In this paper, we show that CD1d is expressed on bronchial epithelial cells, both in primary and airway epithelial cell lines, and that there are at least six spliced variants likely specific to epithelial cells. We felt that examining the function of these epithelial CD1d transcripts was beyond the scope of this paper so it is unclear currently how these proteins function in the context of the lungs. However, it is possible to speculate on the potential implications of CD1d expression on these structural cells. Due to the large surface area of the airway epithelium, one prospect of CD1d expression here is the rapid presentation of both self and exogenous glycolipids to infiltrating NKT cells. This could maintain NKT cells proliferation in situ and prolong their protective effect in the lungs. The opposite is also possible – that of enhancing NKT cell's harmful effect e.g. airway hyper-responsiveness observed in murine models of asthma. In the ovalbumin asthma model, CD1d−/−
mice do not develop airway hyper-responsiveness 
A potentially interesting part of our results is the observation of the same profile of transcripts in epithelial but not haematopoietic cells. The epithelial cell lines from skin, lungs, cervix and colon showed “V2” to “V6” expression (“V2” and “V3” very faint in the cervical epithelial cell line, TZM-bl). The exception to this is A549, a type II alveolar epithelial cell line which only expressed “V4” and “V5”. Type II alveolar epithelial cells have the unique property of producing surfactants, which regulate the surface tension of the alveoli and also to contribute to host defence. In the latter, two subsets of surfactant (SP-A and SP-D, part of collectins family) act as opsonins to facilitate elimination of pathogens by alveolar macrophages 
. The CD1D transcripts on these cells (“V4” and “V5”, discussed later as potential ß2m-independent and soluble forms respectively) could serve as innate host defence molecules rather than antigen-presenting molecules.
Against expectation, we observed only very low/undetectable levels on full length mRNA in PBMC and DCs (). There have been very few studies on the variety of CD1D mRNAs in these cells and their relative expression to the full-length transcript. Kojo et al mentioned existence of multiple CD1D transcripts in PBMCs, however, the level of expression was so low that they required nested PCR to identify this; only three healthy volunteers were studied, among these, one had extremely low/undetectable levels of non-spliced full-length CD1D. Our low expression is in keeping with their finding, and the low/undetectable expression by one-round of PCR in MOLT-4 and THP cells support the finding in primary hematopoietic cells Since the CD1d protein has been identified in these cell populations, one possibility is that the mAbs used detected proteins translated by the variants rather than the full length CD1D.
Some potential functions of the different variants could be gleaned from knowledge of the various regions that are spliced (). Lipid-binding groove in CD1d is formed by α1 and α2 domains 
. The protein for “V1”, deficient in α1, is unlikely to participate in antigen loading and presentation. Hydrophobic amino acid residues encoded by transmembrane (TM) exon are required for insertion of the CD1d protein into plasma membrane. The TM exon is spliced out in “V2” and “V5”. In addition, “V3” and “V6” as a consequence of shift in reading frames contain early stop codons in α3 and TM, respectively. So, it is likely that proteins coded by these variants will be either secreted (soluble isoform) or retained intracellularly. There is precedence for this - Woolfson et al
showed expression of CD1A and CD1C TM-deficient transcripts, and detected corresponding soluble protein isoforms 
. Similarly, soluble HLA-G, an MHC class I molecule which has limited polymorphism and shares high structural similarity with CD1D 
, is encoded by a TM-deficient variant 
, and was recently shown to activate NK cells via NFκB signalling 
“V4” is the most highly expressed transcript in NHBE cells; and is potentially an interesting variant which could have epithelial-specific function. Its protein product is predicted to lack ß2
m binding ability but it has intact TM and cytoplasmic tail (T) exons, so it can produce a cell surface-expressed β2
m-free CD1d isoform. On human intestinal epithelial cells, the major form of CD1d is a non-glycosylated, ß2
m-independent molecule, although these cells also express a ß2
m-associated, fully glycosylated form of CD1d 
. The importance of ß2
m association with CD1d with respect to the CD1d biosynthetic pathway is not established. Association of ß2
m with CD1d could have a role in regulating the extent of CD1d glycosylation and the maturity of the attached carbohydrate side chains 
. Although, in most cases, forms of MHCI protein that do not associate with ß2
m do not fold their antigen presenting domains properly, cells from ß2
m -deficient mice are able to stimulate proliferation of CD1d-restricted T cell clones 
, suggesting presence of a CD1d variant that do not rely on ß2
m for antigen presentation and function.
Regardless of the transcripts, identification by antibodies confirms the presence of protein on the surface of bronchial epithelial cells. The CD1d antibody, clone 42, has been widely used as a blocking antibody to prevent activation of NKT by CD1d-expressing cells 
, therefore flow cytometry results verify presence of (at least) one functional CD1d (likely full length protein) with the ability to associate with ß2
m on the surface human bronchial epithelial cells.
In summary, we have identified for the first time, CD1D expression on human bronchial epithelial cells and detection of at least some of the variant protein by CD1d monoclonal antibodies. This provides a basis for further investigations for the role of this molecule in lung immune responses.