|Home | About | Journals | Submit | Contact Us | Français|
CD1 proteins display lipid antigens to T cell receptors. Studies using CD1d tetramers and CD1d-deficient mice provide important insight into the immunological functions of invariant NK T cells (iNKT) during viral and bacterial infections. However, the mouse CD1 locus is atypical because it encodes only CD1d, whereas most mammalian species have retained many CD1 genes. Viewed from the perspective that CD1 is a diverse gene family that activates several of classes of T cells, new insights into lipid loading and infection response are emerging.
CD1 antigen presentation was discovered using human T cells that recognize CD1a, CD1b or CD1c proteins [1,2]. Separately, a distinct population of CD3+ cells that persist in MHC knockout mice were designated “invariant NK T cells” (iNKT) based on their (nearly) invariant TCR Vα 14 chains and Vβ chains, as well as natural killer (NK) locus encoded markers [3,4]. Later, mouse iNKT cells were found to recognize CD1d, so that the previously separate fields of CD1 and NKT cells merged . Human T cells with Vα24 TCRs were found to have the same molecular recognition properties as Vα14 mouse iNKT cells [6,7] as well as a shared lineage specific transcription factor, promyelocytic leukemia zinc finger (PLZF) . In addition, certain mouse and human T cells recognizing CD1d were found to lack the conserved TCRs  and antigen reactivity [10,11], that normally characterize iNKT, so that the NKT definition was expanded to include also “diverse” NK T cells.
Here we review recent advances in understanding the role of these various populations of CD1-reactive T cells during infection. Increasingly, differences in the cellular expression patterns, subcellular trafficking, antigen binding grooves and phenotypes of the responding T cells make the case that CD1a, CD1b, CD1c, CD1d and CD1e proteins have distinct functions. Further, new insights into non-human CD1 genes show that CD1 gene families are large and vary from species to species. These studies emphasize that CD1-restricted T cells and NK T cells are not synonymous, and make the case that understanding the functions of CD1 involves looking at and beyond NKT.
Many crystal structures of CD1 proteins bound to lipid antigens show that the alkyl chains are inserted into a hydrophobic groove, allowing presentation of carbohydrate, peptidic or inorganic components of amphipathic antigens . Recent studies of ternary complexes show how the T cell receptor α and β chains of iNKT contact the CD1-glycolipid complex to form a binding footprint [13,14]. The NKT footprint is quite different from that of TCRs contacting peptide-MHC . The iNKT TCRs are rotated and pushed laterally so that the α chain binds near the center of CD1d, and the TCR β chain makes limited contact at the margin of CD1d.
These new crystal structures explain in detail why certain Vα and Vβ chains are conserved in natural iNKT populations. The CDR3α loop plays the dominant role in binding to the CD1d platform, and the direct contacts with the protruding galactose unit are mediated by Jα18 residues. Based on mutational studies , molecular models  and other data , the global orientation of the TCR and other aspects of the recognition event visualized in these crystal structures are also likely conserved for natural α-linked glycolipid antigens (Fig. 1) [17-19]. Whether this rotated and laterally displaced footprint is used by diverse TCRs that recognize the glycolipid, lipid and lipopeptide antigens presented by CD1a, CD1b, CD1c or CD1d, remains to be seen.
Unlike CD1a, CD1b and CD1c, the CD1d protein is expressed in the liver and on certain gastrointestinal epithelia. Recent studies implicate CD1d and iNKT cells in controlling bacterial colonization of the gastrointestinal tract of mice . Small intestinal colonization with both gram-negative and gram-positive organisms was increased in CD1d knockout mice, and organisms translocated across the intestinal epithelium. NKT cells triggered CD1d-expressing Paneth cells to secrete antimicrobial peptides .
Invariant NKT cells respond to Borrelia burgdorferi, the causative agent of Lyme disease. Mice that are usually resistant to infection become more susceptible when the CD1d gene is deleted , and levels of protective Borrelia-specific IgM are reduced . An antigenic target of the response was identified as the B. burgdorferi glycolipid II (BbGL-II), an α-galactosyl diacylglycerol that constitutes 12 percent of lipid in this pathogen . This antigen has obvious structural homology to α-galactosyl ceramide (Fig. 1), and BbGL-II loaded CD1d tetramers stain liver NKT cells during infection, indicating that the molecular mechanism of activation involves CD1-glycolipid-TCR contact. Infection of Jα-deficient mice with B. burgdorferi resulted in prolonged arthritis and bacterial persistence, raising the possibility that lipid recognition by iNKT is relevant to a chronic syndrome . Lastly, the recognition of borrelia antigens by mouse NK T cells may be relevant to human Lyme disease because human NKT cells recognize a variant of BbGL-II , and unpublished studies have identified CD1 gene expression in human skin affected by acute borrelial infection (Yakimchuk and Moody, unpublished).
For CD1a, CD1b and CD1c proteins, the most extensively studied bacterial pathogens are M. tuberculosis and M. leprae. Following the discovery of free mycolic acid  and glucose monomycolate antigens , glycerol monomycolate isolated from Mycobacterium bovis was recently found to stimulate a human CD4+ T cell clone (Fig. 1) . Additionally, polyclonal mononuclear cells from humans latently infected with M. tuberculosis produced IFN-γ in response to glycerol monomycolate at a higher frequency than cells from non-infected controls or actively infected tuberculosis patients. Along with studies of mannosyl phosphomycoketides, mycolic acids, glucose monomycolates and sulfated trehalose lipids [27-29], this patient study supports the hypothesis that tuberculosis infection promotes expansion of human lipid reactive T cells in vivo. However, whether or not such responses are durable and subject to recall, such that vaccination might provide protection from infection, remains unknown.
NKT cells respond to viral infections involving HIV  HSV  and influenza . These new observations raise the question of whether the mechanism of virus recognition involves cognate recognition of a virally produced antigen by the TCR, indirect recognition of cellular changes induced by viruses, or both. To date, no virally-derived CD1 ligands have been identified. However, new evidence shows that CD1c presents an N-terminally acylated lipopeptide similar in sequence to HIV nuclear envelope factor (Nef) . This finding supports the hypothesis that cellular lipidation of viral proteins may generate antigens presented by CD1 . In addition, viruses trigger Toll-receptors and cause other cellular changes in ways that can activate NK T cells indirectly via IL-12, altered CD1d expression, or increased production of endogenous sphingolipids [18,35,36]. For example, TLR ligation affects glycosphingolipid biosynthesis by dendritic cells and is associated with increased IFN-γ release by NKT cells [37,38].
CD1d is constitutively expressed on thymocytes, B cells, monocytes, macrophages and on myeloid dendritic cells (DC) at various stages of maturation. In contrast, CD1a, CD1b and CD1c proteins are absent on blood monocytes in the circulation, and two new studies help explain why. Serum immunoglobulin (Ig) and activators of the peroxisome proliferator activator receptor-γ (PPAR-γ) are present in the serum and tonically inhibit CD1a, CD1b and CD1c on human monocytes [39,40]. A human patient with common variable immunoglobulin deficiency expressed CD1a, CD1b and CD1c, and this expression was down-regulated after restoring immunoglobulin (Ig) to physiologic levels suggesting that Ig is necessary and sufficient for control of CD1 expression on circulating monocytes .
When monocytes exit the circulation, they are presumably released from inhibitory signals found at high concentrations in the serum, and they also encounter stimuli that increase CD1a, CD1b or CD1c gene expression in tissues, as seen in patients with autoimmune disease  or infection . The localized upregulation of CD1a, CD1b and CD1c proteins on maturing myeloid DCs at sites of inflammation may allow CD1-expressing DCs and CD1-restricted T cells to generate pro-inflammatory positive feedback loops . These CD1-inducing signals involve GM-CSF, IL-4, Toll-like receptor (TLR) 2 and TLR 5 [44,45]. Mycobacteria produce both ligands for CD1 proteins and signals that induce CD1, so they might provide dual signals to promote CD1-restricted T cell activation at the site of infection .
Several studies of monocyte derived DCs have found that CD1a-, CD1b and CD1c-expressing cells decline in number after exposure to mycobacteria in culture [47-50]. These in vitro studies led to the speculation that drastic losses of CD1 expression might occur at the site of mycobacterial infection in vivo and might represent a physiological means of immune evasion. However, other in vitro studies failed to confirm CD1 down-regulation [44,51]. More importantly, studies of CD1 expression in the lungs, lymphoid tissues and skin of humans with tuberculosis and leprosy do not support the immune evasion hypothesis because CD1a, CD1b and CD1c expressing cells are found at high levels at sites of infection [42,45,52]. Although mycobacteria do not prevent CD1 expression in a general way in all humans, a subset of humans with the lepromatous form of leprosy have reduced levels of CD1 expression at the site of infection [42,53].
Viral infection also downregulates cell-surface expression of CD1. The HIV peptide Nef interacts with human CD1d, leading to decreased expression on the cell surface and diminished activation of CD1d-restricted NKT cells . Kaposi sarcoma-associated herpesvirus and Herpes Simplex Virus 1 downregulate CD1 surface expression using distinct mechanisms involving ubiquitination and lysosomal targeting, respectively [31,55]. The detailed molecular mechanisms of rerouting identified here, as well as the precedent of virally mediated MHC class I immunoevasion, now provide a rationale to examine CD1d expression during in vivo infection with viruses.
T cells recognizing human CD1a, CD1b and CD1c or their mammalian orthologs do not fall under historical or modern definitions of NK T cells because they are not known to commonly express NK receptors or invariant TCRs and do not recognize CD1d (Fig. 2). Lacking a catchy jargon term like iNKT, they are designated according to a simple, descriptive and accurate naming convention: CD1x-restricted T cells, where x is the identifying CD1 gene (Fig. 2). CD1-restricted T cells have functions that are distinct from iNKT cells because they express diverse TCRs, present chemically diverse antigens and recognize different types of cells (Fig. 2). Also, the study of gene induction patterns on myeloid DCs makes clear that CD1a, CD1b, CD1c and CD1e are linked to one another, whereas CD1d is different [44,56].
Further, each of the five CD1 human proteins is emerging to have a distinct personality (Fig. 2). Such gene specific functions can be most readily understood for CD1e. After exiting the endoplasmic reticulum to the golgi apparatus, CD1e is diverted directly to endosomes without evidence of expression at the surface, suggesting that CD1e does not display antigens at the cell surface . Recent studies show that unlike other CD1 proteins, CD1e is released into the lumen by proteolytic cleavage , where it can float freely and promote the molecular trimming of phosphatidylinositol antigens and their subsequent presentation by CD1b [59,60].
CD1b is emerging as the CD1 isoform that focuses on presenting large, exogenous foreign antigens that are taken up into lysosomes. With an interior volume of approximately 2300 cubic angstroms, the CD1b groove is much larger than that in CD1d nearly twice the volume of the CD1a groove . Correspondingly, the polyacylated trehaloses and mycolates, including the new glycerol monomycolate antigen, are lipids in the size range of C70-80, much larger than the C18-48 lipids presented by other CD1 isoforms. In fact, the longest C84-86 mycobacterial mycolates exceed the predicted volume of the CD1b groove and may protrude through a small opening at the bottom of the groove in the C′ pocket . The insertion of such large lipids into CD1b may be more dependent on lipid transfer proteins and acid-mediated steric changes than seen for other CD1 proteins [59,63,64]. CD1a and CD1c proteins show fewer requirements for acid-mediated loading and less prominently accumulate in the most acidic lysosomal compartments [65,66]. These biophysical properties of CD1b suggest that it is specialized to capture exogenous long chain foreign lipids in preference to shorter self phosphodiacylglycerols, sphingolipids and other self lipids that comprise mammalian membranes (Fig 2). Correspondingly, T cells autoreactive to CD1b have been less frequently observed than those directly recognizing CD1a or CD1c [1,41,67,68].
The discovery of an avian CD1 gene [69,70] and new evidence that it folds to form an antigen binding pocket  proves that the CD1 system predates the emergence of mammals. However, unlike classical MHC class I molecules, which are present in all jawed vertebrates including fish, CD1 has not been identified in fish . Also, recent studies suggest that CD1d proteins and NKT cells are apparently lacking in ruminants [73,74]. Figure 2 illustrates how modern species have survived while lacking any one of the five CD1 gene types. On the other hand, most mammalian species have preserved large gene families, some with up to 14 CD1 genes. Also, no mammalian species lacking all CD1 proteins has been identified since the discovery of the CD1 locus more than twenty years ago, implying that CD1 has an indispensable role in the mammalian immune system .
Thus, it appears CD1d and NK T cells per se are not universally conserved, but instead that the CD1 family is represented in some form in all amniote species. If all mammalian species express at least one CD1 protein, this implies that CD1 proteins have important immunological functions that were positively selected by evolutionary forces. Because one of the main functions of CD1 proteins is to present lipid antigens from pathogens, we speculate that the size and composition of CD1 genes present in any given species reflects the results of pathogen exposure and selection pressure on an evolutionary time scale.
We thank Tan-Yun Cheng for advice and lipid antigen graphics. This work was supported by NIH AI071155 and AI049313, the Wellcome Trust GR078283, the Burroughs Wellcome Fund and the Howard Hughes Medical Institute Kwa-Zulu Natal Research Institute for Tuberculosis and HIV (K-RITH).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
**Borg in Nature
The ternary structure of a human NKT TCR bound CD1d and α-galactosyl-ceramide explains why NK T cells naturally express certain conserved Vα and Vβ chains: the same chains present in natural iNKT mediate the contacts with antigen and CD1d. In contrast to observed interactions between human TCR and peptide-MHC, docking of NKT TCR and CD1d is parallel as opposed to diagonal in orientation.
*Van Rhijn in JEM
These data show that human CD1c presents a lipopeptide antigen to T cells and that recognition is specific for peptide sequence. Whereas most studies focus on glycolipids, the discovery of lipopeptide antigens provides possible links to viral or self antigens made through protein lipidation reactions.
**Savage in Immunity
The transcription factor promyelocytic leukemia zinc finger (PLZF) controls development of the NKT cell lineage in mice and was lacking in other T cell populations, indicating that it is a lineage specific transcription factor. Human NKT cells express PLZF mRNA, suggesting that PLZF may play a role in human NKT cell development.
*Layre in Chemistry and Biology
These studies provide evidence for a third type of mycolate antigen for CD1b. Interestingly, people latently infected with M. tuberculosis produced cytokines in response to glycerol monomycolate at levels higher than seen in patients with active tuberculosis. These findings implicate glycerol monomycolate specific T cells in the human immune response to M. tb and further enhance the argument that this isoform binds particularly large lipids.