Department of Microbiology and Immunology, Vanderbilt University School of Medicine, Nashville, TN, USA
L. Van Kaer, PhD Department of Microbiology and Immunology Vanderbilt University School of Medicine Medical Center North Room A-5301 1161 21st Avenue South Nashville TN 37232-2363 USA Tel: +1 615 343 2707 Fax: +1 615 343 2972 e-mail: email@example.com
L. Van Kaer, PhD Department of Microbiology and Immunology Vanderbilt University School of Medicine Medical Center North Room A-5301 1161 21st Avenue South Nashville TN 37232-2363 USA Tel: +1 615 343 2707 Fax: +1 615 343 2972 e-mail: firstname.lastname@example.org
Invariant natural killer T (iNKT) cells are a subset of T lymphocytes that react with glycolipid antigens presented by the major histocompatibility complex class I-related glycoprotein CD1d. Although iNKT cells express an antigen-specific receptor of the adaptive immune system, they behave more like cells of the innate immune system. A hallmark of iNKT cells is their capacity to produce copious amounts of immunoregulatory cytokines quickly after activation. The cytokines produced by iNKT cells can influence the level of activation of many cell types of the innate and adaptive immune systems as well as the quality of an adaptive immune response. As such, iNKT cells have emerged as important regulators of immune responses, playing a role in microbial immunity, autoimmunity, tumor immunity, and a variety of inflammatory conditions. Although several endogenous and exogenous glycolipid antigens of iNKT cells have been identified, how these glycolipids orchestrate iNKT-cell functions remains poorly understood. Nevertheless, iNKT cells hold substantial promise as targets for development of vaccine adjuvants and immunotherapies. These properties of iNKT cells have been investigated most extensively in mouse models of human disease using the marine sponge-derived agent α-galactosylceramide (α-GalCer) and related iNKT-cell antigens. While these preclinical studies have raised enthusiasm for developing iNKT-cell-based immunotherapies, they also showed potential health risks associated with iNKT cell activation. Although α-GalCer treatment in humans was shown to be safe in the short term, further studies are needed to develop safe and effective iNKT-cell-based therapies.
Cells of the innate immune system can sense microbial pathogens through germ line-encoded receptors, called pattern recognition receptors, which interact with conserved pathogen-associated molecular patterns. In turn, the cytokines and costimulatory molecules elaborated by cells of the innate immune system are critically important for initiating an effective adaptive immune response. B and T cells of the adaptive immune system recognize microbial pathogens through diverse, antigen-specific receptors that are generated by somatic DNA rearrangement. Invariant natural killer T (iNKT) cells are a distinct subset of T lymphocytes that express a receptor of the adaptive immune system, yet their specificity and behavior are more reminiscent of cells of the innate immune system. Here, we will review the salient characteristics and functions of these fascinating cells and discuss their potential applications in the development of vaccine adjuvants and immunotherapies.
iNKT cells and their cousins
Nearly two decades ago, it was reported that a small subset of murine T lymphocytes expressed NK1.1 (also called CD161), a marker that was previously considered to be uniquely expressed by natural killer (NK) cells [reviewed in (1)]. This finding, together with parallel studies from other investigators characterizing a T-cell subset expressing intermediate T-cell receptor (TCR) surface levels with a repertoire skewed toward Vβ8 and a subset of CD4−CD8− (double-negative) T cells producing immunoregulatory cytokines, indicated the existence of a unique subpopulation of T cells defined by NK1.1 expression. Subsequent studies showed that the majority of these cells were reactive with the major histocompatibility complex (MHC) class I-related protein CD1d, and these cells were called natural killer T (NKT) cells to reflect their hybrid surface expression phenotype of conventional T cells and NK cells. However, it soon became clear that not all NK1.1-expressing cells recognized CD1d and that some CD1d-restricted T cells failed to express the NK1.1 marker. The need for a redefinition of these cells was further compounded by the realization that the NK1.1 marker (recognized by the PK136 antibody) is only expressed in mouse strains such as C57BL/6 but not in many other commonly used mouse strains (although these strains express allelic variants of NK1.1 that are not recognized by the PK136 antibody). These lymphocytes have now been further categorized into three distinct subsets (1):
(a) Type I NKT cells, also called iNKT cells or conventional NKT cells, are CD1d-restricted T cells that express a semi-invariant TCR formed by the combination of Vα14–Jα18 and Vβ8.2/Vβ7/Vβ2 chains in mice and homologous Vα24–Jα18 and Vβ11 chains in humans. The majority of these cells express NK1.1 (CD161 in humans) (Figure 1A).
(b) Type II NKT cells, also called nonclassical NKT cells, are also CD1d restricted but have a more diverse TCR repertoire, and many of these cells express NK1.1.
(c) NKT-like cells consist of a diverse range of cell types, including CD1d-independent, NK1.1-expressing T cells and multiple other T-cell subsets that express semi-invariant TCRs. To avoid confusion with type I and type II NKT cells, and because these cells are CD1d independent, it has been recommended that these cell types should not be called NKT cells (1).
This review paper focuses on type I NKT cells, which we will call iNKT cells. A key issue in studying iNKT cells is to reliably identify these cells, which can be accomplished in mice and humans by using multimeric forms of CD1d (most commonly CD1d tetramers) loaded with the prototypic iNKT-cell antigen α-galactosylceramide (α-GalCer, see below). Functional studies of iNKT cells in vivo can be performed most reliably with TCR gene Jα18-deficient mice. Analysis of CD1d-deficient mice, together with Jα18-deficient mice, can be used to distinguish between type I and type II NKT-cell functions.
iNKT cells are found with the highest frequency in the liver and the bone marrow of mice and at significant numbers in the thymus, spleen, and peripheral blood [for reviews regarding iNKT cells, see (2–5)]. In humans, the frequency of NKT cells is usually much lower and with a high degree of variability between individuals. Most murine iNKT cells express the coreceptor CD4, and the remaining cells express neither CD4 nor CD8, whereas a significant proportion of iNKT cells in humans express CD8α. In addition to NK1.1 (or CD161), iNKT cells express a variety of other receptors that are characteristic of the NK-cell lineage, including NKG2D, CD94 and members of the Ly49 family of NK-cell receptors. iNKT cells also constitutively express a variety of markers such as CD44, CD69 and CD122 that are typical of activated or memory T cells.
Because of its limited diversity, the TCR of iNKT cells more closely resembles the pattern recognition receptors expressed by cells of the innate immune system than the diverse antigen-specific receptors expressed by cells of the adaptive immune system. Because of their propensity to react with autologous cells, it has been generally assumed that iNKT cells recognize both endogenous and exogenous glycolipid antigens. All iNKT cells from mice and humans react with α-GalCer (6) (Figure 1B), a glycosphingolipid that was isolated from the marine sponge Agelas mauritianus during a screen of natural products with antimetastatic activities in mice. α-GalCer is a very potent iNKT-cell agonist and has been extensively studied with regard to its interaction with CD1d and the invariant TCR of iNKT cells, its immunomodulatory activities, and its therapeutic properties. The trimolecular complex formed between α-GalCer, CD1d and the invariant TCR has been investigated at the atomic level and has shown an unusual and unexpected recognition mode (7, 8). Numerous structural variants of α-GalCer have been synthesized and analyzed (see below and Figure 1B). Several of the known naturally occurring exogenous iNKT-cell ligands are pathogen-derived glycolipid antigens, such as glycosylceramides from the cell wall of gram-negative Novosphingobium (previously called Sphingomonas) species (9–11) and diacylglycerols from Lyme disease-causing Borrelia burgdorferi bacteria (12) (Figure 1B). Interestingly, because the Novosphingobium glycolipids are structurally similar to α-GalCer, and because marine sponges are commonly colonized by bacteria, including Novosphingobium species, it is likely that α-GalCer was isolated from Novosphingobium bacteria that colonized the marine sponge. Several studies have also shown reactivity of iNKT cells with select cellular, tumor-derived, and other microbial glycolipids, but these reagents were only recognized by small subsets of iNKT cells. An endogenous, lysosomal glycosphingolipid, isoglobotrihexosylceramide (iGb3, Figure 1B) is capable of activating murine and human iNKT cells (10), albeit at lower levels than α-GalCer or the bacterial glycolipids, as would be expected for an endogenous, physiological ligand. The role of iGb3 in the development and function of iNKT cells is still under investigation.
Presentation of glycolipid antigens to iNKT cells
CD1d molecules are expressed by professional antigen-presenting cells (APCs), including dendritic cells (DCs), macrophages, and B cells (CD1d is particularly abundant on marginal zone B cells) as well as by double-positive thymocytes and hepatocytes. Soon after synthesis in the endoplasmic reticulum (ER), CD1d molecules bind with endogenous phospholipids that facilitate the exit of CD1d from the ER and its transport to the cell surface (13–15) (Figure 2). Current evidence indicates that the loading of phospholipids onto CD1d in the ER involves the action of microsomal triglyceride transfer protein, which is a lipid transfer protein (LTP) that plays an important role in the assembly of apolipoprotein (apo) B with lipids in the ER. Surface CD1d is then internalized by clathrin-mediated endocytosis that depends on a tyrosine-based motif in the cytoplasmic tail of CD1d. In addition to this pathway, it has been reported that a subset of CD1d molecules reaches endosomal compartments after association with the MHC class II-associated invariant chain in the ER. Within late endosomes, the ER-loaded phospholipids are exchanged for antigenic lipids that can activate iNKT cells. These glycolipids can be derived from endogenous sources, such as lysosomal iGb3, or from exogenous sources. Multiple pathways might be engaged to deliver exogenous glycolipids to the CD1d-loading compartment: (a) phagocytosis of microbes or their products, (b) engagement of mannose within glycolipids by mannose receptors, (c) engagement of modified low-density lipoprotein (LDL) by scavenger receptors, and (d) internalization of apoE-containing lipid complexes through the LDL receptor or related lipoprotein receptors. After their arrival in late endosomal compartments, endogenous or exogenous glycolipids might undergo processing (e.g. by carbohydrate hydrolases), followed by loading onto CD1d with the assistance of a variety of LTPs, including saposins, GM2 activator and Niemann–Pick C2 protein. These antigenic CD1d–glycolipid complexes are then displayed at the cell surface for recognition by iNKT cells.
In recent years, it has become clear that multiple viral pathogens, including Kaposi sarcoma-associated herpes virus, HIV-1, herpes simplex virus-1, vesicular stomatitis virus, and vaccinia virus, can interfere with CD1d trafficking and, thus, impede glycolipid antigen presentation to iNKT cells (16). Furthermore, Chlamydia trachomatis downregulates CD1d expression on human epithelial cells by targeting its degradation by both cellular and chlamydial proteasomes (17).
Like conventional T cells, iNKT cells develop in the thymus (18). iNKT cells develop relatively late during T-cell ontogeny, owing to structural constraints in the recombination of TCR Vα14 and Jα18 gene segments. Generation of the canonical Vα14–Jα18 rearrangement is a random event. After expression of surface TCRs at the double-positive thymocyte stage, iNKT cells diverge from conventional T cells. iNKT cells require positive selection by CD1d–glycolipid complexes in the thymus. However, in sharp contrast with the positive selection of conventional T cells, which is dependent on cortical thymic epithelial cells, iNKT cells are selected by CD1d-expressing, double-positive thymocytes. This selection likely involves endogenous iNKT-cell antigens, and iGb3 has been suggested to play such a role (10). This selection process also involves homotypic interactions between members of the signaling lymphocyte activation molecule (SLAM) family of receptors (19). These receptors signal through the adaptor protein SLAM-associated protein, the Src kinase Fyn, and the downstream transcription factor nuclear factor (NF)-κB (20). The selected iNKT cells subsequently expand and undergo multiple maturation steps that ultimately result in the characteristic surface phenotype and innate-like effector functions of iNKT cells. These maturation steps involve a variety of signal transducers, transcription factors, and costimulatory molecules. Importantly, recent studies have shown a critical role of the transcription factor promyelocytic leukemia zinc finger (PLZF) in directing the innate-like effector differentiation of iNKT cells during thymic development (21, 22). Furthermore, the licensing of mature iNKT cells for cytokine production requires granulocyte monocyte colony-stimulating factor (GM-CSF) signaling (23). After acquisition of their innate-like phenotype, iNKT cells egress from the thymus in a mechanism that involves lymphotoxin-β receptor signaling. Nevertheless, in mouse, a significant proportion of iNKT cells remain in the thymus, where they become long-lived residents. Recruitment of iNKT cells to peripheral tissues requires expression of the sphingosine 1-phosphate 1 receptor, and entry into the liver requires expression of the chemokine receptor CXCR6.
Pathways of iNKT-cell activation
In most immune responses where iNKT cells have been implicated, the mechanism of iNKT-cell activation remains poorly understood. Nevertheless, studies with microbial pathogens have shown two general pathways of iNKT-cell activation (14, 24) (Figure 3). In the direct pathway of iNKT-cell activation (Figure 1A), which is engaged by Novosphingobium species and B.burgdorferi, cognate glycolipids bind CD1d and activate iNKT cells directly. However, iNKT cells also become activated to many microbes that lack cognate iNKT-cell antigens (25). iNKT-cell activation during infection with such microorganisms is driven by cytokines. In this indirect pathway of iNKT-cell activation (Figure 3B), engagement of signaling pattern recognition receptors, most notably toll-like receptors, with microbial products leads to the production of cytokines such as interleukin (IL)-12, IL-18, and/or type I interferons (IFNs) that can activate iNKT cells. The latter mechanism has been implicated in the activation of iNKT cells by several bacterial and viral microorganisms and their products. For some microorganisms, iNKT-cell activation also requires CD1d expression on the APCs, in which case cytokines might function to amplify low-affinity interactions of the iNKT-cell receptor with complexes between CD1d and endogenous glycolipids such as iGb3. Similar mechanisms might be involved in the activation of iNKT cells during acute and chronic inflammatory conditions.
A hallmark of iNKT cells is to rapidly produce copious amounts of cytokines quickly after TCR engagement (2–5). Interestingly, however, iNKT cells are capable of producing a wide array of cytokines (Figure 1A), including IL-2, IL-4, IL-6, IL-10, IL-13, transforming growth factor (TGF)-β, IFN-γ, and tumor necrosis factor (TNF)-α, and some iNKT cells can also produce IL-17. iNKT cells also produce growth factors for hematopoietic cells such as GM-CSF and multiple chemokines such as macrophage inflammatory protein (MIP)-1α and RANTES. In addition, these cells have cytolytic activity, owing to high levels of granzyme B, perforin, FasL, and TNF-related apoptosis-inducing ligand. The effector functions exhibited by iNKT cells are strongly influenced by the nature, strength, and context of the stimulus as well as by the APC type and activation status. For example, activation by cognate iNKT antigens typically results in IFN-γ and IL-4 production by iNKT cells, whereas the cytokine-driven pathway of iNKT-cell activation typically fails to induce substantial amounts of IL-4, suggesting that IL-4 production by these cells largely depends on TCR engagement. DCs appear to be the most effective APCs in presenting glycolipids and activating iNKT cells in vivo. Furthermore, although CD1d was thought to lack significant polymorphism, a recent study has identified several murine CD1d alleles that can affect the presentation of endogenous and exogenous antigens to iNKT cells (26).
iNKT cells can be divided into subsets based on CD4 and NK1.1 (in mice) expression (27, 28). These distinct subsets of iNKT cells show substantial diversity in their ability to produce cytokines (29), which might explain their functional differences. In humans, CD4+ iNKT cells produce both Th1 and Th2 cytokines, whereas CD4− iNKT cells primarily produce Th1 cytokines. Although differences in cytokine production are less dramatic in mouse iNKT-cell subsets, differences in the capacity of CD4+ and CD4− iNKT cells to regulate immune responses have been observed (28). IL-17 production by murine iNKT cells appears to be specific to a subpopulation of CD4−NK1.1− cells (29).
iNKT cells are capable of extensive cross talk with a variety of other cell types, including DCs, macrophages, neutrophils, NK cells, and conventional B and T cells, but also B-1 B cells, marginal zone B cells, γδ T cells, different subsets of regulatory T (Treg) cells, type II NKT cells, and myeloid-derived suppressor cells. In addition, because of their capacity to promote cytokines that can bias adaptive immune responses toward T helper type 1 (Th1), Th2, Th17, or Treg-cell differentiation, iNKT cells can have a profound impact on the quality of an immune response.
iNKT cells play important roles in a variety of immune responses, as shown in studies with iNKT-cell-deficient mice (because of CD1d or Jα18 gene disruption), Vα14–Jα18 transgenic animals, and/or CD1d-blocking antibodies. iNKT cells play a protective role in infections with a variety of pathogens (25), including Pseudomonas aeruginosa, B.burgdorferi, Ehrlichia muris, Chlamydia pneumoniae, Novosphingobium capsulatum, Streptococcus pneumoniae, Leishmania major, Schistosoma mansoni, Cryptococcus neoformans, influenza virus, Theiler’s murine encephalomyelitis virus, and herpes simplex viruses 1 and 2. However, iNKT cells can also play a pathogenic role during microbial infection, such as during the immune response to Chlamydia muridarum. These studies also showed that the genetic background of the animals used can have a significant impact on the contribution of iNKT cells to antimicrobial immunity. For example, iNKT cells protected against cerebral malaria in BALB/c mice but exacerbated disease in C57BL/6 mice. Likewise, iNKT cells played a protective role in the immune response to respiratory syncytial virus (RSV) in BALB/c mice, but exacerbated disease in C57BL/6 mice. In the case of infection with lymphocytic choriomeningitis virus, iNKT cells inhibited viral replication in the pancreas and liver but not in the spleen (30), suggesting that these cells can induce tissue-specific antiviral effects. Because a significant proportion of human iNKT cells express CD4 and the chemokine receptors CCR5 and CXCR6, these cells are effective targets for HIV-1 infection and might serve as an early reservoir of HIV-1 replication in infected individuals (31). iNKT cells play a protective role in host immunity to tumors (32), which has been shown in transplantable, chemically induced, and genetic tumor models. iNKT cells generally also play a protective role in autoimmunity (33), such as during the progression of diabetes in nonobese diabetic (NOD) mice and in some mouse models of multiple sclerosis (MS) and systemic lupus erythematosus. However, iNKT cells can also contribute to pathology in autoimmunity, as shown in models of collagen-induced arthritis and antibody-mediated arthritis, and during the development of primary biliary cirrhosis in transgenic mice expressing a dominant-negative TGF-β receptor or in mice infected with Novosphingobium aromaticivorans(34). iNKT cells also make an important contribution to the development of tolerance in a variety of experimental models, including anterior chamber-associated immune deviation in the eye, anti-CD4-mediated and costimulatory blockade in transplant tolerance, ultraviolet- and burn-wound-induced immune suppression, fetal tolerance, and graft-vs-host disease. iNKT cells play an important pathogenic role in the development of a variety of inflammatory diseases and hypersensitivities in mice, including atherosclerosis, contact hypersensitivity, transplant rejection, oxazolone-induced colitis, concanavalin-A-induced hepatitis, lipopolysaccharide-induced inflammation and shock, hepatic ischemia–reperfusion injury, and allergen-, ozone-, and virus-induced airway hyperreactivity. Finally, iNKT cells, through their capacity to produce colony-stimulating factors such as GM-CSF, also contribute to hematopoiesis during steady-state conditions and situations of immune activation.
While these studies have shown critical roles of iNKT cells in both health and disease, iNKT-cell-deficient mice do not develop any obvious spontaneous pathologies, except for lupus-like nephritis in aged Jα18 knockout mice (35). In addition, iNKT cells appear to be absent in ruminants (36), suggesting functional redundancy with other innate-like or CD1-restricted T-cell subsets.
Response of iNKT cells to glycolipid antigens
The in vivo response of iNKT cells to glycolipid antigen stimulation has been explored most extensively with α-GalCer (37). Within hours after α-GalCer treatment, iNKT cells produce copious amounts of cytokines, including IL-2, IL-4, and IFN-γ. Activated iNKT cells produce IL-4 more rapidly than most other cytokines, and cytokine production is short lived, lasting only a few days. iNKT cells activated in this manner also profoundly downregulate their surface TCRs, which become nearly undetectable by 8–12 h but are mostly restored by 24 h after α-GalCer injection. This phenomenon is largely responsible for the transient ‘disappearance’ of iNKT cells following α-GalCer injection when these cells are identified with reagents that bind with surface TCRs (anti-TCR-β antibodies or CD1d tetramers). Activated iNKT cells also downregulate the surface expression of the NK1.1 marker, and this downregulation is long lasting (up to 6 months). α-GalCer-activated iNKT cells also rapidly proliferate in vivo, resulting in profound expansion of the iNKT-cell population in multiple organs, which peaks around day 3 after α-GalCer treatment. In some studies, the iNKT-cell population was shown to expand up to 10- to 15-fold in the spleen and 2- to 3-fold in liver. After this expansion phase, the iNKT-cell population contracts through homeostatic mechanisms that involve the proapoptotic Bcl-2 family member Bim, reaching pretreatment levels around 10–15 days.
A cardinal feature of adaptive immunity is the development of immunological memory. However, this is not the case for iNKT cells, which develop an anergic phenotype after α-GalCer treatment (37). This anergic phenotype persists for up to 6 weeks after the initial α-GalCer treatment. A similar anergic phenotype is seen following challenge of mice with multiple bacterial pathogens or their products (38–40) and following injection of sulfatide (41), a ligand of type II NKT cells. Thus, the development of anergy might be a common outcome of iNKT-cell activation. Acquisition of an anergic phenotype by activated iNKT cells likely provides a means to avoid chronic cytokine production by these cells that could cause uncontrolled inflammation. The anergic phenotype induced by α-GalCer was intrinsic to these cells, and the proliferative defect could be overcome by culture in the presence of IL-2. iNKT-cell anergy was associated with sustained expression of the inhibitory, costimulatory receptor programmed death-1 (PD-1) (42, 43). Blockade of PD-1 interactions with its ligands, PD-L1 and PD-L2, was able to prevent the generation of iNKT-cell anergy by α-GalCer but not by bacteria or sulfatide. The critical role of the PD-1/PD-L pathway in the induction of iNKT-cell anergy was further shown in PD-1 knockout animals, which are resistant to α-GalCer-induced iNKT-cell anergy (43). Furthermore, certain glycolipid treatment modalities can effectively activate iNKT cells while avoiding the induction of anergy. This includes the delivery of α-GalCer in the context of DCs, loaded onto a recombinant CD1d molecule, formulated as a nanoparticle, or injected subcutaneously at low doses.
Numerous structural analogs of α-GalCer have been synthesized and analyzed (37). Some reagents, such as the sphingosine-truncated α-GalCer analog called ‘OCH’ (Figure 1B) and the C20:2 variant containing a diunsaturated N-acyl chain (Figure 1B), induce a Th2-biased cytokine production profile by iNKT cells. Conversely, the C-glycoside analog α-C-GalCer (Figure 1B) promotes a Th1 bias in the cytokine production profile of iNKT cells. Furthermore, iNKT cells stimulated with threitolceramide (Figure 1B), a nonglycosidic α-GalCer analog, produce a cytokine profile similar to that of α-GalCer but recover more quickly from activation-induced anergy (44).
Immunomodulatory activities of iNKT-cell ligands
The cytokines secreted by α-GalCer-activated iNKT cells can rapidly activate a variety of other cell types, including DCs, macrophages, NK cells, B cells, and conventional T cells (37, 45). This is evidenced by expression of activation markers such as CD69 on NK cells, B cells, and T cells and induction of costimulatory molecules on DCs, macrophages, and B cells. Activation of DCs involves interactions between CD40 and OX40 ligand on these cells with CD40 ligand and OX40, respectively, on iNKT cells. Activated DCs subsequently produce proinflammatory cytokines such as IL-12 and TNF-α, which further contribute to the cytokine storm. IL-12 derived from DCs and IFN-γ derived from iNKT cells also induce NK cells to produce IFN-γ. Furthermore, α-GalCer-activated iNKT cells enhance the immune responses mediated by cytotoxic T cells (CTL), CD4+ T cells, and B cells. α-GalCer promotes not only T-cell-dependent but also T-cell-independent antibody responses, probably owing to the capacity of iNKT cells to interact with CD1d on B cells, including B-1 and marginal zone B cells. These characteristics of activated iNKT cells form the basis for the adjuvant activities of iNKT-cell ligands (46).
α-GalCer also influences the quality of an adaptive immune response (37, 45). A single α-GalCer injection induces a substantial rise in serum immunoglobulin E levels, suggesting Th2 deviation of the adaptive immune response. Th2 deviation is characteristically observed in chronic α-GalCer injection protocols, which might be associated with the capacity of anergic iNKT cells to produce residual amounts of Th2 but not of Th1 cytokines.
The cytokine storm that ensues following α-GalCer treatment can also lead to significant pathology (47). Because many iNKT cells reside in the liver, a single α-GalCer treatment induces transient hepatitis in mice. α-GalCer treatment in some mouse strains can also induce abortion, probably because a substantial number of iNKT cells reside in the uterus. Furthermore, a single intranasal administration of α-GalCer sensitizes mice to the development of airway hypersensitivity. Nevertheless, α-GalCer treatment in humans, who have fewer iNKT cells than mice, has proven to be safe, at least in the short term (48). However, at present, it is unclear whether α-GalCer treatment in humans can elicit sufficiently strong biological responses to bring forth its adjuvant and therapeutic activities.
iNKT-cell antigens as vaccine adjuvants
The capacity of iNKT-cell ligands to enhance the efficacy of vaccines against multiple microbial pathogens has been tested (46, 49). Inclusion of α-GalCer in a malaria vaccine resulted in enhanced antimalarial immunity, in a manner that was dependent on IFN-γ production by iNKT cells. α-C-GalCer, which induces a Th1-biased cytokine production profile in iNKT cells, showed profoundly enhanced adjuvant activity in the malaria vaccine compared with α-GalCer. α-GalCer was also effective as a mucosal adjuvant in different vaccine formulations directed against influenza virus. Likewise, α-GalCer enhanced the immunogenicity of an HIV-1 DNA vaccine, enhancing CD4 and CD8 T-cell responses as well as humoral immunity.
The adjuvant activities of iNKT cells have also been tested in multiple cancer vaccines (46, 49). α-GalCer substantially enhanced the priming and boosting of CD8 T cells directed against ovalbumin antigens or the human cancer/testis antigen NY-ESO-1, resulting in the eradication of established tumors bearing these antigens. Coadministration of α-GalCer with irradiated plasmacytoma cells also enhanced tumor immunity, in a manner that involved DC maturation. Similarly, injection of α-GalCer-loaded and CD1d-transfected B16 melanoma cells provided protection against rechallenge of the melanoma cells.
iNKT cells as targets for immunotherapy
α-GalCer was originally discovered by Kirin Brewery Company, Ltd. (Gunma, Japan) as a natural product with antimetastatic activities in mice (50). The antitumor activities of α-GalCer have been shown in a variety of transplantable, chemically induced, and genetic tumor models (48). Although iNKT cells themselves have cytolytic activity, the tumor target cells do not need to express CD1d for therapeutic efficacy, suggesting that iNKT cells are not the relevant effector cells. Instead, NK cells and antigen-specific CTLs appear to be the critical effector cells through their cytolytic activities and their capacity to produce IFN-γ, which has antiangiogenic activities. Delivery of α-GalCer in the context of DCs provided superior antimetastatic activities, in part because this treatment avoids the induction of iNKT-cell anergy. Likewise, combined treatment with α-GalCer and antibodies that block the PD-1/PD-L pathway, thus avoiding anergy induction, was superior to α-GalCer treatment alone (42, 43). Furthermore, α-C-GalCer was more effective than α-GalCer, which is probably because of its superior capacity to induce IFN-γ production by iNKT cells, to prime DCs, and to transactivate NK cells and CD8 T cells. These preclinical studies have paved the way for phase I and phase II trials in cancer patients (48). These studies used free α-GalCer, α-GalCer presented by DCs, or transfer of cells expanded in vitro from peripheral blood mononuclear cells cultured with α-GalCer. Although these treatments were safe, it has generally been challenging to obtain strong biological responses, which is likely because of the low frequency of iNKT cells in humans, as well as the iNKT-cell dysfunction that has been observed in cancer patients. The most encouraging trial to date involved a study where patients with non-small cell lung cancer were administered four times with peripheral blood mononuclear cells cultured in the presence of α-GalCer, IL-2, and GM-CSF (51). This treatment was well tolerated and was accompanied by the induction of iNKT-cell-dependent responses. The patients with increased IFN-γ-producing cells, compared with poor responders, had an increased median survival time (responders: 31.9 months, n = 10; poor responders: 9.7 months, n = 7; P = 0.0015).
The capacity of α-GalCer to modulate microbial infection has also been investigated (52). α-GalCer treatment was generally associated with enhanced pathogen clearance and an improved disease course, as shown for Mycobacterium tuberculosis, S.pneumoniae, P.aeruginosa, Trypanosoma cruzi, C.neoformans, hepatitis B virus, and influenza virus. However, for C.muridarum infection, delayed rather than enhanced clearance was observed. Mechanisms of action appeared to be diverse. In some cases, divergent results were obtained with regard to pathogen clearance and disease. For example, α-GalCer treatment of RSV-infected mice delayed viral clearance but improved weight loss. For Plasmodium species, α-GalCer treatment promoted the clearance of organisms from mice infected with the sporozoite stages of the parasite but delayed clearance of mice infected with erythrocyte stages. This is likely because of the critical role of IFN-γ and CD8+ T cells in clearing the liver but not blood stages of the organism, which is consistent with the enhanced capacity of α-C-GalCer to clear malaria parasites from sporozoite-infected mice. Importantly, the time window in which α-GalCer treatment was effective is rather narrow, typically within a few days of infection. As such, iNKT-cell activation will likely provide little benefit to established infections. No benefit of α-GalCer treatment was observed in patients chronically infected with hepatitis C virus (53).
Because of its capacity to promote Th2 immune responses, particularly when injected repeatedly, α-GalCer has been tested in multiple mouse models of autoimmunity (33, 50). Chronic α-GalCer treatment protected NOD mice against diabetes, in a manner that involved Th2 deviation of autoantigen-specific immune responses, but the role of Th2 cytokines in this process remains unproven. Foxp3-expressing Treg cells, tolerogenic DCs, and anergy induction in pathogenic T cells all have been implicated as factors contributing to disease protection. OCH also protected against disease, and C20:2 appeared to be more effective than both α-GalCer and OCH. α-GalCer and OCH were also effective in protecting mice against the development of experimental autoimmune encephalomyelitis (EAE) in mice, a model for MS. However, these studies also showed that the timing of iNKT-cell activation can significantly impact treatment efficacy, with early treatment ameliorating disease and late treatment exacerbating disease in some studies. In these EAE models, disease amelioration was typically associated with Th2 deviation and disease exacerbation was associated with Th1 deviation. Nevertheless, both Th1 and Th2 cytokines have been implicated in disease protection, which is consistent with a pathogenic role of Th17 cells in EAE. α-GalCer also protected mice against experimental autoimmune myasthenia gravis, an antibody-dependent disease, in a mechanism that appeared to rely on Foxp3-expressing Treg cells. α-GalCer and OCH were able to protect C57BL/6 mice against collagen-induced arthritis, and this was associated with Th2 deviation. α-GalCer also attenuated collagen-induced arthritis in DBA/1 mice, with a critical role for IL-10. Surprisingly, however, a single injection of α-C-GalCer was also protective in the DBA/1 model of collagen-induced arthritis, and this was suggested to be because of general suppression of T-cell responses. In contrast with its effects on collagen-induced arthritis, α-GalCer moderately enhanced joint inflammation in an antibody-mediated arthritis model. α-GalCer, OCH, and α-C-GalCer were all protective in an experimental model of ocular autoimmunity, and surprisingly, α-C-GalCer was most effective (54). This protective effect of α-C-GalCer was associated with dampening of Th1 and Th17 effector functions. α-GalCer was generally protective in chemically induced and genetic models of lupus-like disease, but in some mouse strains, or when administered late in the disease process, it exacerbated disease (55). Disease protection was typically associated with Th2 deviation. α-GalCer was also protective in a model for autoimmune thyroiditis, in a manner that correlated with reduced IFN-γ and reduced autoantibody production. Collectively, these studies have shown that iNKT-cell activation typically protects against autoimmunity but that treatment efficacy is influenced by a variety of parameters, including the nature and dose of the iNKT-cell antigens, the frequency and route of injections, the timing of treatment relative to disease initiation and progression, the particular animal model used, and the genetic background of the animals (33, 50). Many of these issues pose problems for translation to treatment of human autoimmunity.
A single injection of α-GalCer into recipient mice significantly reduced morbidity and mortality of graft-vs-host disease (56). This was dependent on stimulation of host iNKT cells and Th2 deviation of donor T cells.
Although iNKT cells exhibit potent therapeutic properties in a number of diseases, iNKT-cell activation can also exacerbate disease such as allergic reactions and atherosclerosis in mice (33, 55).
Conclusions and outlook
iNKT cells are unique lymphocytes that bridge the innate and adaptive immune systems. These cells acquire their innate-like phenotype during thymic development by interaction with CD1d–glycolipid complexes on double-positive cells and under direction of the transcription factor PLZF. iNKT cells can respond to endogenous and exogenous glycolipid antigens but can also become activated indirectly by cytokine-driven mechanisms. Quickly after activation, these cells produce a variety of cytokines with potent immunomodulatory activities, but they then enter a period of quiescence. iNKT cells regulate a variety of immune responses. iNKT-cell ligands exhibit potent immune-enhancing properties that are being exploited for the development of vaccine adjuvants. iNKT-cell activation also has therapeutic effects in mice, and clinical trials with α-GalCer for cancer are underway. The preclinical studies with α-GalCer and related iNKT-cell antigens have also shown potential risks associated with iNKT-cell activation that will need to be overcome in order to develop safe and effective iNKT-cell-based vaccine adjuvants and therapies. Future studies should focus on developing means to better control the outcome of iNKT-cell activation on immune responses and disease using diverse treatment modalities and within genetically diverse hosts.
We apologize to colleagues whose work we did not cite because of space constraints or omission. We thank many colleagues, especially Dr Sebastian Joyce (Vanderbilt University School of Medicine), for helpful discussions. The authors’ work was supported by grants from the National Institutes of Health (to LVK), a discovery grant from the Diabetes Research and Training Center at Vanderbilt University School of Medicine (to LW), a predoctoral fellowship from the National Institutes of Health (to CLG), and a postdoctoral fellowship from the National Multiple Sclerosis Society (to VVP).