CD1d is able to bind and present a diverse array of foreign and self-lipids, some of which can stimulate an NKT cell response. For example, iNKT cell receptors can be stimulated by CD1d in complex with mycobacterial phosphatidylinositol mannosides (Fischer et al., 2004); microbial cell wall glycuronosylceramide antigens such as a-galacturonosylceramide and analogues thereof (Kinjo et al., 2005; Mattner et al., 2005; Sriram et al., 2005; Wu et al., 2005; Wu et al., 2006); glycuronosylceramide homologues such as a-galactosylceramide and analogues thereof; the tumor derived disialoganglioside GD3 (Wu et al., 2003); a diacylglycerol glycolipid from Borella burgdorferi (Kinjo et al., 2006); the self-lipid isoglobotrihexosylceramide (iGb3) (Zhou et al., 2004) and the self glycerophospholipid phosphatidylcholine (PC) (Giabbai et al., 2005). Although iNKT cell receptors are able to recognise diverse ligands presented by CD1d they are able to selectively discriminate between analogues that have slight structural modifications (Goff et al., 2004; Wu et al., 2006; Yu et al., 2005; Zajonc et al., 2005a). This raises the questions: How can iNKT cell receptors with a semi-invariant TcR recognise diverse lipid antigens with structurally distinct head groups? How do iNKT cell receptors differentiate between structurally similar ligands?.
The three-dimensional structure of mouse or human CD1d has been determined in complex with various self and foreign lipids. The self-glycolipid/CD1d complexes include, the glycerophospholipid phosphatidylcholine (PC) (Giabbai et al., 2005) and the glycosphingolipid 3' sulfogalactosylceramide (sulfatide) (Zajonc et al., 2005b). The foreign glycolipids include 1) a-galactosylceramide (a-GalCer), a glycosphingolipid from the marine sponge Agelas mauritanius (see Fig. 2a) (Koch et al., 2005); 2) PBS-25, a short-chain variant of a-GalCer (see Fig. 2b) (Zajonc et al., 2005a); 3) a-glucuronosylceramide (GalA-GSL), a microbial glycosphingolipid from Sphingomonas (see Fig. 2c) (Wu et al., 2006); 4) the synthetic glycerophospholipid dipalmitoyl-PIM2 (see Fig. 2d) (Zajonc et al., 2006). The growing number of structures provide invaluable insight into 1) the nature of the antigen-binding cavity 2) the common features of glycolipids that can bind to CD1d 3) generalities in the mode of binding of foreign glycolipids in the antigen-binding cavity and 4) CD1d residues likely to interact with the NKT cell receptor. Some general trends have been observed:
1) The ligands are composed of two long hydrophobic acyl chains, one bound in the A' and one bound in the F' pocket of the antigen-binding cavity. The larger A' pocket can accommodate up to 26 carbons (Koch et al., 2005), however shorter length acyl chains can still be accommodated and are often stabilised by the binding of spacer-lipids in the A' pocket (Wu et al., 2006; Zajonc et al., 2005a).
2) The acyl chain in the A' pocket has been reported as being either blocked (Zajonc et al., 2006) or accessible (Zajonc et al., 2005b) to TcR access depending on the glycolipid that is presented. The F' pocket of the antigen-binding cavity is optimal for an 18-carbon length chain (Koch et al., 2005) and accommodates the sphingoid base of a glycosphingolipid. This pocket is capped in one structure such that the sphingosine tail is denied access to the CD1d surface (Zajonc et al., 2005a).
3) As yet there is no consensus as to whether the sn1- and sn2-fatty acids (eg. see Fig. 2d) of the glycerophospholipids will bind in the A' or the F' pocket (Giabbai et al., 2005; Zajonc et al., 2006). However, the glycerolipid tails of the glycerophospholipids do appear to be more flexible than the acyl tails of the glycosphingolipids (Zajonc et al., 2006).
4) The hydrophobic tail of the glycolipid is linked via an a- or p-anomeric link to a hydrophilic head group. The composition of the head group is a single or multiple-linked sugar moiety with or without modifications. The head group of the glycolipid protrudes out of the antigen-binding cleft where it is available for TcR recognition (Koch et al., 2005; Wu et al., 2006; Zajonc et al., 2006; Zajonc et al., 2005a; Zajonc et al., 2005b).
5) Stabilising interactions are formed between residue Asp80 on the a1 helix of CD1d and the 3'-hydroxyl and in some cases the 4'-hydroxyl groups of the sphingoid base (Koch et al., 2005; Wu et al., 2006; Zajonc et al., 2005a) of a glycosphingolipid. Glycosphingolipids with a sphinganine backbone, such as GalA-GSL, lack the 4'-hydroxyl and only make contact with CD1d through the 3'-hydroxyl of the sphinganine backbone. The fewer contacts lead to a less rigid head group conformation (Wu et al., 2006) which may also alter the conformation of TcR exposed residues like Arg79 on the a1 helix of CD1d. The extent of these interactions also subsequently determines the depth that the glycolipid sits in the groove as well as the lateral disposition, orientation and the rigidity of the sugar moiety (Wu et al., 2006).
6) An aspartate on the a2 helix of CD1d (D151 in human CD1d; D153 in mouse CD1d) forms hydrogen bonds with the 2'-hydroxyl and/or the 3'-hydroxyl of the sugar moiety (Koch et al., 2005; Wu et al., 2006; Zajonc et al., 2005a). The configuration of the 2'-hydroxyl is correlated with the lack of stimulatory activity of a-mannosylceramide (a-ManCer) compared to a-GalCer, therefore such subtle differences are crucial to TcR recognition (Kawano et al., 1997).
7) The substitution of CD1d residues (R79, D80 and D153) involved in hydrogen bonds with the short chain a-GalCer ligand resulted in abolished NKT cell stimulation, most likely because the conformation of the sugar moiety was severely disrupted (Kamada et al., 2001).
8) The importance of the conformation of the sugar moiety on NKT cell recognition is reinforced by the inability (Kawano et al., 1997) or minimal ability (Parekh et al., 2004) of p-galactosyl-ceramide (P-GalCer) to induce NKT cell proliferation and cytokine release (Ortaldo et al., 2004; Parekh et al., 2004) compared to the highly potent a-GalCer. p-GalCer is predicted to make fewer contacts with the a2 helix of CD1d such that the sugar moiety adopts an alternate perpendicular orientation to the CD1d binding groove (Zajonc et al., 2005a), once again influencing TcR recognition and cytokine release.
9) It was previously considered that the length of the lipid tail in the binding groove had an impact on the antigenic potency. However, a-GalCer (C26) and the identical, but otherwise shorter tailed PBS-25 (C8) have identical antigenic potency (Brossay et al., 1998; Zajonc et al., 2005a). The cytokine response has been shown however to be altered for other a-GalCer analogues with both a shortened lipid tail and the introduction of unsaturations (Yu et al., 2005) and also with other truncated, saturated lipids (Goff et al., 2004; Miyamoto et al., 2001).
Was this article helpful?