing liver NKT cells at time points after 35 days postinfection compared with day 7 (Fig. 5 a, □). This may reflect the difference in responsiveness to the powerful pharmacological αGalCer agonist compared with the in vivo signals for NKT cell activation during infection, which are likely to be weaker TCR agonists and/or cytokines such as IL-12 and IL-18.
NKT cell unresponsiveness to αGalCer stimulation is followed by recovery from the unresponsive state and both of these changes are independent of thymic influences
Most NKT cells in the periphery are derived from the thymus where positive selection is mediated by CD1 expression on CD4+CD8+ thymocytes (1, 2, 3). We investigated what role the thymus plays in the induction of peripheral NKT cell unresponsiveness and late recovery from the anergic state. Mice were thymectomized at 8 wk of age to allow development and population of the periphery with mature NKT cells. Three to 4 wk later, thymectomized mice were infected with M. bovis BCG i.v. We examined NKT cell cytokine-producing ability upon TCR stimulation 14–18 days after infection and noted that NKT cells showed impaired cytokine production upon αGalCer stimulation, compared with control NKT cells from uninfected mice (Fig. 6,a). However, NKT cells from 35 days postinfection no longer showed a defect in cytokine production upon stimulation (Fig. 6,a). These changes in responsiveness occurred while the levels of bacterial burden in the tissues persisted at similar levels (Fig. 6 b). These results indicate that both the induction of NKT cell unresponsiveness and the recovery of NKT cells from unresponsiveness occur during the natural course of infection in vivo and these effects are thymus independent.
NKT cells become unresponsive after LPS exposure
The NKT cell unresponsiveness noted above occurred in mice that had ongoing active M. bovis BCG infection, assessed by CFU that remained at the same levels throughout the 21-day analysis (data not shown). This made it unclear as to whether or not the postinfection NKT cell anergy was the physiological response that occurs following infection-induced activation or whether it was a result of persisting infection. To model a one-time NKT cell activation without concomitant infection, we used a single injection of the TLR4 agonist LPS. Previously, we have shown that NKT cells can be activated by a TLR agonist, LPS, through its ability to stimulate DC to produce IL-12, which in turn amplifies NKT cell reactivity against self-lipids presented by CD1d (14). Recently, Nagarajan and Kronenberg (39) also reported that NKT cells produce IFN-γ following LPS injection into mice. Therefore, we treated C57BL/6 mice with LPS i.v. and analyzed CD69 surface expression on NKT cells. We noted that CD69 was up-regulated on NKT cells 24 h after LPS i.v. (Fig. 7,a), consistent with NKT cell activation, while there was no significant reduction in CD3/TCR levels on NKT cells following LPS i.v. (Fig. 7,a). Next, we studied the cytokine production by NKT cells of LPS-treated mice upon αGalCer stimulation. At the indicated time points after an i.v. injection of LPS, mice received αGalCer-loaded BM-DC to determine whether NKT cells could be stimulated to secrete cytokines. Approximately 50% of NKT cells from the control vehicle-treated animals were positive for IFN-γ upon αGalCer stimulation, whereas at 2–3 days after LPS injection, only 3% of NKT cells from both liver and spleen became cytokine positive (Fig. 7,b). However, the cytokine-producing ability of NKT cells upon TCR stimulation recovered by day 7 after LPS injection (Fig. 7 b). This striking difference indicates that NKT cells become unresponsive to restimulation by 2 days after LPS activation, just as they had, but with a different time course, when they became unresponsive 14 days after live mycobacterial infection.
Two major mechanisms have been proposed for NKT cell activation. The first mechanism is direct Ag recognition by NKT TCR. For instance, αGalCer or exogenous Ags from microbial organisms, such as α-glucuronosylceramide from Sphingomonas and α-galactosyldiacylglycerol from B. burgdorferi, are presented by CD1d (16, 40, 41). The second mechanism is amplification of NKT cell self-reactivity by inflammatory cytokines such as IL-12 and IL-18 (14, 39). In addition, the TLR 9 agonist CpG oligodeoxynucleotide-activated DC were reported to stimulate NKT cells by producing type I IFN and de novo- charged glycosphingolipid (42). Therefore, NKT cell activation during infection with microorganisms is thought to be mediated either by microbial lipids that are cognate Ags for the NKT TCR and/or by inflammatory cytokines and self-lipid Ags. Even though NKT cell activation has been well studied by using αGalCer, a strong mimic of NKT cell Ags, little is known about how activated NKT cells respond physiologically in infection. In this study, we examined the natural course of the NKT cell response after M. bovis BCG infection in vivo and found that NKT cells are distinct from MHC-restricted T cells in the timing and in the process of activation termination. First, the expansion and effector function of NKT cells peaked before those of MHC-restricted T cells, consistent with their role as innate-like lymphocytes participating in the early phase of host defense. Then, the NKT cells underwent apoptosis after activation and those NKT cells that remained became unresponsive to TCR stimulation. This second phase of response termination occurs once the MHC-restricted T cell response is underway. Thus, the NKT and the MHC-restricted T cell responses are synchronized. Importantly, the NKT cell response itself goes through a cycle of activation and expansion followed by a contraction and inactivation phase, finally culminating in the recovery in capacity for activation.
MHC-restricted T cells are important for immunity against mycobacterial infection, including both Th1 CD4 T cells and CD8 T cells (43, 44, 45). The process for the development of a MHC-restricted T cell response requires 2–3 wk postinfection. Consistent with M. tuberculosis, the MHC-restricted T cell proliferation and IFN-γ production to M. bovis BCG were active after 2–3 wk postinfection. In addition, BrdU experiments demonstrated that MHC-restricted T cells were still increasing at day 21. In contrast, we found that NKT cell number and IFN-γ production peaked at day 7 postinfection. Although the factor responsible for NKT cell activation during mycobacterial infection is not known for certain, there are several TLR agonists found in mycobacterium, including the 19- kDa lipoprotein, PIM2, PIM6, and lipoarabinomannan, and whole M. tuberculosis or these TLR ligands have been shown to activate APCs via TLR2 or TLR4 (46, 47, 48). Thus, activation of NKT cells by the cytokine-mediated mechanism is likely.
Importantly, during M. bovis BCG infection, the early expansion of NKT cell number was then followed by a rapid contraction so that cell numbers were almost back to baseline by day 14. Furthermore, BrdU incorporation in NKT cells at the latter time point was comparable to that observed in control uninfected mice, suggesting that they were no longer actively proliferating. Highly potent but nonphysiological αGalCer activated NKT cells are known to down-modulate surface markers and become undetectable by CD1d tetramer staining, even while actively proliferating. However, this was not observed during M. bovis BCG infection, as we did not detect TCR down-modulation of NKT cells before their disappearance and NKT cell number did not increase after the contraction phase. This difference compared with αGalCer-mediated activation may be the result of less strong and more gradual physiological stimulation of NKT cells in microbial infection and the mechanism of cytokine-driven stimulation.
Upon the entry of pathogen into the host, Ag-specific MHC-restricted T cells expand and differentiate into effector cells. When the pathogen is cleared from the host, most T cells die by activation-induced cell death or as a result of the reduction in inflammatory cytokines, and only a small fraction of memory cells persists in the host (30, 32, 33). We demonstrated that there were more active caspase 3-positive cells among postinfection NKT cells than among NKT cells form uninfected mice, suggesting that NKT cells also undergo apoptosis after expansion. Furthermore, we found that NKT cells up-regulated the expression of Fas, suggesting at least one pathway by which apoptosis may be induced. It has been reported that the number of NKT cells was decreased as a result of apoptosis after an acute infection with lymphocytic choriomeningitis virus (LCMV) (49, 50). However, following LCMV infection, NKT cells did not expand before the reduction in cell number. In addition, the NKT cell death after acute LCMV infection was shown to be independent of the FasL-Fas pathway and this may result from a direct LCMV infection to NKT cells. We noted that the kinetics of active caspase 3-positive NKT cells was different in the liver and spleen. However, the proportions of Fas high-expressing NKT cells in the liver and spleen were similar at days 7 and 9 following M. bovis BCG infection (data not shown). It has been reported that NKT cells and MHC-restricted T cells share cytokines for cell survival (51). MHC-restricted T cells produce cytokines that inhibit survival or proliferation of other cell types. Although it is unknown whether MHC-restricted T cells regulate NKT cell survival, more rapid MHC-restricted T cell proliferation in the liver compared with that in the spleen may be related to the NKT cell death at the early time point following M. bovis BCG infection. Further studies are required to understand the mechanisms of NKT cell death in infection.
Interestingly, the NKT cell contraction occurred during the active phase of infection and the NKT cells persisting postcontraction still expressed activation markers (such as CD69), but they had stopped producing IFN-γ. In contrast, at the same time point, MHC-restricted T cells were proliferating and producing IFN-γ. To understand why NKT cells ceased responding to the infection, we investigated whether NKT cells were still able to become activated and produce cytokines upon pharmacological stimulation by αGalCer. We found that the persisting NKT cells were unresponsive to stimulation with αGalCer. Upon i.v. administration of αGalCer-loaded BM-DC from uninfected mice, NKT cells in mice infected with M. bovis BCG 14–21 days before produced neither IFN-γ nor IL-4. These results indicate that NKT cells shutdown their capacity for Ag-driven responsiveness during M. bovis BCG infection by at least two processes, contraction of cell number and unresponsiveness to TCR stimulation. This shutdown of NKT cells seems different from that of MHC-restricted T cells where persisting memory T cells after contraction have a potential to proliferate and survive restimulation, enabling anamnestic response against a subsequent infection with the same pathogen. NKT cells, on the other hand, as innate-like lymphocytes appear to return to a basal state and display a finite, self-terminating pathogen response. NKT cells account for a significant proportion of T cells in the liver and spleen. The frequency of naive Ag-specific MHC-restricted T cells is 1 in 104–106 cells, whereas the NKT cell proportion among T cells is 3% in the spleen and 50% in the liver. Overactivation of NKT cells is known to mediate immunopathology such as liver damage, arthritis, atherosclerosis, and contact hypersensitivity (52, 53, 54, 55, 56, 57, 58). Therefore, it would have an adverse effect on the host if such a large population of cells continued expanding and producing effector cytokines without resolution. The timing of NKT cell shutdown once the adaptive immune response is manifest suggests a synchronized hand off to continue an effective T cell response.
To determine the mechanism of unresponsiveness in NKT cells, we analyzed changes in expression of selected stimulatory and inhibitory receptors on NKT cells after M. bovis BCG infection or LPS exposure. We did not observe up-regulation of inhibitory Ly49C, Ly49I, Ly49G2, or Ly49A NK receptors on NKT cells during the period of anergy. Although stimulatory NK receptors such as NK1.1 and NKG2D were down-regulated on NKT cells following infection, the timing of NK1.1 down-regulation did not correlate with that of NKT cell unresponsiveness (Fig. 1,c and data not shown). PD-1 expression was markedly up-regulated on MHC-restricted T cells following infection; however, only minimal changes in the expression of PD-1 and CTLA-4 were noted on NKT cells (data not shown). In addition, in contrast to the marked reduction in surface TCR levels noted following αGalCer injection, no significant reduction in TCR/CD3 levels on NKT cells was noted after infection or LPS exposure (Figs. 1,a and 7 a). Furthermore, only live NKT cells were analyzed for cytokine-producing ability upon αGalCer stimulation. Therefore, NKT cell death is not responsible for unresponsiveness following M. bovis BCG infection or LPS exposure.
We demonstrated that NKT cells also became unresponsive in mice after administration of LPS, suggesting that NKT cell unresponsiveness is likely to occur in other infectious conditions, including infection with the vast majority of microorganisms that contain TLR agonists capable of stimulating cytokine-driven NKT activation. Together, our finding regarding NKT cells reveals an elaborate complementation of innate-like and adaptive T cells in the host response to infection (Fig. 8).
We thank National Institutes of Health tetramer facility for providing the mouse CD1d tetramer. We also thank Manfred Brigl for reading this manuscript.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by the Arthritis foundation (to A.C.) and the National Institutes of Health (R01AI063428 to M.B.B.). G.S.B. acknowledges support in the form of a Personal Research Chair from James Bardrick, Royal Society Wolfson Research Merit Award, as a former Lister Institute-Jenner Research Fellow, the Medical Research Council, and the Wellcome Trust (081569/Z/06/Z).
Abbreviations used in this paper: αGalCer, α-galactosylceramide; DC, dendritic cell; BCG, bacillus Calmette-Guérin; BM-DC, bone marrow-derived DC; MFI, mean fluorescence intensity; LCMV, lymphocytic choriomeningitis virus.