Peripheral NK1.1⁻ NKT Cells Are Mature and Functionally Distinct from Their Thymic Counterparts¹

Type I NKT cells (hereafter referred to as NKT cells) are T lymphocytes that express a semi-invariant TCR capable of recognizing glycolipid Ags in the context of CD1d (1). There are strong associations between NKT cell defects and diseases involving autoimmunity, cancer, and infection that collectively suggest NKT cells play a pivotal role in immune regulation (2, 3). A wide array of regulatory activity has been attributed to the NKT cell compartment as a whole, and it is most interesting that they appear to be capable of suppressing or promoting immune responses depending on the setting. This functional diversity has led to the suggestion that distinct subsets exist, a view increasingly supported by studies in humans and mice (4, 5, 6, 7). The most clearly defined NKT cell subsets exist in the thymus, where the differential expression of cell surface Ags can be used to identify cells at different stages of development (8, 9). The most significant of these Ags is NK1.1 (CD161c in humans), whose expression marks a crucial checkpoint late in NKT cell development and is used to broadly distinguish immature and mature cells (10, 11, 12).

There is considerable evidence that immature NKT cells do not express NK1.1. NK1.1⁻ NKT cells appear before NK1.1⁺ NKT cells during ontogeny in the thymus and periphery, and thymic NK1.1⁻ NKT cells differentiate to become NK1.1⁺ (but not vice versa) following intrathymic or i.v. transfer (10, 12, 13). Thymic NK1.1⁻ NKT cells are also functionally distinct from NK1.1⁺ NKT cells, producing higher levels of IL-4 and less IFN-γ than their NK1.1⁺ counterparts (11, 12). It is therefore curious that most NKT cells exported to the periphery carry the immature NK1.1⁻ phenotype. The reasons for this are not clear, but many recent thymic emigrant (RTE)⁴ NKT cells up-regulate NK1.1 within days of export (12), suggesting that the late stages of NKT cell development that coincide with NK1.1 expression can occur in the thymus or periphery, whereas many NK1.1⁺ NKT cells are retained in the thymus for reasons that are not clearly understood (14).

The up-regulation of NK1.1 by RTE in the periphery has supported the idea that peripheral NK1.1⁻ NKT cells are at a comparable developmental stage to those in the thymus (8, 15). However, a direct study of the functional and developmental status of peripheral NK1.1⁻ NKT cells is lacking. We questioned whether RTE alone could support the consistently high (often >20%) levels of peripheral NK1.1⁻ NKT cells reported in multiple studies (10, 16, 17), because RTE make up <1% of mainstream peripheral T cells (18, 19) and fewer than 5% of RTE are NKT cells (10, 12). It also seems unlikely that that the consistently high levels of NK1.1⁻ NKT cells reported in young naive mice, in specific pathogen-free facilities, could be due to the phenomenon of NK1.1⁺ NKT cells down-regulating NK1.1 as a result of activation (20, 21). This study explores the alternative view that the majority of peripheral NK1.1⁻ NKT cells are neither immature, nor previously activated, and instead form a previously unrecognized, fully mature NKT cell subset.

Inbred CD45.2⁺ and CD45.1⁺ C57BL/6 mice were bred at the Department of Microbiology and Immunology Animal House, University of Melbourne. The studies were reviewed and approved by the appropriate University of Melbourne animal ethics committee.

The upper part of the thoracic cavity of anesthetized mice was carefully opened to enable removal of the thymus with an aspirator. Sham thymectomized mice had the thymus exposed, but not removed. The wound was closed with surgical staples, and the mice were kept warm until fully recovered. The completeness of thymectomy was checked when the mouse was killed and those with remnants of thymus tissue were excluded from analysis.

Lymphocytes were isolated from the thymus and spleen by gentle grinding between two frosted glass slides into PBS containing 2% FCS (FCS.PBS). The hind femur was flushed with FCS.PBS to collect bone marrow. Liver lymphocytes were isolated by gently pressing perfused liver tissue through 200-μm mesh sieves into FCS.PBS, then removing hepatocytes and cellular debris via a 33% isotonic Percoll (Amersham Biosciences) density gradient at room temperature. Liver, spleen, and bone marrow cells were depleted of RBC using red cell lysis buffer (Sigma-Aldrich). For NKT cell enrichment of thymocytes, cells were labeled with anti-CD8 (clone 3.155; grown in house) and anti-CD24 (clone J11D; grown in house). Ab-bound cells were then depleted using rabbit complement (C-six Diagnostics). Clumping of dead cells was avoided using DNase and viable cells isolated using a Histopaque 1083 density gradient (Sigma-Aldrich) conducted at 400 × g at room temperature. Cells were washed before being surface labeled for flow cytometric sorting.

Cell suspensions were labeled with mixtures comprised of the following fluorochrome-conjugated Abs: anti-CD3 (clone 145–2C11), anti-αβTCR (H57–597), anti-CD4 (RM4–5), anti-NK1.1 (PK-136), anti-CD45.1 (A20), anti-CD45.2 (104), anti-IFN-γ (XMG1.2), anti-IL-4 (11B11), anti-NKG2A/C/E (20d5), anti-NKG2D (CX5) and anti-Ly49C/I (5E6). All flow cytometry reagents were purchased from BD Biosciences, unless otherwise indicated. Intracellular staining for cytokines and BrdU was performed using the appropriate staining kits from BD Biosciences. Data were collected on a LSR2 flow cytometer (BD Biosciences). Fluorochrome-labeled CD1d tetramer loaded with α-galactosylceramide (αGC) was produced in house using a construct provided by M. Kronenberg (La Jolla Institute for Allergy and Immunology, La Jolla, CA). To avoid nonspecific binding of Abs to FcRγ, cells were routinely incubated with anti-mouse CD16/32 (clone 2.4G2) (grown in house). Following cell surface labeling, cells were sometimes sorted using a FACSAria (BD Biosciences). A sample of sorted cells was routinely analyzed to assess the purity of these populations, which was always greater than 95% unless otherwise stated. Data was analyzed using CellQuest (BD Biosciences) or FlowJo (Tree Star) software.

Washed cell suspensions were labeled in 1 ml 0.1% BSA.PBS with 4 ml of 1 mM CFSE (Molecular Probes) for 10 min at 37°C in the dark. The reaction was quenched with 20% FCS.PBS before cells were washed twice in RPMI 1640 culture medium and immediately used.

Thymic lobes were removed from embryos at day 15 of gestation and cultured for 6 days in culture on the surface of 0.45-μm pore size filters resting on Gelfoam gelatin sponges placed (and previously soaked) in 2 ml FTOC culture medium (RPMI 1640 supplemented with 10% v/v FCS, 2 mM glutamax, 10 mM HEPES, 0.5 mg/ml folic acid, and 0.2 mg/ml glucose (RPMI-FTOC)). After culture, the lobes were placed in Terasaki plates, 2 lobes/well, containing sorted populations of 1.5–3.3 × 10⁵ NKT cells in 30 μl of RPMI-FTOC medium. The Terasaki plates were gently inverted, forming a hanging drop, and incubated overnight at 37°C, 5% CO₂. The lobes were then cultured for 5 days in 2-ml cultures in RPMI-FTOC. Lobes were carefully disrupted using glass coverslips to release the cells for FACS analysis.

Before intracellular flow cytometry, cells were stimulated for 2 h in medium supplemented with PMA (10 ng/ml), ionomycin (5 ng/ml), and Golgistop (0.067%) (BD Biosciences).

Mice were treated with 0.8 mg/ml BrdU (Sigma-Aldrich) in drinking water for 6 days before harvest. Water was shielded from light and changed every 48 h. BrdU incorporation was determined by flow cytometry using a commercially available kit (BD Biosciences).

Because NKT cells leave the thymus in an immature NK1.1⁻ state, we directly tested whether the peripheral NK1.1⁻ NKT cell pool was mainly comprised of RTE. We reasoned that if peripheral NK1.1⁻ NKT cells were immature and destined to become NK1.1⁺, removing the thymus from adult mice should result in a progressive loss of NK1.1⁻ cells, because no “new” immature RTE NK1.1⁻ precursors would be exported from the thymus, and preexisting NK1.1⁻ NKT cells would reach maturity and become NK1.1⁺.

Groups of 8-wk-old mice were thymectomized (or sham thymectomized) and individual mice were sacrificed every 4 wk over 4 mo. NK1.1⁻ NKT cells were clearly identifiable in thymectomized mice at all timepoints and the number and frequency of these cells in the spleen, liver, and bone marrow was measured (Fig. 1). Consistent with previous reports, thymectomy caused some variability between the groups (22, 23), with moderate falls in overall T cell (data not shown) and NKT cell numbers (particularly in spleen) in thymectomised mice (Fig. 1,C). However, there was no selective effect on either of the NK1.1⁻ or NK1.1⁺ subsets and the relative proportions of NK1.1⁻ and NK1.1⁺ NKT cells remained remarkably stable (Fig. 1,B). The expression of CD4 by NKT cells was also unaffected by thymectomy, as was the expression of activation markers such as CD69 and CD44 (Fig. 1 A and data not shown).

The persistence of peripheral NK1.1⁻ NKT cells after thymectomy suggested that most of these cells were a mature, stable population. We directly examined this idea in a number of ways. We first compared the differentiation potential of peripheral and thymic NK1.1⁻ NKT cells. Previous studies have established that thymic NK1.1⁻ NKT cells readily differentiate and become NK1.1⁺ when transferred into a recipient thymus (10, 12, 24), but peripheral NK1.1⁻ NKT cells have not been similarly tested.

To directly assay the progenitor status of peripheral NK1.1⁻ NKT cells, we employed a novel in vitro approach, where FACS-sorted, thymus-derived, or liver-derived NK1.1⁻ NKT cells were CFSE labeled and introduced to cultured embryonic thymus lobes. The thymic lobes and sorted NKT cells were cultured together overnight in hanging drops to allow donor cells to enter the thymic microenvironment, with the lobes then cultured for a further 5 days under standard fetal thymic organ cultures (FTOC) conditions to permit differentiation of donor cells. Consistent with earlier studies, NK1.1⁻ NKT cells of thymus origin readily entered and matured in the thymic lobes and most became NK1.1⁺ by 5 days after transfer (Fig. 2) (10, 12). Some NK1.1⁺ NKT cells also developed from liver-derived NKT cells, but the recovery was far lower and fewer had differentiated to the NK1.1⁺ stage. The reduced recovery was reminiscent of the reported resistance of mature mainstream T cells to re-enter the thymic microenvironment (25, 26) and is therefore consistent with liver NK1.1⁻ NKT cells being more mature than their thymic counterparts. Absolute levels of reconstitution are difficult to calculate precisely because the FTOC technique relies on an oversupply of potential donor cells, but the differences are clearly illustrated by the incorporation of nearly 10-fold more donor NKT cells in FTOCs treated with thymus NKT cells, even when 2-fold more liver NKT cells were initially introduced to the cultures (Fig. 2 and data not shown). The use of CFSE served as a convenient marker of donor-derived NKT cells (although very few endogenous NKT cells are present in embryonic thymic lobes), and also revealed that thymic NK1.1⁻ NKT cells had a higher rate of division than liver-derived NK1.1⁻ cells (Fig. 2). An earlier study (12) showed that immature NKT cells proliferate at a higher rate than mature NKT cells, and our results serve to further highlight the differences between thymus and liver NK1.1⁻ NKT cells. We used CD1d tetramer to sort NK1.1⁻ NKT cells, because we, and others, have previously found no evidence that TCR ligation by CD1d tetramer impacts on NKT cell development (Ref. 13 and data not shown). However, it is important to note that we have achieved similar results to those shown in Fig. 2 in FTOC and also following in vivo intrathymic injection, using sorted CD4⁺NK1.1⁻ (CD8/HSA-depleted) donor cells, which enriches for immature NKT cells without these being sorted using CD1d tetramer (data not shown). Although these experiments show that both thymus and liver NK1.1⁻ NKT cells include immature precursors, they suggest that thymus NK1.1⁻ NKT cells contain much higher precursor potential than those derived from the liver. This is consistent with more liver-derived NK1.1⁻ NKT cells being mature, but also supports the expected presence of a small minority of immature recent thymic emigrant NK1.1⁻ NKT cells.

In the thymus, immature NK1.1⁻ NKT cells produce higher levels of IL-4 and less IFN-γ than their mature NK1.1⁺ counterparts (10, 11, 12), but parallel studies directly comparing thymic and peripheral NK1.1⁻ NKT cells have not been performed. We first compared the cytokine production of NK1.1⁻ NKT cells from the thymus and spleen of naive mice following in vitro stimulation with PMA/ionomycin. Consistent with previous reports, fewer NK1.1⁻ than NK1.1⁺ NKT cells from the thymus produced IFN-γ (∼30 vs ∼60%), and a higher proportion produced IL-4 (∼70 vs ∼50%) (Fig. 3, A and B). In contrast, the cytokine profiles of peripheral NK1.1⁻ and NK1.1⁺ NKT cells resembled the profile of mature NK1.1⁺ thymic NKT cells. The differences between thymic and peripheral NK1.1⁻ NKT cells, and the similarities between peripheral NK1.1⁻ and NK1.1⁺ NKT cells, further strengthens the likelihood that most peripheral NK1.1⁻ NKT cells are part of a mature subset, quite distinct from their thymic counterparts.

We also directly compared NKT cells from normal and thymectomized mice following in vivo stimulation with the NKT cell agonist α-GalCer (Fig. 4, A and B). This protocol does not stimulate thymic cells (27), but consistent with the in vitro study, there was no evidence that peripheral NK1.1⁻ cells produced more IL-4 than their NK1.1⁺ counterparts. Approximately 80% of NKT cells in liver stained positive for IFN-γ and ∼30% for IL-4, with both NK1.1⁻ and NK1.1⁺ NKT cells showing similar profiles (Fig. 4 A). Slightly fewer NK1.1⁻ NKT cells in the spleen produced cytokines than NK1.1⁺ NKT cells, but this applied to both IFN-γ and IL-4 production, again quite distinct from the bias toward IL-4 seen among NK1.1⁻ NKT cells from the thymus. Attempts to directly measure the cytokine profiles of NKT cell RTE were inconclusive due to the scarcity of these cells (data not shown). It was significant to note that the cytokine profile of peripheral NK1.1⁻ NKT cells was, at best, only marginally altered in thymectomized mice compared with control mice, which supports the idea that most peripheral NK1.1⁻ NKT cells are not recent thymic emigrants. The subtle increase in the mean percentage of IFN-γ and decrease in the mean percentage of IL-4⁺ NK1.1⁻ NKT cells might reflect the small fraction of RTE that are lost from thymectomized mice.

We next examined the expression of NK cell receptors (NKRs), such as NKG2D, which are more commonly associated with NK cells, but are also expressed by most mature NKT cells. We were particularly interested in NKRs because differences in the distribution of these stimulatory and inhibitory receptors could conceivably explain how the activity of NKT cell subsets with a shared specificity for Ag might be differentially regulated. A striking difference was observed between peripheral NK1.1⁻ and NK1.1⁺ NKT cells in their expression of NKG2A/C/E, NKG2D, Ly49A, Ly49C/I, and Ly49G2 (Fig. 5 and data not shown). The expression of these NK cell receptors was almost exclusively restricted to the NK1.1⁺ compartment, suggesting a far greater propensity of these cells to be influenced by MHC family ligands. In this regard, the data was quite distinct from the cytokine analysis in that thymus and peripheral NK1.1⁻ cells were quite similar to one another. This again highlights the clear distinctions between peripheral NK1.1⁻ NKT cells and other NKT cell subsets in the thymus and periphery.

Our results strongly suggested that most peripheral NK1.1⁻ NKT cells were mature, but we could not formally exclude the possibility that some were derived from mature NK1.1⁺ NKT cells that had down-regulated NK1.1 following Ag-mediated activation (27, 28, 29). We felt this was an unlikely explanation for the high NK1.1⁻ frequency because previously activated NK1.1⁻ NKT cell are refractory to restimulation with αGalCer (20, 21), whereas the cells we tested were potent cytokine producers, comparable to the mature NK1.1⁺ NKT cells (Fig. 4). We also saw no difference in the expression of the activation marker CD69 between NK1.1⁻ and NK1.1⁺ populations (data not shown) and all mice in this study were “naive” in the sense that they are housed in an specific pathogen-free facility and were not deliberately exposed to Ags before experimentation. Nevertheless, we directly tested the possibility that mature NK1.1⁺ NKT cells were reverting to an NK1.1⁻ phenotype by adoptively transferring CFSE-labeled NK1.1⁺ NKT cells into control and thymectomised mice. This also tested whether thymectomy was somehow triggering NK1.1⁻ down-regulation. If thymectomy induced an environment in which NK1.1⁺ NKT cells became activated and were induced to down-regulate NK1.1, we would expect to observe a loss of NK1.1 expression and/or increased cell division among the transferred cells. However, 3 wk after transfer, virtually all donor NKT cells in the thymectomized hosts remained NK1.1⁺ and the CFSE dilution was consistent with previously reported levels of basal proliferation (24, 30) (Fig. 6).

The stability of the NK1.1⁻ compartment following thymectomy suggested that this subset was maintained independently of thymic output. However, at least some of the NK1.1⁻ cells must be RTE (10, 12) and we wished to determine why the undoubted loss of these cells following thymectomy did not produce a significant fall in the overall NK1.1⁻ frequency. One possibility is that RTE levels were far lower than that of mature NK1.1⁻ NKT cells, and that the loss of RTE had little impact on the overall frequency of NK1.1⁻ cells. The overall proportion of NK1.1⁻ cells could also be supported if NK1.1⁺ cells died more rapidly, although we felt this unlikely because of the overall stability of the total NKT cell numbers and proportion of the NK1.1⁻ subset. Another possibility was that peripheral NK1.1⁻ NKT cells divide more rapidly than NK1.1⁺ cells (as they do in the thymus), which would help to maintain overall NK1.1⁻ levels as recent thymic emigrants were lost. To examine the proliferative history of NK1.1⁻ and NK1.1⁺ NKT cell subsets, we administered BrdU in the drinking water for 6 days and measured its incorporation by flow cytometry to determine the extent of cellular division. As previously observed, thymic NK1.1⁻ NKT cells divided far more extensively than NK1.1⁺ thymic NKT cells (Fig. 7) (12, 31). Importantly, BrdU incorporation was higher among NK1.1⁻ peripheral NKT cells compared with NK1.1⁺ NKT cells in thymectomized and control mice, although the rate was well below that of NK1.1⁻ cells in the thymus. This is consistent with the peripheral NK1.1⁻ NKT cell population being maintained by a higher basal rate of proliferation in the absence of thymic input, and acts as another example of a functional distinction between NK1.1⁻ NKT cells in the thymus and periphery. Interestingly, both NK1.1⁻ and NK1.1⁺ NKT cells in thymectomised mice showed less division than in normal mice, indicating that thymectomy did not result in the homeostatic expansion of these cells. This difference between the thymectomized and normal mice could be a result of the removal of rapidly dividing immature NKT cells.

An important late stage in NKT cell development is the CD1d-dependent transition from NK1.1⁻ to NK1.1⁺ that can take place in either the thymus or periphery. NK1.1 expression is regarded as a sign of full maturity among NKT cells and as a corollary of this, thymic and peripheral NK1.1⁻ NKT cells are generally regarded as immature (8, 12, 15, 32, 33). Mature NK1.1⁺ NKT cells can transiently down-regulate NK1.1 after being activated (27, 28, 29), but in this study, we have demonstrated that most peripheral NK1.1⁻ NKT cells in naive mice fall into a third category, that of a stable, mature, and previously unrecognized NKT cell subset.

Previous studies showed that most NKT cells exported from the thymus are NK1.1⁻ (10, 12) and that many up-regulate NK1.1 soon after arriving in the periphery (12). The implication was that peripheral NK1.1⁻ NKT cells in normal naive mice represent immature RTE. We have revealed that this is not the case for most NK1.1⁻ NKT cells because their proportion remains virtually unchanged weeks after thymectomy. Peripheral NK1.1⁻ NKT cells also displayed a Th0-like cytokine profile (high IFN-γ and IL-4 production) similar to that of NK1.1⁺ NKT cells, and quite distinct from Th2-biased thymic NK1.1⁻ cells. Their strong cytokine response strongly suggested that peripheral NK1.1⁻ NKT cells were not simply derived from recently activated NK1.1⁺ cells that had down-regulated NK1.1, because those cells are refractory to further stimulation (20, 21). Moreover, mice in this study were not deliberately exposed to Ags and the NKT cells showed no overt signs of activation. Most importantly, transfer of marked NK1.1⁺ NKT cells showed no evidence of NK1.1 down-regulation in thymectomized mice. Although we do not exclude the possibility that some mature NK1.1⁻ NKT cells could become NK1.1⁺, the most fitting conclusion is that the majority of peripheral NK1.1⁻ NKT cells represent a unique, stable, and previously unrecognized mature NKT cell population that is quite distinct from NK1.1⁻ cells from the thymus.

It is important to state that we are not suggesting all peripheral NK1.1⁻ NKT cells are mature. Multiple studies, including our own, have shown that many NK1.1⁻ NKT cells exported from the thymus become NK1.1⁺ shortly thereafter (12, 24). On reflection, however, these studies also show a significant proportion of NK1.1⁻ cells remaining NK1.1⁻ for the full experimental period (ranging from 1 day to 6 wk). Data from these reports are consistent with our new findings, but it is nevertheless reasonable to have expected the loss of RTE caused by thymectomy to induce a fall in the overall frequency of NK1.1⁻ NKT cells as some immature RTE differentiated and became NK1.1⁺. The reason that no significant shift occurred is probably due to two main factors. Firstly, mature NK1.1⁻ NKT cells appear to greatly outnumber immature cells, so changes that only affect the immature subset (such as thymectomy) would have little impact upon the overall NK1.1⁻ compartment. Secondly, the increased rate of basal proliferation we identified among NK1.1⁻ NKT cells compared with NK1.1⁺ NKT cells would help counterbalance the loss of immature cells and maintain the overall NK1.1⁻ frequency.

Having established that a population of mature NK1.1⁻ NKT cells exists, the next important question relates to the functional significance of these cells. A very recent study has identified a population of NK1.1⁻ NKT cells in the lung and liver capable of producing high levels of IL-17 that promoted neutrophil recruitment (34), but their relevance to the broader NK1.1⁻ NKT cell pool is unclear because the study also reported that the cells were poor producers of IL-4 and IFN-γ, making them quite different to the NKT cells we studied in the thymus, spleen, and liver. We did not observe much (if any) difference in the IFN-γ and IL-4 cytokine profiles of peripheral NK1.1⁻ and NK1.1⁺ NKT cells, but this does not necessarily indicate the subsets are functionally equivalent because mouse CD4⁻ cells are far more effective at promoting anti-tumor responses than CD4⁺ NKT cells, yet the subsets have similar IFN-γ and IL-4 profiles (5, 20).

Another important aspect of the current study showed that both thymic and peripheral NK1.1⁻ NKT cells lacked expression of other NK cell receptors. The significance of these receptors has not been clearly determined for NKT cells, but there is some evidence they affect NKT cell development (35, 36) and function (37, 38). Certainly, the influence of NKRs on mainstream T cells suggests they have the potential to be important regulators of NKT cell function (39, 40, 41). The differential expression of NKRs by mature NKT cell subsets may also prove to be an important pointer to future investigations of functional heterogeneity within the NKT cell pool because it provides an explanation for how NKT cell subsets sharing the same TCR specificity might respond independently of one another in different circumstances.

As is the case for CD4⁺ and CD4⁻ NKT cells, the factors that determine whether a mature NKT cell will be NK1.1⁻ or NK1.1⁺ are unclear. It is difficult to envisage how NK1.1⁻ and NK1.1⁺ NKT cells could avoid similar developmental experiences, but it remains possible that their phenotype has been determined by distinct thymic or peripheral interactions (or a lack thereof) with CD1d-restricted Ags. Expression of NK1.1 is, at least partly, a CD1d-dependent event (34), and the mature NK1.1⁻ phenotype could potentially reflect failed, or perhaps low affinity TCR ligation postselection. Our data showing strong responses of mature NK1.1⁻ and NK1.1⁺ NKT cells to αGalCer would argue against this, but a comparison of the TCR characteristics of both subsets (for example, comparing avidity to different glycolipid Ags and/or Vb repertoires) could shed more light on the issue.

In summary, we have identified a previously unrecognized mature and stable NKT cell subset that shares similarities with NK1.1⁻ NKT cells in the thymus and NK1.1⁺ NKT cells in the periphery, but is clearly distinct from both. We support earlier findings that many recent NKT cell emigrants from the thymus are NK1.1⁻ cells destined to quickly become NK1.1⁺, but can now provide several independent lines of evidence that most peripheral NK1.1⁻ NKT cells are long-lived and already mature. Collectively, these findings have important implications for our understanding of NKT cell development, but the very clear differences in NKR expression by mature NK1.1⁻ NKT cells could also provide an avenue for exploring the basis of the apparently contradictory roles attributed to NKT cells in so many different immune settings.

We thank the Picci Brothers Foundation for their support of Department of Microbiology and Immunology Flow Cytometry Facility, David Taylor and Department of Microbiology and Immunology animal house staff for animal husbandry, and Kon Kyparissoudis for technical assistance.

The authors have no financial conflict of interest.