The epochal discovery of a technique for producing mAbs of defined specificity by Georges Köhler and César Milstein (1) led to many important advances in life sciences and spawned a large industry for production of these Abs for research, diagnosis, and therapy. The first mouse anti-human mAb produced by this technique was NA1/34. It recognized an Ag called HTA-1 (later called CD1a), expressed on human thymocytes and some B lymphoma cell lines (2). Because this and related proteins were among the first human lymphocyte markers identified with the newly invented mAb technology, they entered the immunological lexicon as the first cluster of differentiation molecules or “CD1.” Genomic and sequence analysis established that this protein and four paralogues on human chromosome 1 encoded closely related proteins called CD1a, CD1b, CD1c, CD1d, and CD1e, which were distantly related to the H chains of human MHC class I proteins HLA-A, -B, and -C encoded on chromosome 6 (3, 4). Moreover, these proteins were associated with β2-microglobulin, which is encoded on chromosome 15 and also found in HLA proteins. On the basis of sequence comparisons, the molecules could be divided into group 1, comprising CD1a, CD1b, CD1c, and CD1e, and group 2, containing only CD1d. Surprisingly, however, the mouse contains no group 1 CD1 genes but has two group 2 CD1d genes, whereas the rat contains only one CD1d, and other rodents and mammals may contain 8–12 group 1 CD1 genes. Why had these MHC class I-like genes with limited polymorphism evolved separately from their longer known polymorphic cousins?
Their function or functions were also obscure, but once again, simple but logical direct questions, followed by experiments, led to the answer. If the function of the polymorphic MHC class I proteins was to present peptides with different sequences to the immune system, was it possible that the CD1 proteins were also Ag-presenting molecules? Were there circulating T cells that would recognize CD1 molecules?
The answers were provided in three important papers (5–7), the first of which is reprinted here in Pillars of Immunology. Because CD1 proteins are distinct in sequence and structure from MHC proteins, T cells that recognize CD1 should not be MHC restricted. An unusual CD4− CD8− double negative (DN) TCRγδ T cell line IDP2 that showed MHC-independent lysis had been described. Among several dozen DN T cell lines that had been generated, equally divided among DN TCRαβ and TCRγδ lines, one additional DN TCRαβ cell line was found that also lysed MOLT4 cells that do not express any MHC proteins (5). However, MOLT4 was well known to express all three CD1a, CD1b, and CD1c molecules. mAb blocking experiments established that the TCRs were each involved in the lysis phenomenon. Moreover, the unusual DN TCRγδ cell line was specific for CD1a, whereas the DN TCRαβ cell line was specific for CD1c. Later, CD4+ (and some CD8+) T cells that recognize CD1 proteins were found, and the DN characteristic that led to the discovery diminished in importance.
The door was open. In two subsequent papers, it was established first that a microbial Ag could be presented by a CD1 molecule (6) and then that the foreign Ag was a lipid (7). PBLs were stimulated with an extract of Mycobacterium tuberculosis, and a cell line, DN1, was obtained that proliferated to the extract and to that of M. leprae, but not to several other organisms. The Ag extract was presented by CD1-expressing monocytes and was not MHC restricted. Moreover, the response was blocked by CD1b mAb, but not by CD1a or CD1c mAb, nor by pan-MHC class I or MHC class II mAbs. Further, only CD1b transfectants, not CD1a or CD1c transfectants, could present Ag in the extracts. Thus, in three short papers, novel T cell lines specific for CD1a, CD1b, or CD1c had been described. The Ag recognized by the DN1 clone was protease resistant and extractable into organic solvents. Reverse phase HPLC was used to show that the active component coeluted with mycolic acid, a lipid that forms the outer coat of mycobacteria. A new class of Ag-presenting proteins had been found for lipid Ags.
The floodgates were open, and a profusion of papers have appeared in the past 15 y (reviewed in Refs. 8–10). Important advances were made in four specific areas: identification of the lipids recognized; crystallization of several CD1 proteins, revealing the manner in which lipid moieties lead to binding of these Ags; the different trafficking pathways for CD1a, CD1b, and CD1c; and, most importantly, in vivo studies of the possible role or roles of these CD1 proteins in the pathogenesis of and/or protection from infection. The additional MHC-like proteins present on human chromosome 1, group 2 CD1d and MR1, are beyond the scope of this commentary.
A large number of lipids that bind to the three CD1 proteins have been identified. Although these molecules have been referred to as “lipid Ags,” this label seems a slight misnomer because the lipid moieties themselves embed these molecules to the CD1 proteins, whereas hydrophilic head groups of the Ags are recognized by T cells. The lipids are mainly glycolipids and lipopeptides, and only in rare cases, like mycolic acid, is the lipid itself recognized. “Lipid-linked Ags” would be a more precise term. Many of the lipids recognized come from important human pathogens such as M. tuberculosis, M. leprae, Leishmania, and Borrelia burgdorferi (the causative agent of Lyme disease). In mycobacteria, for example, up to 40% of the cell wall may be composed of lipids.
CD1a and CD1b were crystallized. The resulting structures demonstrated different ways that lipids are bound to these proteins. Both use hydrophobic pockets called A′ and F′ to bind lipids (related in position and function to the A and F pockets in MHC class I proteins that bind amino acid side chains). CD1b also has a C′ pocket that mainly provides a portal to the exterior and a T′ (for tunnel) pocket, features that may allow accommodation of very large lipids (11–13). A computational model of the CD1c structure has recently appeared. It is distinguished from its close cousins by the size of the F′ pocket; the absence of pockets C′ and T′; and the possible presence of a new portal, D′ (14).
The distinctive trafficking pathways of the group 1 CD1 proteins that lead to loading of the lipid-linked Ags are most interesting (10). After biosynthesis in the endoplasmic reticulum and Golgi, where binding of self-lipids occurs, these molecules first travel to the cell surface. Direct loading of lipids may occur at the surface, but in addition these proteins, although they resemble MHC class I proteins structurally, traffic through the endosomal compartments, where loading and/or exchange with exogenous lipid-linked Ags occurs, in a manner similar to the MHC class II proteins. The three group 1 CD1 proteins have distinct trafficking pathways. CD1a, which has only a short cytoplasmic tail, remains largely at the surface and follows the shallow endosomal recycling pathway of MHC class I. By contrast, CD1b and CD1c have longer tails encoding endosomal sorting motifs that use AP2 and AP3 as trafficking carriers to late endosomal compartments. CD1c appears to traffic broadly to both early and late endosomes (lysosomes), whereas CD1b mainly traffics both to the same endosomes in which peptide loading on MHC class II proteins occurs and more deeply to late endosomes. In these pathways, they encounter lipid-linked Ags derived from pathogens in phagolysosomes. Lipid loading presumably requires some form of assistance. Saposins and apolipoproteins (15, 16) have been suggested to play an important role in this process, but in addition the fourth group 1 CD1 molecule, CD1e, may also have a significant part in lipid loading (17, 18). Interestingly, loading of peptides onto MHC class II molecules requires the assistance of the class II-like protein HLA-DM. Even the classical MHC class I molecules have among them HLA-F that resides in the endoplasmic reticulum and has no known function. Could it also, in some circumstances, be involved in loading of peptides onto MHC I proteins?
After loading lipid-linked Ags, CD1b and CD1c molecules return to the cell surface, where they can be recognized by T cells. Thus a new system for recognizing lipid-linked Ags was described in addition to the longer known peptide recognition system involving MHC class I and MHC class II proteins.
However, important work often ends in important questions. What is the role of CD1 proteins in the pathogenesis of and/or protection from infection? Can the information gained be used to develop vaccines against important pathogens, particularly M. tuberculosis, which causes worldwide morbidity and mortality nearly equal to that of HIV and has been a plague at least since the time of the pharaohs? Expansion of CD1-restricted T cells and secretion of IFN-γ has been clearly demonstrated in infections, particularly those with M. tuberculosis (19, 20). At this point, though, data indicating that these T cell responses are protective are still lacking. Microorganisms also have been reported to be able to either upregulate or downregulate expression of CD1 proteins on monocytes (10, 21, 22) (K. Yakimchuk, C. Roura-Mir, T.-Y. Cheng, S. R. Granter, R. Budd, A. Steere, V. Pena-Cruz, and D. B. Moody, submitted for publication). Because the mouse lacks group 1 CD1 proteins, a mouse model has not been available. However, the recent construction of a mouse transgenic for human CD1a, CD1b, and CD1c may make it possible to design an experiment in which protection can be demonstrated (23). Similarly few data on vaccination with lipid-linked Ags are available (24–27). Of particular interest is a recent study of cattle immunized with mycobacterial glucose monomycolate (27). Cattle express CD1a, CD1b, and CD1c, but no CD1d. In this species, infections with M. bovis and M. avium subsp. paratuberculosis cause important economic problems. Strong T cell responses to immunization were generated. Much work in this area remains to be done, but the basic scientific foundations have been laid.
Disclosures The author has no financial conflicts of interest.
This work was supported by National Institutes of Health Research Grant AI049524.
Abbreviation used in this paper: