β-Glycosphingolipids as Immune Modulators

Abstract

β-Glycosphingolipids have emerged as a family of potential ligands for natural killer T (NKT)- regulatory lymphocytes. This subset of regulatory lymphocytes has been implicated in the regulation of autoimmune processes. The major histocompatibility complex (MHC) Class I-like CD1d glycoprotein is a member of the CD1 family of antigen-presenting molecules and is responsible for selection of NKT cells. β-Glycolipids have been shown to alter immune responses in the opposing settings of autoimmune diseases or cancer. In this review, we discuss the potential use of β-glycoshpingolipids for NKT-based immunotherapy.

Keywords :

• Ligands
• glycosphingolipids
• NKT cells

Abbreviations
NKT	=	(natural killer T-cell)
β-GC	=	(β-glucosylceramide)
β-GalCer	=	(β-galactosylceramide)
α -GalCer	=	(α -galactosylceramide)

Plasticity of Regulatory Lymphocytes

The term “plasticity” is not well defined for regulatory lymphocytes. Several authors have used this term interchangeably with “flexibility” or to describe the ability to pursue distinct actions in different immune settings (Prohaska et al., 2002; Stockinger et al., 2006; Rothenberg, 2007). It is not yet clear whether plasticity is a requisite property of regulatory cells, represents different behaviors in various environments, or is a unique ability for intercellular cross-talk. In some situations it may require a duality in function, such as the capability for both T_H1 and T_H2 cytokine secretion. The plasticity of a regulatory cell may evolve from the use of different ligands or from signals in the immune microenvironment (Rothenberg, 2007). It may also depend on cell-cell interaction, or on the influence of cytokines or co-stimulatory molecules (Weaver et al., 2006). Antigen presentation and antigen-presenting cells (APC) may play an important role in this setting (Morel et al., 2003).

On the other hand, plasticity may simply be the result of the natural programs of different subsets of cells (Rothenberg and Dionne, 2002). Different functions of lymphocytes depend on membrane structure and, more specifically, on lipid rafts(Bendelac et al., 1997) and the potential functional effects of the integration of glycosphingolipids within the membrane structure(Godfrey et al., 2004). Thus, one may theoretically classify the generation of different immune responses triggered by the same ligand into two types of plasticity: Type-I plasticity, or “true” plasticity, in which identical cells and ligands produce different immune responses; and, Type-II plasticity, or “apparent” plasticity, in which exposure to glycolipid ligands alters cellular properties so that the cellular population in question is no longer homogenous, with the resultant net effect being a function of the balance between these cellular subpopulations.

NKT Lymphocytes: Development and Classification

Natural killer T (NKT)-lymphocytes are a unique lineage of T-lymphocytes that share properties with both NK cells and memory T-lymphocytes (Bendelac et al., 1997). The identification of NKT cells is based on further classification of T-lymphocytes according to the cell surface markers they present, by which they resemble NK cells and T-lymphocytes, and to biological function in terms of cytokine production and ligand reactivity (Godfrey et al., 2004). This subset of lymphocytes is CD1d-reactive, may be either CD4⁺ or double negative (DN; CD4⁻ and CD8⁻ negative) and express NK cell and memory T-lymphocyte markers. They are unique in their invariant Vα 14Jα18 T-cell receptor (TCR) α -chain, and their TCR β-chain is biased towards Vβ8.2, Vβ2, and Vβ7. NKT cells are also unique in their glycolipid antigen reactivity and in their marked cytokine production (Bendelac et al., 2007).

Studies in mice have helped to identify a subset of α β T-cell receptor (α β-TCR)⁺ lymphocytes with the following properties: (1) DN for CD8 and CD4; (2) intermediate level of TCR expression; and (3) Vβ8 expression that is 2- to 3-fold higher than normal T-lymphocytes (Godfrey et al., 2004). Other reports have demonstrated a subset of α β-TCR⁺ cells that express the NK1.1 marker (Nkrp1c or CD161c), which was once thought expressed exclusively by NK cells. Among the NK1.1⁺ T-lymphocytes, two further subgroups have been identified: CD4⁺ and DN (Godfrey et al., 2004). Additional characteristics have been derived from analysis of the TCRs, in which both α and β chains follow certain patterns. The TCR-α chain is invariant and contains Vα 14Jα 18, while the β-chain is biased towards Vβ8.2, Vβ2, and Vβ7 (Bendelac et al., 2007). Other characteristics of the NK1.1⁺α β-TCR cells have been defined by their biological interactions. Development of these cells requires β2-microglobulin, despite the lack of CD8 expression, and is independent of major histocompatibility complex (MHC) Class II expression (Ohteki and MacDonald, 1994). NKT cells are reactive to the MHC Class I-like molecule CD1d (Bendelac et al., 1995) and have also been reported to be highly reactive to an α -structured-glycolipid known as α -galactosylceramide (α -GalCer) (Brigl and Brenner, 2004).

Double negative α β-TCR⁺ lymphocytes have been reported to produce high amounts of cytokines such as interleukin-4 (IL-4), interferon-γ (IFNγ), and tumor necrosis factor (TNF) (Zlotnik et al., 1992). Furthermore, the fact that certain subsets of CD4⁺ T-lymphocytes in the thymus produce large amounts of cytokines (Hayakawa et al., 1992), while NK1.1⁻CD4⁺ thymocytes do not (Arase et al., 1993), suggests that the fraction of CD4⁺ cells that are highly active in cytokine production may be NKT cells. NKT cells were originally identified in mice, and similar cells were later found in rats, non-human primates (Motsinger et al., 2003), and humans (Porcelli et al., 1993; Dellabona et al., 1994). In humans, the cells are characterized by expression of invariant Vα 24Jα 18 TCR α -chain and Vβ11 TCR β-chain (Godfrey et al., 2004).

One of the main challenges in the initial classification of NKT lymphocytes has been the fact that the molecular characteristics used to identify NKT lymphocytes varied among mouse strains, and between mice and humans. For example, the identification of NKT lymphocytes based on expression of NK1.1 is appropriate when dealing with mouse strains, such as C57BL/6, that express the NK1.1 marker. However, NK1.1-positivity is irrelevant in other widely used strains, such as Balb/c, that do not express NK1.1 (Godfrey et al., 2006). Furthermore, in humans, the expression of the NK1.1 homologue, CD161, is not limited to T-lymphocytes with NKT cell characteristics (Godfrey et al., 2004). Because these cells cannot be defined solely by their molecular characteristics, this subgroup of lymphocytes has been referred by several terms, including NK T or NKT cells, NK1.1+ (like) T-lymphocytes, natural T-lymphocytes, iNKT cells, and Vα 14 invariant (Vα 14i) T-lymphocytes (Godfrey et al., 2000). “CD1d-dependent natural killer-like T-lymphocytes” may be the most accurate designation; however, the term NKT cell is the most widely used term (Godfrey and Kronenberg, 2004; Berzins et al., 2005).

NKT cells can be categorized into three subgroups: (1) Human NKT with invariant TCR, Vα 14 invariant or Vα 24 invariant NKT lymphocytes expressing semi-invariant TCRs that recognize CD1d encoded by Vα 14 and Jα 18 gene segments in mice, or Vα 24 and Jα 18 (Tsuji, 2006); (2) NKT lymphocytes that are also restricted to CD1d but use more diverse TCRs than the first group (Tsuji, 2006); and (3) non-CD1d-restricted NKT lymphocytes that is not dependent on CD1d molecules (Bendelac et al., 2007).

NKT cell development begins in the thymus and continues in the periphery (Bendelac et al., 2007). Common thymocytic precursors, positive for both CD4 and CD8 (DP, double positive), undergo random TCR gene rearrangement resulting in the expression Vα 14Jα 18 in conjunction with either Vβ8.2, Vβ7, or Vβ2, leading to CD1d-dependent selection and formation of mature NKT (Bendelac et al., 2007). Expression of NK1.1 generally follows migration out of the thymus, but can also precede it (Godfrey et al., 2004). Unique molecular interactions govern NKT development and emigration from the thymus; yet it remains unclear whether there is a single autologous antigen responsible for both positive selection and peripheral activation of these regulatory lymphocytes (Kronenberg and Engel, 2007).

Although the majority of peripheral NKT cells are mature, there is a substantial constitutive population of immature NKT cells that migrate from the thymus to the periphery. These immature cells are functional, but are likely to behave differently than their mature counterparts (Godfrey et al., 2004). The importance of TCR or CD1d in the development of NKT cells is evidenced by the selective deficiency of NKT cells following disruption of TCR-α chain or CD1d results in, with other types of lymphocytes remaining intact (Crowe et al., 2002).

In mice, NKT cells are most commonly found in the liver and account for 30% to 50% of hepatic lymphocytes (Crispe, 2003). However, NKT cells may be present wherever regular T-lymphocytes are found, accounting for 20–30% of bone marrow lymphocytes, 10–20% thymic lymphocytes, and constitute smaller proportions in the lung (7%), blood (4%), spleen (3%), and lymph nodes (0.3%) (Godfrey et al., 2000). In humans, the exact distribution of NKT cells in different organs has not been fully determined; however, in the liver, the fraction of NKT cells varies between 4% and 20% (Doherty et al., 1999; Ishihara et al., 1999).

Immune Function of NKT Lymphocytes

NKT cells, CD1d-restricted NKT cells in particular, have several functions; some of these are characterized as direct effector functions and others as immunomodulatory functions (Burdin et al., 1999; Van Kaer, 2004a). Upon activation, NKT cells rapidly produce different cytokines, including IFNγ and IL-4, express activation markers (e.g., CD69), down-regulate NK1.1 surface antigen, and disappear from the tissues in which they are usually found, probably due to activation-induced cell death (Eberl and MacDonald, 1998; Matsuda et al., 2000; Osman et al., 2000; Laloux et al., 2002; Van Kaer, 2004a). Activated NKT cells display cytotoxic capability, mediated by Fas, granzyme A/B, perforin, and granulysin (Taniguchi and Nakayama, 2000; Gansertet al., 2003). This process eventually results in the activation of both the innate and the adaptive immune system (Carnaud et al., 1999; Kitimura et al., 1999; Hermans et al., 2003; Van Kaer, 2007). Co-administration of α -GalCer with an antigen shifts the antigen specific T-lymphocyte response towards the T_H1 cytokine profile (Cui et al., 1999), whereas other reports have demonstrated a T_H2 polarization following α -GalCer administration (Singh et al., 1999). The ability of NKT cells to generate either T_H1 or T_H2 responses indicates their importance as immunoregulatory cells and the complexity of their modulatory machinery-different responses can be generated by the same ligands (Godfrey and Kronenberg, 2004; Zigmond et al., 2007). In addition, NKT activation can lead to NK cell activation, proliferation of memory CD4⁺ and CD8⁺ T-lymphocytes, or expression of the early activation marker CD69 on the surface of T- and B-lymphocytes (Erbl et al., 2000; Nishimura et al., 2000).

NKT lymphocytes have been implicated in the regulation of autoimmune processes in both mice and humans (van Kaer, 2004b). Colitis was induced in C57/B6 mice by intracolonic instillation of trinitrobenzenesulfonic acid (TNBS). Oral tolerance was induced by administration of five oral doses of proteins extracted from TNBS-colitis colonic wall (Trop et al., 2003). Clinical, macroscopic, and microscopic scores were used for colitis assessment. To evaluate the putative role of AsGm-1 in tolerance induction, depletion of AsGm-1 expressing cells was performed. Then to evaluate the mechanism of tolerance induction, liver-associated NKT lymphocytes were harvested 14 d following tolerance induction, and cultured with concanavalin A (ConA) and colitis-extract proteins. Orally tolerized mice exhibited significant alleviation of the clinical, macroscopic, and microscopic parameters of colitis, with increased CD4⁺ILA⁺/CD4⁺IFNγ⁺ lymphocyte ratio, increased IL-4, and decreased IFNγ and IL-12 serum levels (Trop et al., 2003).

In contrast, orally fed mice that were AsGm-1-depleted displayed evidence of severe colitis. These mice exhibited significantly decreased CD4⁺IL-4⁺/CD4⁺IFNγ⁺ ratios, and an increase in IFNγ and IL-12, accompanied by decreased IL-4 levels. NKT cells harvested from tolerized mice secreted high levels of anti-inflammatory cytokines. In contrast, in non-tolerized mice, NKT cells secreted primarily pro-inflammatory cytokines. Thus, in a tolerized environment, both NK1.1- and AsGm-1-expressing cells are essential for disease alleviation. In contrast, in a non-tolerized environment, AsGm-1 expressing cells support an anti-inflammatory immune paradigm, while NKT lymphocytes support a pro-inflammatory shift (Trop et al., 2003).

The aim of one recent study was to evaluate the in vivo effect of adoptive transfer of immune-programmed NKT cells (Shibolet et al., 2004b). Colitis was induced in C57/B6 mice, from which NKT, CD4, and CD8 lymphocytes, and dendritic cells (DC) were prepared from spleens of naive mice, animals with colitis, and animals with colitis that were orally tolerized. Subsets of splenocytes, NKT, CD4, CD8, NKT⁺CD4, NKT⁺CD8, and NKT⁺DC lymphocytes were prepared. Assessment of the T_H1/T_H2 cytokine secretion paradigm in vitro was performed before and after exposure to the antigen.

Adoptive transfer of ex vivo immune-programmed lymphocytes from each group was performed into recipient mice, followed by colitis induction. Ex vivo exposure of NKT cells harvested from mice with colitis-to-colitis proteins [colitis-extracted proteins (CEP)] led to a T_H2 cytokine shift (Shibolet et al., 2004b). The IL-4/IFNγ ratio increased for NKT harvested from colitis-harboring mice following exposure to CEP. Adoptive transfer of ex vivo-educated NKT lymphocytes harvested from colitis-harboring mice significantly alleviated experimental colitis in vivo. The number of intrahepatic NKT lymphocytes increased significantly in mice transplanted with NKT lymphocytes harvested from colitis-harboring donor mice that were ex vivo-exposed to CEP, similar to mice transplanted with NKT lymphocytes harvested from tolerized donors. Exposure of NKT cells to the disease-target antigen induced a significant increase in the IL-4/IFNγ cytokine ratio. Adoptive transfer of a relatively small number of immune-programmed NKT cells induced a systemic T_H1 to T_H2 immune shift and alleviated immune-mediated colitis (Shibolet et al., 2004b).

In order to determine the role of NK1.1⁺ cells in the induction and maintenance of pro-inflammatory and/or tolerizing responses, colitis was induced in C57/B6 donor mice (Menachem et al., 2005). These donor mice received five oral doses of colonic proteins extracted from TNBS-colitis colonic wall. Depletion of NK1.1⁺ lymphocytes was performed before lymphocyte harvesting. Splenocytes were harvested and separated into T-cell subpopulations, then transplanted into recipient mice before intracolonic instillation of TNBS. The adoptive transfer of CD4⁺ and NK1.1⁺ cells harvested from tolerized mice markedly ameliorated the colitis in recipient mice (Menachem et al., 2005).

In contrast, the adoptive transfer of CD8⁺ or DN lymphocytes failed to transfer this tolerance. Recipients of splenocytes from tolerized mice exhibited an increase in CD4⁺IL-4⁺/CD4⁺IFNγ⁺ ratio. In contrast, recipients of splenocytes from NK1.1-depleted-tolerized mice exhibited severe colitis with a significant decrease of the CD4⁺IL-4⁺/CD4⁺IFNγ⁺ ratio. However, adoptive transfer of splenocytes from non-tolerized NKT-depleted mice led to an alleviation of colitis with a relative increase of the CD4⁺IL-4⁺/CD4⁺IFNγ⁺ratio. These data suggest that NK1.1⁺ lymphocytes play a critical role in immune regulation. These cells may be accountable for the alteration of the inflammatory response and the CD4⁺IL-4⁺/CD4⁺IFNγ⁺ ratio in immune-mediated colitis and in peripheral tolerance induction (Menachem et al., 2005).

Reduced numbers and defective function of NKT lymphocytes have been demonstrated in non-obese diabetic (NOD) mice. Transplantation of NKT lymphocytes, introduction of a Vα 14Jα 281 transgene into the NOD background, or activation of NKT lymphocytes by administration of α -GalCer ameliorates diabetes in this model (Miyake and Yamamura, 2007; Novak et al., 2007). Administration of α -GalCer and OCH, a sphingosine-truncated analogue of α -GalCer, had protective effects in murine experimental autoimmune encephalomyelitis (EAE) and collagen-induced arthritis, respectively (Miyake and Yamamura, 2005). In these models, disease amelioration was associated with a shift in the immune balance from a pathologic T_H1 type response towards a protective T_H2 type response.

Altered number or function of NKT cells has also been described in several human autoimmune diseases (Miyake and Yamamura, 2005). Patients with systemic lupus erythematosus, scleroderma, diabetes, multiple sclerosis, Sjogren syndrome, and rheumatoid arthritis were found to have lower numbers of peripheral blood NKT cells compared to healthy subjects. Invariant Vα 24Jα Q cells were found to have a regulatory role in human subjects with scleroderma (Miyake and Yamamura, 2007).

NKT cells are also the prominent driver in asthma development (Jinquan et al., 2006). Asthmatic NKT cells migrate from the thymus, spleen, liver, and bone marrow into blood vessels, and then concentrate in airway bronchi mucosa. This recruitment is dependent on high expression of CCR9 and engagement of CCL25/CCR9. NKT cells promote asthma via two different pathways. One is an indirect pathway. NKT cells contact with CD3⁺ T-lymphocytes and induce them to secrete large quantities of T_H2 cytokines (IL-4, IL-13). This requires the participation of dendritic cells and the synergic action of CCL25/CCR9 and CD226 (Jinquan et al., 2006). The other is a direct pathway. Circulating asthmatic NKT cells highly selectively express T_H1 cytokines. Once they reach the airway epithelium, most NKT cells shift to a T_H2-bias, highly expressing IL-4 and IL-13, but not IFNγ. Both pathways lead to airway hyper-responsiveness and inflammation, with resultant asthma (Jinquan et al., 2006).

Recent studies indicate that invariant TCR⁺ CD1d-restricted natural killer T (iNKT) cells play an important role in regulating the development of asthma and allergy (Meyer et al., 2007). iNKT cells can function to skew adaptive immunity toward T_H2 responses, or can act directly as effector cells at mucosal surfaces in diseases such as ulcerative colitis and bronchial asthma (Meyer et al., 2007). In mouse models of asthma, NKT cell-deficient strains fail to develop allergen-induced airway hyperreactivity (AHR), a cardinal feature of asthma, and NKT cells are found in the lungs of patients with chronic asthma, suggesting a critical role for NKT cells in the development of AHR. Interestingly, Vα 14 NKT cells are not essential for the induction of specific IgE response, but instead tend to induce suppression of specific IgE upon α -GalCer activation in vivo (Iwamura and Nakayama, 2007).

The suppression of IgE production is not detected either in Vα 14 NKT cell-deficient mice or in IFNγ-deficient mice. Therefore, activated Vα 14 NKT cells are able to exert a potent suppressive activity on T_H2 cell differentiation and subsequent IgE production by producing a large amount of IFNγ (Iwamura and Nakayama, 2007). In an OVA-induced asthma model, α -GalCer administration inhibited airway inflammation and airway hyperreactivity by IFNγ from activated Vα 14 NKT cells, thus suggesting the negative regulation of T_H2-responses by the activated Vα14 NKT cells (Iwamura and Nakayama, 2007).

Ligands for NKT Regulatory Cells

Unnatural glycosphingolipids have been synthesized and used to provide insight into the various functional pathways associated with receptor binding of these ligands (Bittman, 2004). The biological activities of ceramides and glycosphingolipids are highly variable with apparently small changes in molecular structure. For example, recent studies suggest that sphingolipid analogs can be used in various systems: as anticancer agents; as probes of protein targets of bioactive lipids and of glycosphingolipid transbilayer distribution in bilayers with and without sphingomyelin; as modulators of cholesterol-enriched microdomains (rafts) that may facilitate fusion of alphaviruses with target membranes; as enhancers of membrane permeabilization induced by cholesterol-specific cytolysins; and as probes for the selective internalization of glycosphingolipids in caveolae of living mammalian cells (Bittman, 2004).

The trans 4, 5-unsaturation of the sphingosine backbone promotes closer packing and lower compressibility of ceramide analogs in the lipid-water interface relative to comparable saturated species (Miyake and Yamamura, 2007). Therefore, trans unsaturation in the vicinity of C4 of the sphingoid backbone may augment intramolecular hydration/hydrogen bonding in the polar region (Brockman et al., 2004). A truncated, non-isosteric, glycosidase-resistant C-α GalCer analog in which the anomeric carbon is bonded directly to the C1 of the phytoceramide backbone was a less potent agonist for NKT cells than the O-glycoside α GalCer form. However, this analog induced cytokine production with the highest IFNγ :IL-4 and IFNγ : IL-13 ratios. These data suggest that a new mimetic of α GalCer may preferentially promote T_H1-immune responses (Lu et al., 2006).

Variation in the length of the long-chain base and in the structure of the carbohydrate-containing polar head group of (2S, 3R) (or D-erythro)-β -lactosylceramide (LacCer) did not alter the mechanism of endocytic uptake from the plasma membrane. LacCer is taken up from the plasma membrane almost exclusively via caveolae in human skin fibroblasts and other cell types (Singh et al., 2003). The three unnatural stereoisomers [(2R,3R)-, (2S,3S)-, and (2R,3S)-] of dipyrromethene difluoride (BODIPY)-LacCer were evaluated recently as probes for the role of stereochemistry of the long-chain base of LacCer in the mechanism of endocytic uptake (Liu and Bittman, 2006). The stereochemistry in the sphingosine backbone of LacCer dramatically affects the mechanism of LacCer endocytosis. The principal basis for this change in mechanism is how the sphingosine moiety is associated with other lipids within membrane microdomains (rafts). Only the D-erythro stereoisomeric form associates with microdomains.

Several glycosphingolipids and phospholipids derived from mammalian, bacterial, protozoan, and plant species have been identified as possible natural ligands for NKT lympho-cytes (Zajonc et al., 2005). Some have been crystallized in CD1d-bound forms, revealing their tertiary structure (Tsuji, 2006). The two main subgroups of NKT lymphocytes recognize CD1d molecules. The CD1 molecules, which present non-peptide, mostly lipid antigens, resemble MHC molecules, especially MHC Class I molecules in that their heavy chains bind non-covalently with β2-microglobulin (Ulrichs and Porcelli, 2000; Gumperz and Brenner, 2001). In humans, CD1 molecules are encoded on chromosome 1, with four possible isoforms: CD1a, CD1b, CD1c, and CD1d. Sequence homology allows further classification of the family into two groups. Group 1 consists of CD1a, CD1b, and CD1c, and does not exist in mice. Group 2 consists of CD1d, and is common to both mice and humans (Gumperz and Brenner, 2001). CD1d molecules are constitutively expressed by several different cells, including antigen-presenting cells such as macrophages and dendritic cells, and B- and T-lymphocytes in the thymus and liver (Brossay et al., 1997; Roark et al., 1998). CD1d molecules present antigens, primarily glycosphingolipids, to NKT lymphocytes (Bendelac et al., 1995).

Several potential natural glycosphingolipids have been suggested to activate NKT lymphocytes, including glycosphingolipids such as glycosylphosphatidylinositol (Brutkiewicz, 2006), iso-globotirhexosylceramide (Zhou et al., 2004), and α -glucuronsylceramide (Kinjo and Kronenberg, 2005; Mattner et al., 2005), and phospholipids such as phosphatidylcholine and phosphatidylinositol (De Silva et al., 2002; Giabbai et al., 2005). It is noteworthy that not all naturally occurring lipid ligands are stimulatory. Gangliotriaosylceramide, which is secreted by certain murine T-lymphocyte lymphoma cells, inhibits CD1d-mediated antigen presentation (Sriram et al., 2002).

The semi-invariant α β -TCRs can recognize isoglobotrihexosylceramide (iGb3), a mammalian glycosphingolipid, and a microbial α -glycuronylceramide found in the cell walls of Gram-negative, lipopolysaccharide-negative bacteria (Zhou et al., 2006). iGb3 was proposed as one of the candidates recognized by NKT lymphocytes under pathological conditions such as cancer and autoimmune disease (Hansen and Schofield, 2004; Zhou, 2006). However, no direct evidence for its physiological role has been provided. Recent studies in mice deficient in iGb3 synthase revealed that these mice developed, grew, and reproduced normally, while exhibiting no overt behavioral abnormalities (Porubsky et al., 2007).

Consistent with the notion that iGb3 is synthesized only by iGb3S, the lack of iGb3 in the dorsal root ganglia of iGb3S-deficient mice (iGb3S^−/−) was confirmed. iGb3S^−/− mice showed normal numbers of iNKT cells in the thymus, spleen, and liver, and these cells expressed selected TCR Vβ chains identical to controls (Porubsky et al., 2007). Upon administration of α -galactosylceramide, activation of iNKT and dendritic cells was similar in iGb3S^−/− and iGb3S^+/− mice, as measured by up-regulation of CD69, as well as intracellular IL-4 and IFNγ in iNKT cells, up-regulation of CD86 on dendritic cells, and a rise in serum concentrations of IL-4, IL-6, IL-10, IL-12p70, IFNγ, TNFα, and Ccl2/MCP-1 (Porubsky et al., 2007). These results suggest that iGb3 is unlikely to be an endogenous CD1d lipid ligand determining thymic iNKT selection.

The dual recognition of self and microbial ligands underlies the innate-like anti-microbial functions mediated by CD40L induction and massive cytokine and chemokine release by NKT lymphocytes (Skold and Behar, 2003). There are multiple ways by which NKT lymphocytes are activated during microbial infection, and some may be associated with proteins that control lipid metabolism (Tupin et al., 2007). During an infectious assault, the presentation of a neo-self glycolipid by APCs activates iNKT cells, which release pro-inflammatory or anti-inflammatory cytokines and jump-start the immune system (Stanic et al., 2003b).

In 1994, several glycosphingolipids were extracted from the Okinawan marine sponge Agelas mauritanus (Van Kaer, 2005). These agelasphins, which consist of D-galactose and ceramide, showed anti-tumor activity in mice (Taniguchi et al., 2003). Because only small amounts of agelasphins are present in each sponge, the modified derivative α -GalCer was produced (Wilson et al., 2002). Indeed, the only efficient method for selectively stimulating NKT lymphocytes in vivo is via the sea sponge-derived agent α -GalCer (Van Kaer, 2005). Multimers of CD1d1-α -GalCer and α -GalCer analog-loaded complexes demonstrate cooperative engagement of the Vα 14Jα 18 iNKT cell receptor (Taniguchi et al., 2003; Bezbradica et al., 2005). Administration of α -GalCer causes potent activation of NKT lymphocytes, rapid and robust cytokine production, and activation of a variety of cells of the innate and adaptive immune systems (Van Kaer, 2005). Administration of α -GalCer induces secretion of both IL-4 and IFNγ. Repeated administration favors the production of T_H2 cytokines (Yamamura et al., 2004; Miyake and Yamamura, 2005). OCH is a unique analogue that selectively stimulates NKT to produce IL-4 and IL-10, and may be more beneficial for suppression of T_H1-mediated diseases.

α-GalCer is a potent ligand for NKT lymphocytes after binding to CD1d expressed by APCs. The CD1d/α -GalCer complex is recognized by NKT lymphocytes via Vα 14i TCR in mice and Vα 24i TCR in humans (Schmieg et al., 2005). Such NKT lymphocyte activation results in the production of both T_H1 and T_H2 cytokines, an observation common to both species (Kawano et al., 1997; Burdin et al., 1999). The necessity of the CD1d molecule and the specificity of Vα 14i-NKT cells in this process have been demonstrated by the absence of the reaction in CD1d- and Vα 14i NKT-deficient mice (Kawano et al., 1997).

CD1d-mediated recognition of α -GalCer is highly conserved. Brossay et al. (1998) showed that human NKT lymphocytes recognize α -GalCer presented by mouse CD1d molecules and vice versa (Brossay et al., 1998). However, α -GalCer has been shown to be hepatotoxic in mice, limiting its use in human testing (Osman et al., 2000). Our understanding of the processes involved in the stimulatory or inhibitory NKT pathways has been advanced through investigations using several analogs of α -GalCer in different models: an α -GalCer analog with a shortened sphingosine chain (OCH) were even more effective in EAE models than α -GalCer, provoking increased IL-4 production (Miyamoto et al., 2001); KRN 7000, an analog with a di-unsaturated 20 carbon chain (C20:2), was reported to stimulate IL-4 production, yet inhibit IFNγ (Yu et al., 2005); and α -C-GalCer, a C-glycoside analog, was reported to generate a longer lasting reaction, with a cleaner T_H1 response (Gonzalez-Aseguinolaza et al., 2000, 2002). In addition to α -GalCer and its analogs, β-GalCer (C12) has also been reported to be a CD1d ligand capable of stimulating NKT lymphocytes (Ortaldo et al., 2004; Parekh et al., 2004).

A recent study evaluated 16 analogs of α -GalCer for the CD1-mediated TCR activation of naive human NKT lymphocytes and their anti-cancer efficacy (Chang et al., 2007). In vitro, glycosphingolipids containing an aromatic ring in their acyl or sphingosine tail were more effective than α -GalCer in inducing T_H1 cytokines/chemokines, TCR activation, and human NKT lymphocyte expansion (Chang et al., 2007). However, none of these glycosphingolipids can directly stimulate human dendritic cell maturation, except for a glycolipid with an aromatic ring on the sphingosine tail.

Activation of the NKT cell TCR with phosphorylation of CD3η, ERK1/2, or CREB is correlated with the induction of T_H1 cytokines. Notably, the extent of NKT lymphocyte activation when glycolipid is presented by APCs is greater than when glycolipid was presented by non-antigen-presenting cells (Chang et al., 2007). In vivo, in mice bearing breast or lung tumors, the glycosphingolipids that induced more T_H1-biased cytokines and CD8/CD4 T-lymphocytes displayed significantly greater anti-cancer potency than α -GalCer. These findings indicate that different ligands can be designed to favor T_H1-biased immunity, directing the final effect of NKT lymphocytes. Furthermore, current studies of different unnatural glycosphingolipids may promote better understanding of the relationship between structure and function, thus allowing more accurate and fine-tuned use of such ligands in the future.

Immunomodulatory Activities of α -GalCer and its Analogues

Investigators have studied the role of activated NKT lymphocytes in various types of pathology, including infectious, autoimmune, and neoplastic conditions (Wilson et al., 2003; Miyake and Yamamura, 2007).

1. Hepatitis B Viral Infection: In hepatitis B virus transgenic mice, NKT lymphocyte activation abolished HBV (Kakimi et al., 2000).

2. Tuberculosis: CD1-restricted T-lymphocytes recognize both pure lipid mycobacterium antigens presented by CD1, as well as those processed by macrophages infected by Mycobacterium tuberculosis (Chackerian et al., 2002). Some activated CD1-restricted T-lymphocytes with with granulysin-mediated bactericidal capabilities can lyse uninfected macrophages (Stenger et al., 1998). The role of NKT lymphocytes in granuloma formation has also been demonstrated (Apostolou et al., 1999); however, lipid antigen processing and presentation during mycobacterium lung infection was not efficient enough to induce NKT lymphocyte activation (Chackerian et al., 2002). α -GalCer-mediated activation was protective in this setting.

3. Cryptococcus infections: Cryptococcus neoformans is a ubiquitous fungal pathogen. Among the host factors determining susceptibility and vulnerability is the balance between T_H1 and T_H2 responses; T_H1 predominant responses are protective (Kawakami et al., 1997). α -GalCer-induced NKT lymphocyte activation has been shown to increase IFNγ production and augment local host resistance to C. neoformans infection (Kawakami et al., 2001a, 2001b).

4. Malaria: Administration of α -GalCer enhances malaria immunity in mice that have received sub-optimal doses of irradiated sporozoites or recombinant virus expressing sporozoite antigens (Gonzalez-Aseguinolaza et al., 2002). Interestingly, α -GalCer had a maximal adjuvant effect when co-administered with the antigens, but provided no adjuvant effect when given 2 d prior to antigen administration (Gonzalez-Aseguinolaza et al., 2002).

5. NKT lymphocytes and encephalitis: In an EAE model, α -GalCer-induced NKT lymphocyte activation variously potentiated or prevented disease, depending on the NKT reaction (Jahng et al., 2001). IFNγ secretion was associated with disease exacerbation, whereas IL-4 production was protective against EAE. Prior immunization with α -GalCer resulted in increased IL-4 secretion.

6. Malignancy: The role of NKT lymphocytes in the setting of neoplastic disease remains under investigation. Having demonstrated that NKT lymphocytes are capable of causing tumor regression via IL-13-mediated inhibition of tumor-specific cytotoxic T-lymphocytes (Terabe and Berzofsky, 2004), it was suggested that NKT lymphocytes might normally inhibit tumor immunity. In contrast, the absence of NKT lymphocytes has been associated with increased risk of tumorigenesis (Smyth et al., 2000). Crowe et al. (2002) asked why the potential for NKT lymphocytes to generate an anti-tumor reaction depended on exogenous stimuli such as IL-12 and α -GalCer for some tumors (B16F10 melanoma), but not others (sarcomas). They hypothesized that different tumors display different CD1d-binding glycolipid antigens.

7. Diabetes: In NOD mice in which T_H1 cells mediate diabetes, α -GalCer-activated NKT cells can prevent pancreatic islet β cell destruction (Hammond et al., 1998; Hong et al., 2001).

Several studies have suggested that disease-target antigens can serve as NKT cell ligands. In NOD mice, NKT-dependent amelioration of diabetes is noted following vaccination with disease-target antigens such as insulin, GAD65, and streptozotocin-treated islets (Hauben et al., 2005). Adoptive transfer of ex vivo disease target antigen-exposed NKT lymphocytes alleviated immune-mediated colitis (Trop et al., 1999; Menachem et al., 2005). NKT lymphocytes exposed to tumor antigens suppressed tumor growth (Shibolet et al., 2004a). In asthma, NKT lymphocytes migrate from the thymus, spleen, liver, and bone marrow into blood vessels, and then concentrate in airway bronchi mucosa (Jinquan et al., 2006). This recruitment is dependent on CCR9 expression and engagement of CCL25/CCR9 in response to target antigens. Circulating asthmatic NKT lymphocytes express high levels of the T_H1 cytokine IFNγ. Once they reach the airway epithelium and are exposed to disease-target antigens, most NKT lymphocytes shift to a T_H2-bias, and express high levels of IL-4 and IL-13 (Jinquan et al., 2006).

β-glycosphingolipids as NKT Ligands

Thus far, α -anomeric D-glycosylceramides have not been detected in mammals. However, several recent studies have suggested that endogenous β-structured glycosphingolipids may be potent NKT ligands (Lalazar et al., 2006b). β-Structured glycosphingolipids are normal constituents of cell membranes (Goni and Alonso, 2006; Sonnino et al., 2007). β -Glucosyl-ceramide (β -GC) is a naturally occurring glycolipid that is a metabolic intermediate in glyco-sphingolipid anabolic and catabolic pathways. Its synthesis from ceramide is catalyzed by glucosylceramide synthase (Lalazar et al., 2006b). There is circumstantial evidence pointing to β-GC involvement in NKT lymphocyte regulation from studies of patients with Gaucher's disease, which is the most common lysosomal storage disease.

In this disease, decreased activity of glucosylceramide synthase results in elevated serum β-GC levels (Elstein et al., 2001). Interestingly, patients with Gaucher's disease have altered humoral and cellular immune profiles, including altered NKT cell number and function. Furthermore, findings in patients with Gaucher disease suggest a direct effect of β-GalCer on cellular membranes. Some patients have increased red blood cell aggregation (Adar et al., 2006) due in part to changes in cellular membrane properties.

Following binding to CD1d, β-GC has two equally efficient pathways of action: a direct pathway, resulting from β-GC binding and presentation, and a second pathway, resulting from the inhibition of α -GalCer-mediated activation (Ortaldo et al., 2004). In vitro, CD1d-bound β-GC inhibits NKT lymphocyte activation by α -GalCer (Ortaldo et al., 2004). Glucosylceramide synthase deficiency leads to defective ligand presentation by CD1d, thereby inhibiting NKT activation. β -GalCer-deficient mice exhibit normal NKT development and function, and cells from these animals can stimulate NKT hybridomas (Stanic et al., 2003a). In striking contrast, the same hybridomas fail to react to CD1d1 expressed by a β -D-glucosylceramide (β -D-GlcCer)-deficient cell line. Human β -D-GlcCer synthase cDNA transfer restores recognition of the mutant cells expressing CD1d1 by Vα 14Jα 18 NKT hybridomas.

Suppression of β -D-GlcCer synthesis inhibits antigen presentation to iNKT cells; however, β -D-GlcCer itself is unable to activate NKT hybridomas (Stanic et al., 2003a). β -D-GalCer (C12) efficiently diminished the number of detectable NKT lymphocytes in vivo without inducing cytokine expression. Binding studies have demonstrated that both α-GalCer and β -D-GalCer are equally efficient in reducing the number of NKT lymphocytes. However, in contrast to α -GalCer, β -D-GalCer (C12) is a poor inducer of IFNγ, TNFα, GM-CSF, and IL-4 gene expression (Ortaldo et al., 2004).

In addition to CD1d-binding mediated mechanisms, it has been proposed that β-glycosphingolipids may affect NKT activation via alteration of cell membrane properties, specifically those of lipid rafts (Lalazar et al., 2006a). Lipid rafts are highly dynamic submicroscopic assemblies enriched in sphingolipids and cholesterol. Alteration of raft properties may impair raft receptor localization without necessarily inhibiting ligand-receptor binding (Sehgal, 2003). Raft disruption has been shown to inhibit IL-6/STAT3 and IFNγ/STAT1 signaling (Sehgal, 2003). CD1d is also localized in lipid rafts (Park et al., 2005), thus disruption of lipid rafts can inhibit NKT activation without impairing CD1d-ligand binding (Park et al., 2005). The administration of naturally occurring β-glycosphingolipids can alter lipid raft composition and structure, thereby affecting intracellular signaling machinery (Stanic et al., 2003b; Ortaldo et al., 2004). It has been reported that β -GalCer (C12) stimulates NKT lymphocytes (Parekh et al., 2004). β -Glycosphingolipids seem to affect NKT lymphocytes differently than α -GalCer, inhibiting NKT lymphocyte proliferation without stimulating cytokine expression (Margalit et al., 2005).

As summarized next, the effects of β-glycosphingolipids in immune-mediated disorders has been studied in several animal models:

1. Immune-mediated hepatitis: β -GalCer alleviated ConA induced hepatitis in mice, an effect associated with decreased serum IFNγ levels and reduced expression of the transcription factor STAT1 (Margalit et al., 2005). ConA induces NKT cell-mediated liver damage. Serum AST and ALT levels were markedly reduced in GC-treated mice compared with non-treated animals, and histological damage was markedly attenuated. The beneficial effect of GC was associated with a 20% decrease of intrahepatic NKT lymphocytes, significant lowering of serum IFNγ levels, and decreased STAT1 and STAT6 expression. Intraperitoneal administration of radioactive GC was detected in the liver and bowel (Margalit et al., 2005). In vitro administration of GC led to a 42% decrease of NKT lymphocyte proliferation in the presence of dendritic cells, but this did not occur in their absence.

2. Immune-mediated colitis: In a murine T_H1-mediated colitis model, β-glucosyl-ceramide generated a T_H2 response associated with colitis alleviation (Zigmond et al., 2007).

3. Hepatocellular carcinoma: In mice with hepatocellular carcinoma (HCC), β-GalCer administration suppressed tumor growth and improved survival associated with a T_H1 immune shift (Zigmond et al., 2007). The immunological effect of β-glucosylceramide was assessed by analysis of intrahepatic and intrasplenic lymphocyte populations, serum cytokines and STAT protein expression in the colitis and hepatoma models. Administration of β-glucosylceramide led to alleviation of colitis and to suppression of HCC, manifested by improved survival and decreased tumor volume. These beneficial effects were associated with an opposing immunological effect in the two models: the peripheral:intrahepatic CD4:CD8 lymphocyte ratio increased in the colitis model, while it decreased in the HCC group (Zigmond et al., 2007). The effect of β-glucosylceramide was associated with decreased STAT1 and -4 expressions, and with over-expression of STAT6, along with decreased IFNγ levels in the colitis model, whereas the opposite effect was noted in the HCC model. Taken together, these data suggest that β -glucosylceramide alleviates immunologically incongruous disorders and may be associated with “fine tuning” of immune responses, by changes in plasticity of NKT lymphocytes (Zigmond et al., 2007).

4. Graft versus host disease: β-GalCer treatment was effective in alleviating semi-allogeneic acute and chronic graft versus host disease (GVHD) in mice (Ilan et al., 2007). Acute and chronic GVHD were generated by fusion of splenocytes from C57BL/6 to (C57BL/6xBalb/c) F1 mice, and from B10.D2 donor mice into Balb/c, respectively. Recipients were treated daily with GC. Histological and immunological parameters of GVHD were assessed (Ilan et al., 2007). Treatment with GC significantly alleviated GVHD in both models. The beneficial effect of GC was associated with an increase in the intrahepatic:peripheral NKT lymphocyte ratio in the semi-allogeneic/acute model. However, this ratio was decreased in the chronic GVHD model. A significant increase in intrahepatic CD8 lymphocyte trapping was noted in the semi-allogeneic/acute model, but we observed the opposite in the chronic GVHD model. Administration of GC led to decreased serum IFNγ, but increased serum IL-4 levels in the T_H1-mediated model; whereas in the T_H2-mediated models, the opposite was true. These data suggest that GC ameliorates GVHD in both T_H1- and T_H2-mediated murine models, suggesting it is able to differentially affect the immune system (Ilan et al., 2007).

5. Liver fibrosis: To explore the role of NKT cells in hepatic fibrosis via GC, hepatic fibrosis was induced in male C57Bl/6 mice by bi-weekly intraperitoneal (IP) carbon tetrachloride (CCl₄) administrations for 7 wk (Safadi et al., 2007). Mice were treated with daily IP GC injections. Marked amelioration of hepatic fibrosis was observed in all GC-treated mice without altering the production of reactive oxygen species. As determined by Sirius red-stained liver tissue sections and measured by Bioquant(R) morphometry; all CCl₄-administered groups presented significantly increased fibrosis area relative to naive animals. Western blot analysis for smooth muscle α -actin from liver extracts followed a similar pattern. A significant decrease in liver damage was observed in all GC-treated groups, as noted by a decrease in transaminase serum levels. The beneficial effect of GC was associated with a significant decrease in the intra-hepatic NKT and CD8 lymphocytes, as well as the attenuation of both T_H1 and T_H2 cytokines (Safadi et al., 2007).

One of the most intriguing characteristics of NKT lymphocytes is their plasticity. This subset of lymphocytes generates both T_H1 and T_H2 responses upon activation; however, the mechanisms and consequences of NKT plasticity remain under active investigation. Firstly, it is unclear whether or not the plasticity is mediated by a ligand. Different immune responses are generally associated with typical mediators; however, the same ligand can generate different types of immune responses in different immune microenvironments. In light of the differing effects of a given ligand in vivo and in vitro, the net effect of NKT activation may not result from the binding of a single ligand, but result rather from the sum of the effects of a variety of mediators. Furthermore, organ-specific factors may play a role in NKT plasticity, with different responses generated in different organs by an identical stimulus.

Alternatively, originally identical stimuli may reach NKT lymphocytes via different pathways through presentation by different APC (Godfrey and Kronenberg, 2004; Ilan et al., 2007). Furthermore, NKT lymphocytes are a heterogeneous population of cells that differ from one another in their CD1d reactivity and CD expression, contributing to plasticity. Apart from inherent heterogeneity between different NKT populations, changes in cellular membranes with altered lipid raft properties will affect raft-bound receptors such as CD1d and may add to the variety of responses. NKT plasticity may also be an expression of different immunologic reaction profiles, dictated by genetic heterogeneity. Thus NKT plasticity may likely be considered the result of an amalgam of several of the factors mentioned above, with CD1d ligands being the final link in a chain of factors determining the final response profile.

The overall objective of ongoing research in this field is to determine the NKT ligand-dependent signaling pathways in order to optimize the use of different ligands as putative immunomodulatory agents. Understanding the relationship between NKT intracellular signaling pathways and ligand structure, and the dependency on the CD1d system and specific NKT raft microdomains will contribute to the development of NKT-based immunotherapy.

This work was supported in part by a grant from The Roman-Epstein Liver Research Foundation (to Y. I.).