Search

Biological membranes are compartmentalized into microdomains that exhibit particular lipid and protein compositions. Membrane microdomains, such as tetraspanin-enriched microdomains and lipid rafts, have been suggested to play a role in a variety of physiological and pathological processes. Therefore, the characterization of the protein compositions of these microdomains, which is the focus of this review, appears to be a crucial step to better understanding their function. Proteomics has recently allowed the characterization of tetraspanin-enriched microdomains in colon cancer cells. This demonstrated the presence of different categories of membrane proteins and suggested a variation in the composition of tetraspanin-enriched microdomains during tumor progression. On the other hand, proteomics has permitted the identification of hundreds of proteins in lipid rafts of different origins. However, the diversity of methodologies in sample preparation and of strategies in protein identification led to a broad variability in the data obtained. These methodological issues are discussed. Moreover, proteomics has revealed that different sets of proteins were present in tetraspanin-enriched microdomains as compared with lipid rafts, strengthening the idea that these microdomains are distinct structures.

Tetraspanins are integral membrane proteins involved in a variety of physiological and pathological processes. In cancer, clinical and experimental studies have reported a link between tetraspanin expression levels and metastasis. Tetraspanins play a role as organizers of a molecular network of interactions, the "tetraspanin web". Here, we have performed a proteomic characterization of the tetraspanin web using a model of human colon cancer consisting of two cell lines derived from primary tumor and metastasis from the same patient. The tetraspanin complexes were isolated after immunoaffinity purification and the proteins were identified by MS using LC-ESI-MS/MS and MALDI-FTICR. The high resolution and mass accuracy of FTICR MS allowed reliable identification using mass finger printing with only two peptides. Thus, it could be used to resolve the composition of complex peptide mixtures from membrane proteins. Different types of membrane proteins were identified, including adhesion molecules (integrins, Lu/B-CAM, GA733 proteins), receptors and signaling molecules (BAI2, PKC, G proteins), proteases (ADAM10, TADG15), and membrane fusion proteins (syntaxins) as well as poorly characterized proteins (CDCP1, HEM-1, CTL1, and CTL2). Some components were differentially detected in the tetraspanin web of the two cell lines. These differences may be relevant for tumor progression and metastasis.

By interacting with each others, the tetraspanins are thought to assemble a network of molecular interactions, the tetraspanin web. These tetraspanin/tetraspanin interactions involve in part the palmitoylation of the proteins. We show that tetraspanins interact with cholesterol as indicated by the precipitation of tetraspanin/tetraspanin complexes by digitonin, a cholesterol-precipitating reagent, and the labeling of the tetraspanins CD9, CD81 and CD82 with a photoactivatable cholesterol in vivo. Cholesterol may participate to the interaction of tetraspanins with each other since digitonin-precipitation of tetraspanins was correlated with their mutual interaction, and because these interactions were disrupted following cholesterol depletion by methyl-beta-cyclodextrin (MbetaCD) treatment, or cholesterol sequestration by saponin. A mutant CD9 molecule lacking all palmitoylation sites was not precipitated by digitonin under conditions in which wild-type CD9 was precipitated, indicating a role of palmitoylation for the interaction with cholesterol. Finally, upon ligation of tetraspanins on the surface of a lymphoid B cell line, the tyrosine phosphorylation of several proteins, including the vav nucleotide exchange factor, was inhibited when cells were pretreated with MbetaCD, and increased when they were treated with MbetaCD/cholesterol complexes. Thus, there is a physical and functional link between tetraspanins and cholesterol.

The tetraspans associate with a large number of surface molecules, including a subset of beta1 integrins and, indirectly through CD19, with the complement receptor CD21. To further characterize the tetraspan complexes we have raised and selected monoclonal antibodies (mAb) for their ability to immunoprecipitate a molecule associated with CD9. A unique mAb was identified which recognizes the complement regulator CD46 (membrane cofactor protein). CD46 associated in part with several tetranspans and with all beta1 integrins that were tested (CD29/CD49a, CD29/CD49b, CD29/CD49c, CD29/CD49e, CD29/CD49f) but not with beta4 integrins. These data, together with cross-linking experiments showing the existence in living cells of CD46/integrin complexes, suggest that CD46 associates directly with beta1 integrins and indirectly with tetraspans. CD46 also acts as a receptor for measles virus; however, mAb to various integrins and tetraspans did not modify the virus fusion entry step.

The molecules of the tetraspan superfamily have two unequal extracellular domains separated by four transmembrane (TM) domains. These molecules are associated on the cell surface with each other and with other partner molecules, in particular beta1 integrins. We now show that CD9 associates with the precursor of the beta1 integrin (pre beta1). This association is detected as early as 15 min after metabolic labeling, and the use of Brefeldin A demonstrates that it does not require Golgi modifications of either CD9 or integrin beta1. The specificity of this interaction is demonstrated by the fact that other tetraspans, CD63, CD81, and CD82, do not associate with pre beta1, and CD9 does not associate with immature human histocompatibility leukocyte antigen class I. In order to localize the region of CD9 responsible for the association with the beta1 integrin, we have generated two reciprocal chimeric CD9/CD82 molecules with the junction localized just after the third TM region. The large extracellular loop of CD9 or the fourth TM domain, or both, appear to be sufficient to mediate an association with the mature integrin with a high efficiency, compared to CD82. By contrast, association with pre beta1 requires at least two regions of the molecule. Mutation of CD9 at the consensus site of the tetraspan superfamily, localized between the second and the third TM domain, did not impair the co-precipitation of pre beta1. Finally, because pre beta1 associates with calnexin, we have investigated a possible association of CD9 with this chaperone molecule. CD9 associates with calnexin independently of its association with the beta1 integrin, suggesting that calnexin could be involved in the processing of CD9.

CD9, CD63, CD81, and CD82 are glycoproteins of unknown function which belong to the tetraspan superfamily. These molecules have short cytoplasmic sequences, four transmembrane domains and two unequal extracellular regions. Here, we show that these molecules are associated with each other on cell surface and with other glycoproteins such as very late antigen (VLA) integrins and HLA-DR antigens. Moreover, the VLA integrins and HLA-DR antigens were also found to be associated. The interactions of these molecules were analyzed by transfection experiments. It is demonstrated that overexpression of CD9 antigen in Raji cells leads to a lower efficiency of precipitation of CD81 and CD82, suggesting a direct interaction between these molecules. In these cells, the co-precipitation of CD81 and CD82 was not modified, suggesting that these tetraspans did not compete for association. However, in COS-7 cells, transfection of both CD81 and CD82 led to a marked reduction of the number of CD9/CD81 or CD9/CD82 complexes compared to single-transfected cells, and this was associated with the appearance of CD81/CD82 complexes. Therefore, in this cellular system, CD9 competes with CD81 and CD82 for association with the other tetraspan proteins. Finally, the tetraspans do not compete for the association with integrins or HLA-DR. Indeed, when CD9 was expressed in Raji cells, it was incorporated into the pre-existing complexes of these molecules with CD81 and CD82. These data suggest the existence of a tetraspan network which, by connecting several molecules, may organize the positioning of cell surface proteins and play a role in signal transduction, cell adhesion, and motility.

The CD9 antigen is a cell surface glycoprotein of unknown function which belongs to the tetraspans family. We demonstrate here, by precipitation, Western blotting and co-capping experiments, that this molecule is associated with a large fraction of beta 1 integrins in two cell lines, the pre-B cell line NALM-6 and the megakaryocytic cell line HEL. In HEL cells, CD9 antigen is only associated with VLA-4. In contrast, in NALM-6 cells, CD9 antigen is associated with both VLA-4 and VLA-5. On the other hand, only the beta 1 chain is co-precipitated with the CD9 antigen in transfected L cells. These data show that the CD9 antigen is associated with the beta 1 chain rather than with a particular integrin. CD9 monoclonal antibodies (mAb) did not modify the binding of HEL and NALM-6 cells to fibronectin, laminin or collagen. The association of CD9 antigen to VLA integrins is strengthened by the fact that both CD9 and anti-VLA mAb induce aggregation of the two cell lines and inhibit their migration in Transwell chambers. Because the aggregating effect, but not the inhibition of migration, is observed in CEM or CD9-transfected CEM cells, these two effects are likely to be mediated by different mechanisms.

In this report we show that it is possible to induce the expression of HLA-DR antigens on K562 cells, previously reported to be unresponsive to interferon-gamma (IFN-gamma). However, only low cell concentrations and a high dose of IFN-gamma allowed the induction of HLA-DR antigens. Furthermore, the recombinant glycosylated IFN-gamma is 100-fold more efficient than the unglycosylated form. This induction of HLA-DR antigens on K562 was not related to a stage of differentiation or to the presence of cells subsets specifically sensitive to IFN-gamma, since repeated sorting of K562 HLA-DR-positive and negative cells did not lead to the selection of a cell subset with a different potential of induction for HLA-DR. The difficulty in obtaining induction is due to the production of a soluble endogenous inhibitor of proteic nature, whose action is not restricted to the K562 cell line since it operates also on both epithelial and fibroblastic cells. Treatment of normal human epithelial and fibroblastic cells with conditioned medium from K562 cultures caused a marked decrease in the expression of HLA class II antigens (DR and DP) induced by IFN-gamma (10,000 U/ml), but had no effect on cell growth; however, it also affected expression of HLA class I antigens. This inhibition is not mediated by prostaglandin or an IFN-alpha or IFN-beta-dependent mechanism. Production of this inhibitor by pluripotent human leukemic cells could cause an unbalance in the complex control exerted by the immunological system during hematopoietic differentiation or leukemic progression.

Preimmunization of Lewis rats with anti-S antigen (S-Ag) monoclonal antibodies (mAb) led to protection against experimental autoimmune uveoretinitis (EAU) induced by this retinal autoantigen. High titers of anti-idiotypic (anti-Id) antibodies were raised against three mouse mAb, S2D2 (IgG2b), S6H8 (IgG2a) and S7D6 (IgG1), directed at S-Ag. An almost complete prevention was observed in S2D2 mAb-immunized animals while a partial protection was achieved with S6H8 and S7D6 mAb. No detectable anti-Id antibody nor disease prevention was observed in rats immunized with the mAb S9E2 (IgG2a) which only recognizes bovine and sheep S-Ag, or with control mAb of the same isotypes irrelevant to S-Ag. The mAb treatment did not modify the level of the whole polyclonal antibody response to S-Ag. These results suggest an important role in the pathogenesis of EAU for the epitopes recognized by S2D2-S6H8 and S7D6 in the S-Ag molecule. The success of anti-Id immunization for autoimmune disease suppression may depend on the identification of relevant epitopes.

Mouse monoclonal antibodies (mAb), of either IgG2a or IgG2b isotypes, specific for the retinal S-autoantigen (S-Ag) or a pool of rat anti-S-Ag sera prevented experimental autoimmune uveoretinitis in Lewis rats when injected i.p. at the time of immunization. Control mAb of the same isotypes, irrelevant to S-Ag, had no inhibitory effect. The humoral response to S-Ag, as studied by enzyme-linked immunosorbent assay using a mouse mAb specific for rat kappa chain, was moderately but significantly reduced in suppressed animals. The rapid disappearance of the injected mAb from rat sera, as measured using a rat mAb specific for mouse kappa chain, could be explained by its complexing with either autologous antigen released from the retina at the site of inflammation, or anti-idiotypic antibodies.