The natural killer T lymphocyte: a player in the complex regulation of autoimmune diabetes in non-obese diabetic mice

Abstract

Manipulation of the immune response to specifically prevent autoaggression requires an understanding of the complex interactions that occur during the pathogenesis of autoimmunity. Much attention has been paid to conventional T lymphocytes recognizing peptide antigens presented by classical major histocompatibility complex (MHC) class I and II molecules, as key players in the destructive autoreactive process. A pivotal role for different types of regulatory T lymphocytes in modulating the development of disease is also well established. Lately, CD1d-restricted natural killer T (NKT) lymphocytes have been the subject of intense investigation because of their ability to regulate a diversity of immune responses. The non-classical antigen presenting molecule CD1d presents lipids and glycolipids to this highly specialized subset of T lymphocytes found in both humans and mice. From experimental models of autoimmunity, evidence is accumulating that NKT cells can protect from disease. One of the best studied is the murine type 1 diabetes model, the non-obese diabetic (NOD) mouse. While the NKT cell population was first recognized to be deficient in NOD mice, augmenting NKT cell activity has been shown to suppress the development of autoimmune disease in this strain. The mechanism by which CD1d-restricted T cells exert this function is still described incompletely, but investigations in NOD mice are starting to unravel specific effects of NKT cell regulation. This review focuses on the role of CD1d-restricted NKT cells in the control of autoimmune diabetes.

The progressive development of autoimmune diabetes in the non-obese diabetic (NOD) mouse

Type 1 diabetes (T1D) is an autoimmune disease, where lymphocytes destroy the insulin producing β-cells of the pancreatic islets of Langerhans. The development of T1D is influenced by multiple genetic and environmental factors of which most remain unknown. The non-obese diabetic (NOD) mouse strain [1] spontaneously develops a disease very similar to human T1D, and constitutes a major animal model used for investigating the cause of autoimmune diabetes [2,3]. Disease can be transferred by T lymphocytes, demonstrating that these cells are central to T1D development, but other immune cells such as B cells, dendritic cells, macrophages and possibly natural killer (NK) cells are also known to play an important role in diabetes pathogenesis. The autoimmune activities leading to disease progress over a relatively long period of time, and have been described to transit discrete checkpoints [4] (Fig. 1). As early as around 2 weeks after birth, the pancreas experiences a wave of β-cell death (Fig. 1a), also occurring in non-diabetic mouse strains, leading to the presentation of pancreas-derived antigens in the draining pancreatic lymph nodes [5] (Fig. 1b). In the NOD mouse, this leads to the stimulation of autoreactive T lymphocytes. The insufficiently controlled initiation of T cell autoreactivity, specific for the NOD strain, is regarded as a failure at the first checkpoint. Soon after, at 3–4 weeks of age, leucocytes enter into the pancreas of NOD mice (Fig. 1c). As insulitis proceeds, cells continue to accumulate around the islets, initially as a peri-islet infiltrate (Fig. 1d). The transformation from this benign state to invasive infiltration and β-cell destruction is seen as passage through the second checkpoint (Fig. 1e). When more than 90% of the islets have been destroyed and insulin production is too low to regulate blood glucose levels, diabetes precipitates occurring from around 12 weeks of age (Fig. 1f). At 25–30 weeks of age, around 80% of female NOD mice have been affected (Fig. 1g). Insulitis develops in both female and male NOD mice; however, only 20–30% of the male mice become diabetic, suggesting that the second checkpoint is more effective in males.

Fig. 1.

Development of autoimmune diabetes in the non-obese diabetic (NOD) mouse. At around 2 weeks of age a wave of apoptosis occurs among the pancreatic β-cells of the islets of Langerhans, as a part of natural pancreatic development (a). This leads to the presentation of β-cell derived antigens on dendritic cells in the pancreatic lymph nodes and activation of β-cell specific T lymphocytes (b). From 3 to 4 weeks of age lymphoid cells start to infiltrate the pancreas (c), initially as a non-invasive peri-islet infiltrate (d). With time the benign infiltrate develops into an aggressive state, in which T lymphocytes enter into the islet and destroy β-cells (e). When less than 10% of the β-cell mass remains, starting to occur at 10–12 weeks of age (f), blood glucose levels can no longer be maintained and the mice become diabetic. At 25–30 weeks of age around 80% of female NOD mice have developed disease (g), while only 20–30% of the male mice are affected.

The important role of different types of regulatory T lymphocytes in the control of autoimmune disease is well established, including in T1D of NOD mice [6]. Multiple cellular subsets have been ascribed the ability to inhibit diabetes development. The demonstration of a regulatory role for CD4⁺ T cells in NOD pathogenesis [7,8] was followed by reports on additional cell types with the capacity to inhibit diabetes development, including T cell receptor (TCR) γδ T cells [9], CD25⁺CD4⁺ T cells [10] and CD62L⁺ CD4⁺ T cells [11,12]. More recently, CD1d-restricted natural killer T (NKT) lymphocytes have been implied in the regulation of autoimmune disease in humans and mice [13].

CD1d-restricted natural killer T cell populations in humans and mice

CD1d-restricted natural killer T cells

NKT cells are a subset of TCR αβ cells, of which most recognize lipid antigens presented on the major histocompatibility complex (MHC) class I-like molecule CD1d [14–17] (Fig. 2). NKT cells have been classified as innate-like lymphocytes, which share features of both innate and adaptive immune cells [18]. They have an activated phenotype, are generally CD4⁺ or CD4^–CD8^–, and display some level of autoreactivity to CD1d [19,20]. When activated, NKT cells promptly secrete large amounts of cytokines [21], including interleukin (IL)-4 and interferon (IFN)-γ. NKT cells can regulate the immune response in diverse situations, such as autoimmune diseases in humans and mice [22], tumour rejection [23] and different types of infections [24]. NKT cells are abundant among T lymphocytes in the liver and bone marrow, are also found in thymus and spleen, and more frequent in pancreatic and mesenteric than other lymph nodes [25].

Fig. 2.

Subsets of CD1d-restricted T lymphocytes in the mouse. Natural killer T (NKT) cells were initially defined as T lymphocytes expressing T cell receptor (TCR)αβ together with NK1·1. NK1·1⁺ TCRαβ⁺ T cells are depicted by the yellow square. Recently the term NKT cell is often used synonymously with ‘CD1d-restricted T cells’. CD1d-restricted T cells are represented on the left of the hatched black/grey dividing line. Among NK1·1⁺ TCRαβ⁺ T cells, the majority are CD1d-restricted (those in the yellow box that fall to the left of the hatched black/grey dividing line). However, a subset of NK1·1⁺ TCRαβ⁺ T cells is independent of CD1d (to the right of the dividing bar), and is therefore often called ‘NKT-like’. CD1d-restricted T cells can be subdivided on the basis of the TCR they use. Those expressing the invariant Vα14-Jα18 TCR α-chain rearrangement (the red rectangle) are often called iNKT cells or classical NKT cells. The blue rectangle represents CD1d-restricted T cells which express other, diverse TCR rearrangements. These cells are also termed non-classical NKT cells. Part of the areas in the blue and red rectangles fall outside the yellow square, symbolizing that not all CD1d-restricted T cells express the NK1·1 marker. The nomenclature in this figure follows the recent proposal by Godfrey et al. [16].

The cells were initially termed NKT cells because they co-express the TCRαβ and surface markers found normally on natural killer cells, such as the NK1·1 receptor (NKR-P1c or CD161c). NK1·1 is expressed by some mouse strains such as the C57BL/6 [26]. In strains which do not have NK1·1, such as the NOD mouse, other less specific markers have been used to define the population; alternatively, the cells have been studied in congenic mice expressing the NK1·1 antigen. The use of the term NKT cell is somewhat confusing and not always consistent, prompting a recent attempt to unify the terminology [16]. As illustrated in Fig. 2, most (but not all) TCRαβ⁺ NK1·1⁺ cells in the mouse are CD1d-restricted. Based on the TCR used, CD1d-restricted T cells, in turn, can be divided into the subgroups of classical and non-classical NKT cells (described further below). However, even in mouse strains which express the marker, not all CD1d-restricted T cells will display NK1·1 on the surface. Immature CD1d-restricted T cells are NK1·1^– and, moreover, activation of mature NKT cells can result in the down-modulation of NK1·1. In the current literature, the term NKT cells usually refers to CD1d-restricted T cells regardless of NK1·1 expression. CD1d-independent NK1·1⁺ TCRαβ⁺ T cells may be called ‘NKT-like’[16] (Fig. 2). Following this nomenclature, ‘NKT cells’ will denote CD1d-restricted T cells in the text below.

Classical NKT cells

CD1d-restricted T cells make up a heterogeneous and functionally divergent population. A large proportion of the cells in the mouse display a TCR consisting of an invariant Vα14-Jα18 TCRα-chain, combined with diverse TCRβ-chains using a limited number of Vβ regions [14,20]. This subset is often referred to as invariant NKT cells (iNKT cells) or classical NKT cells (Fig. 2). Cells with this TCR can be activated by the abundantly used synthetic ligand α-galactosylceramide (αGalCer) presented on CD1d [27]. A lysosomal glycosphingolipid was identified recently as an endogenous ligand for iNKT cells [28]. A high evolutionary conservation of the CD1d system is suggested by the presence of a population of human T lymphocytes with an invariant TCR using theVα24-JαQ segments homologous to the murine Vα14-Jα18 regions [29,30]. The human cells have the same antigen specificity and CD1d-restriction as the murine cells.

Non-classical NKT cells in mice and humans

The murine CD1d-restricted cells not using the Vα14-invariant TCR [19,31–33] appear to share the main characteristic features of NKT cells, such as the expression of NK markers, rapid cytokine secretion upon activation and a memory surface phenotype [33–36]. They are thought to utilize a relatively diverse TCR repertoire; however, this subset may also contain sets of cells with invariant receptors different from the Vα14 type [37]. CD1d-restricted cells which do not use the Vα14 invariant TCR are referred to as non-classical NKT cells (Fig. 2). Non-classical NKT cells fail to be activated by αGalCer [38,39], but recently other CD1d-restricted ligands have been identified for this group of NKT cells [40]. CD1d-restricted T cells with diverse TCR may be more frequent than the Vα24 iNKT subset in humans [41–43], emphasizing the need for further analysis of this subset. Studies of the non-classical NKT cell population in both mice and humans have been hampered by the lack of reliable markers expressed by these cells. While αGalCer loaded CD1d-tetramers can be used to study both human and murine iNKT cell populations [44,45], this method does not allow a general identification of non-classical NKT cells, as the majority of their ligands are still unidentified.

Functionally distinct NKT cell subsets in mice and humans

In mice, the iNKT cells display CD69 on the surface and are efficient producers of both IL-4 and IFN-γ upon stimulation. In contrast, cells within the non-classical NKT subset produced high levels of IFN-γ, low levels of IL-4 and were CD69^– CD49b^high [35,36]. These distinct TCR repertoires and functional abilities of NKT cell subsets imply that they can play selective roles in immune responses. In addition, the profile of cytokines released by NKT cells can be modulated by the means of stimulation and the cytokine milieu [46].

Among human CD1d-restricted T cells a similar functional heterogeneity has been defined, in this case within the population of Vα24 iNKT cells [47,48]. The expression of NK cell-associated markers was associated primarily with the CD4^–CD8^– Vα24 iNKT subset. Both CD4⁺ and CD4^–CD8^– Vα24 iNKT cells were capable of Th1 cytokine production, including IFN-γ and TNF-α, while Th2 cytokines such as IL-4 and IL-13 were secreted exclusively by the CD4⁺ subset.

NKT cell activation

The autoreactive nature of NKT cells prompted a search for endogenous lipids which might serve as ligands for CD1d-restricted cells. This resulted in the identification of cellular lipids with the ability to activate NKT cells in a CD1d-restricted manner [38,49–51]. Recently, studies using infection models have shown that natural activation of NKT cells can occur through the presentation of both endogenous and exogenous ligands on CD1d [51–54]. In contrast, the nature of CD1d-bound lipids that mediate NKT cell activation during autoimmunity is less well explored. Sulfatide has been proposed to be a ligand involved in the stimulation of nonclassical NKT cells in experimental autoimmune excephalomyelitis [40], while ligands which activate NKT cells during natural regulation of T1D have not been identified. Experiments using αGalCer suggest that dendritic cells are critical for efficient stimulation of NKT cells in vivo, although B cells could mediate αGalCer-induced production of IL-4 rather than IFN-γ [55,56]. Interestingly, during some infections iNKT cells produce high amounts IFN-γ, but no detectable IL-4 [54,57]. The pattern of cytokines induced may be influenced by factors such as IL-12, which may promote the secretion of IFN-γ [58,59], and it has been shown that structural variants of αGalCer can induce qualitatively distinct responses such as skewed cytokine profiles [60,61].

To increase our understanding of the regulatory abilities of different NKT cell subsets in autoimmunity, and to explore the potential to modulate these cells to prevent or suppress disease, it will be important to investigate in detail the mechanism(s) behind NKT cell regulation of autoimmunity. To this end, the diabetic NOD mouse has provided a useful model system.

Association between CD1d expression, NKT cell regulatory function and autoimmune diabetes development

Defects in CD1d expression and CD1d-restricted T cells in autoimmune diabetes

NOD mice, as well as autoimmune-prone SJL mice, have a numerically reduced NKT cell population. In addition, the remaining NKT cells show diminished production of cytokines such as IL-4 and IFN-γ [62–65]. To investigate a possible link with disease, several laboratories have identified genetic regions which influence the size and functional ability of the NKT cell population. NKT cell numbers were found to be controlled by a locus on chromosome 2 which may contribute to diabetes susceptibility, and a locus on chromosome 1 which appears to contribute to lupus susceptibility (reviewed in [66]). These results lend support to the notion that this cellular subset may have important regulatory functions in T1D.

Given that the NKT cell population of NOD mice is numerically and functionally diminished, genetic ablation of CD1d expression might not be expected to have dramatic effects in this strain. Despite this, three studies show that NOD mice lacking CD1d display an accelerated onset and increased incidence of disease [67–69], while one report failed to demonstrate such a difference [70].

Studies addressing a correlation between NKT cell levels in peripheral blood and T1D in humans have yielded variable results [71–75]. However, the studies of human peripheral blood lymphocytes (PBL) should be interpreted with caution, as the deficiency in the iNKT cell population found in lymphoid organs of NOD mice was not reflected among peripheral blood lymphocytes in the same mice [76]. Therefore, further investigation should be awaited before conclusions can be drawn about a similar deficiency in diabetes patients.

Deficiencies in antigen-presenting cells may add to the reduced functional ability displayed by CD1d-dependent NKT cells in NOD mice. CD1d is highly expressed on murine CD8α⁺ dendritic cells (DC) [68,77], a DC subset which is suggested to play a role in tolerance induction [78]. NOD mice display abnormalities associated with DC, including a selective reduction in the CD8α⁺ DC population [79,80]. In addition, bone marrow-derived DC from NOD mice express lower levels of CD1d than the same cells from C57BL/6 mice [81]. Thus, a poor immunoregulation by NKT cells in NOD mice may result both from deficiencies in the NKT cell population, as well as inadequate CD1d-restricted antigen presentation on DC.

Prevention of diabetes by increasing NKT cells numbers or NKT cell activation

Different strategies have been used to show that improved NKT cell activity reduces autoimmune diabetes in NOD mice. In early experiments, adoptive transfer of CD4^–CD8^– TCRαβ⁺ thymocytes, a population enriched for CD1d-restricted T cells, into young NOD mice was found to prevent the development of diabetes in recipient mice [82]. Further, diabetes development was suppressed in TCR transgenic NOD mice with an amplified iNKT cell population, and transgenic iNKT cells could prevent disease induction by spleen cells from diabetic NOD mice in adoptive co-transfer experiments [83]. Subsequently, it has also been demonstrated that TCR transgenic NKT cells using a non-Vα14 TCR could prevent disease using a similar experimental approach [36]. This suggests that NKT cells of both the classical and non-classical subset can modulate autoimmune diabetes. It remains to be investigated whether the two NKT cell types, which differ in their functional abilities [35,36], regulate diabetes through the same mechanism. It is notable that both subsets of NKT cells have also been suggested to regulate experimental autoimmune encephalomyelitis [40,84–87].

Employing a different approach, investigators have treated NOD mice with the iNKT cell ligand αGalCer and found a delayed onset and reduced incidence of diabetes development [68–70,88]. Importantly, disease was reduced even when αGalCer treatment was initiated after the onset of invasive insulitis.

Proposed mechanisms of CD1d-restricted T cells in the regulation of diabetes pathogenesis in NOD mice

Where and when do NKT cells regulate the pathogenic process in diabetes development?

While the suppressive influence of CD1d-restricted T cells on T1D in NOD mice is well established, the mechanism of disease regulation, in this as well as in other autoimmune disease models, is less well understood. The means whereby NKT cells can modulate the development of autoaggression is starting to be unravelled from a series of studies in the NOD model. These have addressed both the ‘natural regulatory function’ of NKT cells as well as αGalCer induced regulation. Both experimental approaches will be important to increase our understanding of immunomodulation by NKT cells, with the aim of developing treatments for autoimmune diseases. The results indicate that NKT cells may interfere with diabetes development by more than one mechanism, and at multiple stages of diabetes pathogenesis as outlined below and in Fig. 3.

Fig. 3.

Possible influence of CD1d expression and natural killer T (NKT) cells during the autoimmune reactions preceding diabetes in non-obese diabetic (NOD) mice. (a) At around 2 weeks of age β-cell antigens from apoptotic cells are acquired by antigen-presenting cells and transported from the pancreas to the pancreatic lymph nodes [5] (b). This leads to the activation of autoreactive T cells at this site (c). Bone marrow derived dendritic cells (DC) from NOD mice have reduced levels of CD1d [81], suggesting that inappropriate display of CD1d on DC in NOD mice may contribute to reduced NKT cell activation. Administration of α-galactocylceramide (αGalCer) results in the accumulation of iNKT cells and tolerogenic DC selectively in the pancreatic lymph nodes [68,90], which may create an environment which prevents the development of autoaggressive T cells (c). NKT cells could interfere with the activation and expansion of autoreactive T cells in a cytokine or cell-to-cell contact-dependent manner [95], directly or by modulating DC function. (d) Transgenic over expression of CD1d in the islets reduces diabetes incidence [81]. NKT cells are present in the pancreatic infiltrates in young female NOD mice, and also in male NOD mice at late times [68,91]. Within the pancreas NKT cells may interfere with the expression of β-cell destructive behaviour of autoreactive T cells, or directly kill autoaggressive T cells. (e) The proportion of NKT cells is reduced in the late invasive infiltrates in the pancreas of prediabetic female NOD mice.

A role for dendritic cells

DC are instrumental for tolerance induction in T lymphocytes [78], and it is known that CD1-restricted T cells can modulate DC differentiation [89]; thus it is reasonable to envisage that NKT cells could act to suppress autoimmunity via DC. Administration of αGalCer lead to an accumulation of iNKT cells and myeloid DC in the pancreatic lymph nodes of NOD mice [68,90]. Injection of myeloid DC isolated from pancreatic but not inguinal lymph nodes protected female NOD mice from diabetes [68]. The authors suggested that iNKT cell mediated suppression of T1D in NOD mice induced by αGalCer treatment was achieved through the recruitment of tolerogenic myeloid dendritic cells (DC) to the pancreatic lymph nodes [68,90]. This, in turn, would result in an enhanced tolerance of autoreactive T cells at this site (Fig. 3b,c).

TH1/TH2 cytokines in diabetes regulation by NKT cells

Early experiments indicated that regulation of disease by NKT cells was associated with a TH1 to TH2 shift. IL-4 release from NOD NKT cells was shown to be diminished, and a crucial role was assigned to IL-4 and/or IL-10 in the protection from diabetes afforded by transferred CD4^–CD8^– TCRαβ⁺ thymocytes [82]. TCR transgenic NOD mice with elevated numbers of iNKT cells were resistant to cyclophosphamide induced disease; however, simultaneous administration of blocking antibodies to IL-4 abrogated protection [91]. Moreover, in normal female NOD mice, αGalCer treatment could reduce cyclophosphamid- induced diabetes in an IL-4-dependent manner [92]. These data are in agreement with the previously described protective role of IL-4 administration in the NOD model [6]. The role for IL-10 in αGalCer-mediated protection against cyclophosphamide accelerated disease, however, is less clear [88,92]. Taken together, the suppression of autoimmunity exerted by NKT cells clearly depends on IL-4 in some settings, while the role of IL-10 requires further investigation.

In addition to the deficiency in IL-4 production, NOD NKT cells have a reduced ability to secrete IFN-γ[63]. This NKT cell defect was suggested to contribute to diabetes susceptibility, which would be compatible with the protective effect of IFN-γ on diabetes development in NOD mice [93].

Prevention of the development of autoreactive effector T cells

The effects of NKT cell regulation have been analysed in experiments using monoclonal TCR transgenic populations of autoreactive diabetogenic T cells. Lehuen and co-workers have demonstrated that TCR transgenic MHC class II-restricted diabetogenic T cells from BDC2·5 NOD mice were unable to induce disease in TCR transgenic mice with elevated iNKT cell numbers [94]. BDC2·5 T cells underwent initial activation and proliferation in pancreatic lymph nodes, but did not differentiate into IFN-γ secreting cells (Fig. 3c). There was no shift of their cytokine production towards IL-4; however, the BDC2·5 T cells appeared rather to be rendered anergic in the presence of transgenic iNKT cells [94]. The regulation of BDC2·5 T cells in vivo was also achieved when IL-4, IL-10, IL-13 or TGF-β were absent or inhibited [95]. In vitro, the regulation of BDC2·5 cells by iNKT cells required cell-to-cell contacts, and a minor role was ascribed to IL-4 [95]. Using a similar approach, it was shown that diabetogenic transgenic CD8⁺ AI4 T cells were unable to transfer disease to sublethally irradiated NOD mice in which iNKT cells had been activated with αGalCer [90]. αGalCer stimulation of iNKT cells augmented apoptosis and anergization of the transgenic autoreactive CD8⁺ T cells. Thus, in these specific situations, iNKT cells prevented the development of autoaggressive effector cells without a concomitant skewing of the response to a Th2 type.

Regulation of autoaggression by NKT cells within the pancreas?

NKT cells could also interfere with destructive autoaggression in situ in the pancreas (Fig. 3d), for example by direct killing of effector cells, or by suppressing their effector functions. Blood vessel endothelial cells in the pancreas express CD1d [81,96], suggesting that NKT cells could enter the tissue upon ligand recognition on vessel walls. iNKT cells can indeed be found in the pancreatic infiltrates of NOD mice [68,91] (Fig. 3d). The proportion of iNKT cells among infiltrating T cells decreased at the time when peri-insulitis was converted to invasive insulitis and diabetes in female mice (Fig. 3e). In contrast, male NOD mice, which have a reduced incidence of disease, had higher frequencies of iNKT cells in the islets than female mice, consistent with a putative protective effect by these cells in males [68]. Further, transgenic over-expression of CD1d in the pancreatic islets reduced diabetes incidence in female mice, and augmented the αGalCer-induced cytokine production by pancreatic lymph node cells, suggesting that iNKT cell interaction with CD1d in the pancreas can support diabetes regulation [81].

Concluding remarks

Studies of the mechanisms behind the modulation of autoimmune diabetes by CD1d-restricted NKT cells suggest that NKT cells can interfere in the pathogenic process preceding disease in more than one way. Distinct requirements for the regulatory effect, such as the variable dependence of IL-4, found in the different reports are likely to depend on factors such as the stage of disease investigated, the activation state of the autoreactive T cell and whether NKT cells were stimulated by exogenous ligands. Further research is required to define precisely which effector functions are mediating disease regulation during natural and induced regulation. The NOD model is likely to continue to provide crucial information on these issues.

CD1d-restricted NKT cells have characteristics which make them appear as attractive targets for prevention or treatment of autoimmune disease. One is the relatively non-polymorphic nature of the CD1d-molecule, allowing the use of generic ligands for CD1d-presentation to activate NKT cells. Secondly, presence within the NKT cell population of large ‘clones’ reactive to the same ligands, and the presence of the same clones in different individuals, suggest that targeting of such clones could lead to efficient immune responses in a majority of cases. Recent reviews have discussed the potential of αGalCer treatment as a therapy for autoimune diseases [97,98]. To elucidate the full potential of therapy through NKT cell modulation, further information is needed on the ligands recognized by both classical and non-classical NKT cells, the distinct roles of the different NKT cell subsets in immunity and the activation requirements for induction of selective effector functions.