Dendritic Cell Migration and Antigen Presentation Are Coordinated by the Opposing Functions of the Tetraspanins CD82 and CD37

Dendritic cells (DC) are the most potent of the APCs because they have the unique ability to activate naive Ag-specific T cells (1). However, DC function varies with activation state; the classical example being migratory DCs. Unactivated migratory DCs in the periphery efficiently patrol through tissues and are specialized for Ag uptake. These highly endocytic cells are poor stimulators of T cells, because of both a moderate surface expression and a high turnover of MHC molecules. However, once DCs receive danger signals transduced through pattern recognition receptors, they reduce MHC turnover and upregulate expression of MHC/peptide complexes, costimulatory molecules such as CD80 and CD86, and proinflammatory cytokines, all of which increase their capacity to stimulate T cells and direct adaptive cellular immunity (2). Concurrently, DCs undergo morphological changes involving the extension of dendrites thought to promote efficient interactions with T cells, a decrease in Ag uptake and processing, and a modification in cell migration. Thus, rather than migrating randomly through tissue, they now migrate directionally, via the lymphatics to the draining lymph nodes (2–4). The mechanisms of migration used by DCs have been reported to be both dependent on adhesion molecules (5–7) and adhesion-independent ameboid migration, driven chiefly by cytoskeletal protrusion and contractile forces (8). Nonetheless, the activated DC that has migrated to the draining lymph node has now metamorphosed to become a cell specialized at initiating adaptive immunity by presenting Ags to naive T cells and inducing their activation (2). These major functional changes require exquisite coordination between the Ag processing and presentation machinery, and the myriad of adhesion, signaling, and cytoskeletal proteins that regulate DC morphology and migration.

One type of protein known to regulate both Ag presentation and cell motility is the tetraspanins, a superfamily of four-transmembrane molecules, highly conserved in evolution and expressed in all mammalian cells. The best-defined role of tetraspanins in biology is their molecular organization of cell membranes. Tetraspanins directly interact with their molecular partners and organize them into signal-transducing microdomains (9, 10). Components of tetraspanin-enriched microdomains (TEM) include membrane proteins such as integrins, proteases that regulate cell-surface molecule expression posttranslationally, and signaling molecules including kinases and phosphatases. Tetraspanins regulate the spatiotemporal molecular interactions of their partner proteins and thereby influence complex cellular events such as activation, adhesion, and migration. An emerging role for TEMs is in communication between the cell surface and the cytoskeleton. Tetraspanins regulate cytoskeletal-dependent processes such as outside-in integrin signaling and adhesion strengthening, actin polymerization, cellular polarity, and spreading (9–11). On the molecular level the tetraspanin CD81 has been shown to interact with three important regulators of cytoskeleton polymerization: the Rho GTPase Rac, Ezrin, and Moesin. Ezrin and Moesin are peripheral membrane proteins that link the cytoskeleton with the plasma membrane (11).

In APCs, tetraspanins have been implicated in many facets of biology including pattern recognition (12), Ag presentation (13), and cell migration (14). A role for tetraspanins in Ag processing is suggested both by their presence in MHC class II compartments, where they associate with MHC class II and the peptide editors HLA-DM and HLA-DO (15–17), and by their association with MHC I and II at the cell surface (15, 18). However, the consequences of these interactions on Ag presentation are incompletely understood, with studies variously pointing to roles in promoting (19) and restraining Ag presentation (20, 21).

Recently, we have demonstrated that the tetraspanin CD37 plays an important role in linking Ag processing and presentation with cell migration during DC activation. CD37 molecularly interacts with MHC (22), is downregulated upon DC activation (23), restrains MHC-mediated Ag presentation (24), but is also required for optimal DC migration to draining lymph nodes (14). The closest phylogenetic relative to CD37 in the human genome is CD82 (25), a tetraspanin best characterized as a metastasis suppressor that influences cellular adhesion and is a prognostic indicator in various nonimmune human cancers (26, 27). In APCs, CD82 also molecularly associates with MHC, both at the cell surface and in intracellular vesicles (16, 17, 28). We therefore hypothesize that CD82 is also an excellent candidate to regulate the cell migration and Ag presentation machinery in DCs. In this article, we show a central role for CD82 in DC function where remarkably, it shows an effect opposite to CD37; CD82 is upregulated upon bone marrow–derived dendritic cells (BMDC) activation, negatively regulates DC migration, but is required for efficient activation of T cells. Thus, the data support a new concept that by regulating the expression of the functionally opposing tetraspanins CD37 and CD82 during activation, DCs coordinate the cellular machinery important for both Ag presentation and cell migration.

All mice used were backcrossed 10 times to the C57BL/6J [wild-type (WT)] genetic background. CD37^−/− mice were generated by homologous recombination (29). CD82^−/− mice were generated by cre-loxP recombination (30). Mice were bred from established colonies at either the Alfred Medical Education and Research Precinct Animal Services (Prahran, Victoria, Australia), the Animal Research Laboratories (Clayton, Victoria, Australia), or the Radboud University Medical Center (the Netherlands). C57BL/6-OT I, C57BL/6-OT I Ly5.1⁺, and C57BL/6-OT II mice were obtained from the Walter and Eliza Hall Institute (Melbourne, Victoria, Australia). All mice used were male, 6–12 wk of age, and age matched to relevant controls. All experiments were carried out under the ethics approval of the Animal Ethics Committee at the Alfred Medical Education and Research Precinct Animal Services.

Mice were tail or ear clipped upon weaning (14–21 d old), and genomic DNA was isolated via a DNA Isolation Kit (Promega). Oligonucleotides were custom orders (Geneworks). WT PCR was performed using the following primers: forward: 5′-AGGTGTTTGCCCTTCTCCTT-3′ and reverse: 5′-CCACCTGTGACAACCAAGTG-3′. CD82^−/− PCR was performed with the following primers: forward: 5′-TCCTTAAGCCTCAAGAAAACC-3′ and reverse: 5′-TGTGAGGGCTCCAGTCTCC-3′. PCR program was as follows: 95°C for 15 min, cycle: 95°C 30 s, 50°C 30 s, 72°C 45 s, 35 cycles, 72°C for 7 min (final annealing), and hold at 4°C. CD82^−/− mice were shown to lack CD82 mRNA, and levels of CD82 in WT BMDCs were measured via TaqMan real-time PCR for mouse CD82 (CD82^−/− expression of CD82 not shown) (Catalog Mm00492061_m1; Life Technologies). Levels of CD37 in WT BMDCs were measured via TaqMan real-time PCR for mouse CD37 (Catalog Mm00514240_m1; Life Technologies). Levels of 18S were measured by TaqMan real-time PCR for mouse 18S (Catalog Mm_03928990_g1; Life Technologies). Real-time PCR was performed according to the manufacturer’s guidelines.

DCs were isolated as previously described (14). In brief, bone marrow was cultured for 7–9 d with 10 ng/ml GM-CSF and IL-4 (R&D Systems) to obtain BMDCs, and activated with 1 μg/ml LPS. Spleens were treated with enzymatic digestion and density-gradient centrifugation before magnetic bead depletion to obtain splenic DCs. T cells were isolated as previously described (31). In brief, spleens and lymph nodes were prepared into a single-cell suspension before being subjected to magnetic bead depletion.

Human PBMCs were isolated from buffy coats and monocytes were enriched by plastic adherence. Monocytes were cultured with 450 U/ml GM-CSF (Strathmann) and 300 U/ml IL-4 (Strathmann) in complete RPMI 1640 (Invitrogen Life Technologies)/10% FCS for 6 d to generate immature DCs. DC maturation was induced by addition of 200 ng/ml LPS for 24 h. DCs were fixed in 2% paraformaldehyde, mounted on poly-l-lysine–coated coverslips, and stained with anti-CD37 (WR17) and anti-CD82 (B-L2; Serotec) followed by goat anti-mouse Alexa 488 secondary Abs. Samples were analyzed by flow cytometry or confocal laser scanning microscopy.

In vivo responses were performed as previously described (14). In brief, IFN-γ ELISPOT was performed on restimulated splenocytes 2 wk after injection of gamma-irradiated B16-OVA. To measure Ag presentation and T cell priming in vivo, we injected 3 × 10⁶ CFSE-labeled OT-I Ly5.1⁺ T cells i.p. into recipient mice. Two days later, 5 × 10⁵ LPS activated (1 μg/ml), SIINFEKL-pulsed (1 μg/ml, 1 h), and Cell Tracker Orange (0.5 μM; Life Technologies)–labeled BMDCs were injected intradermally into the base of the tail. Seventy hours later, mice were culled, draining lymph nodes harvested, and the number of T cell divisions and migrated BMDCs were measured by flow cytometry. The ratio was calculated by enumerating the total number of T cell divisions and dividing this by the number of BMDCs that had migrated to the draining lymph nodes.

T cell proliferation was performed as previously described (31). In brief, purified WT or CD82^−/− T cells were stimulated with anti-CD3 in the presence or absence of anti-CD28. In vitro Ag presentation was performed by coculturing of OVA-specific OT-I and OT-II T cells or B3Z hybridomas with either splenic or WT, CD82^−/−, and CD37^−/− BMDCs pulsed with either SIINFEKL or ISQAVHAAHAEINEAGR (OVA helper peptide). In costimulation assays, WT T cells were stimulated with anti-CD3, in the presence or absence of WT, CD82^−/−, or CD37^−/− BMDCs. T cell proliferation was measured by [³H]thymidine incorporation (Amersham). B3Z hybridoma activation was measured by colorimetric determination of IL-2 production (24).

Maturation assays were performed on freshly isolated splenic DCs or BMDCs as previously described (32). In brief, DCs were serum staved before pulsing with [³⁵S]methionine/cysteine before a chase of 4.5 h. Levels of MHC maturation were measured via immunoprecipitation of MHC class II (MHC II) (M5/114) and SDS-PAGE and exposure to film.

OT-I or OT-II cells were stained with CFSE (0.5 μM, 15 min, 37°C; Invitrogen) or Cell Tracker Orange (0.5 μM, 15 min; Life Technologies). DCs were stained with either Cell Proliferation Dye eFluor 670 (referred to in this article as CT670; 0.5μM, 15 min, 37°C; eBioscience) or CFSE as described earlier. For flow conjugation, BMDCs were preactivated for 18 h with 1 μg/ml LPS (splenic DCs were not activated in vitro) and after staining immediately incubated with increasing concentrations of OVA peptide [either SIINFEKL or ISQAVHAAHAEINEAGR (OVA helper peptide) for at least 2 h at 37°C], after which cells were mixed together (1:1 T cell/DC ratio), centrifuged at 200 × g for 2 min, and incubated (37°C; 15 min). Cells were placed on ice and immediately analyzed by flow cytometry. For live microscopy, BMDCs were allowed to adhere to glass slides in the presence of 1 μg/ml LPS for 18 h and were pulsed with SIINFEKL or CD4 helper for at least 2 h before being washed three times. Slides were then placed within a temperature- and CO₂-controlled stage and T cells were added for live microscopy. Slides were imaged every 30 s from 5 min after addition of T cells for between 20 and 60 min with a special-order Nikon A1r Plus Confocal (60× lens; driven by NIS-Elements 4.1 acquisition software; CFI Apochromat 40XWI).

DC migration assays were performed as previously described (14). In brief, migration of dermal DCs out of ear explants was performed, where ears were cultured in the presence or absence of CCL19. Transwell migration was performed by the use of LPS-activated BMDCs in transwells migrating toward a CCL19 chemokine gradient. In vivo migration of intradermally injected BMDCs to draining lymph nodes was performed as previously described (14) except 1 μM CT670 (eBioscience) was used in place of SNARF-1.

BMDC basal expression and expression after activation with 1 μg/ml LPS for 18 h was examined. Cells were stained with anti–CD11c-allophycocyanin (BD Pharmingen, San Diego, CA) and either anti–MHC II–FITC (M5/114), anti–CD80-PE, or anti–CD86-PE, anti–LFA-1 FITC, anti–CD11b FITC, CD49dFITC (all generated in-house), anti–MHC I biotin (BD Pharmingen), or anti-CD49e biotin (BD Pharmingen) plus streptavidin-FITC (BD Pharmingen), anti-CCR7 PE (eBioscience), anti–ICAM-1 PE (Molecular Probes), and examined under flow cytometry by FACSCalibur (Becton Dickinson) or special-order LSR Fortessa (Becton Dickinson).

BMDC morphology and adhesion were assessed as previously described (14). In brief, BMDCs were stimulated with 1 μg/ml LPS, adhered to fibronectin-coated slides in the presence of 50 ng/ml PMA before being fixed and stained for phalloidin-FITC. Microscopy images were scored blind before the average size of the cell membrane was calculated. For confocal microscopic analysis, cells were additionally stained with rabbit anti–β-tubulin Ab (Cell Signaling Technology) and then anti-rabbit Alexa 647. Cells were imaged by confocal microscopy [Nikon A1r Plus Confocal, driven by NIS-Elements 4.1 acquisition software; a 60× lens (CFI Apochromat LWD 40xWI) was used]. Images acquired consisted of a stack of 5 × 1 mm steps, and the fluorescence intensity of both phalloidin and β-tubulin staining was determined at the focal plane using ImageJ analysis.

Two million naive BMDCs were allowed to adhere to plastic tissue culture wells for 18 h before being serum staved for 2 h. Rho/Rac-1/Cdc42 Activator I (CN04 toxin) was added for 2 h, before cells were lysed in situ for G protein linked immunosorbent assay. G-LISA RhoA Activation Assay Biochem Kit (catalog no. BK124), G-LISA Rac-1 Activation Assay Biochem Kit (catalog no. BK128), and G-LISA Cdc42 Activation Assay Biochem Kit (catalog no. BK127) were performed as per manufacturer’s guidelines (all from Cytoskeleton, Denver, CO).

CD82^−/− or WT splenic DCs were coincubated with FITC-labeled, carboxylate-modified microspheres (Invitrogen-Molecular Probes, Carlsbad, CA) either at a DC/bead ratio of 500:1 for 40 nm FITC-coated beads or 10:1 for 500 nm FITC-coated beads over 48 h at 37°C. After 48 h, cells were washed three times, stained for CD11c expression (as described earlier), and analyzed via flow cytometry.

BMDCs were cultured at 1 million cells in 1 ml complete RPMI 1640 containing 1 μg/ml LPS for 24 h. Supernatant was then collected and filtered through a 0.22-μm filter and spun twice to remove cellular debris, before a Mouse Inflammation Cytometric Bead Array kit was performed as per the manufacturer’s guidelines (Catalog 552364; Becton Dickinson).

All data are presented as mean ± SEM. Comparisons between groups were determined by Student t test, or, where appropriate, ANOVA. The p values are as follows: p > 0.05 not significant, *p < 0.05, **p < 0.01, ***p < 0.001.

To determine whether the tetraspanin CD82 has a role in the cell-mediated immune response, we studied a newly generated CD82^−/− mouse. WT and CD82^−/− mice were challenged intradermally with gamma-irradiated B16-OVA cells. Two weeks later, the frequency of T cell–specific IFN-γ responses in the spleen was measured by peptide-specific ELISPOT. CD82^−/− mice had significantly decreased Ag-specific CD8⁺ responses (Fig. 1A); however, CD4⁺-specific responses were not above background in a single immunization, as had been previously described in this model (14). The reduced CD8 response could not be attributed to a defect intrinsic to CD82^−/− T cells because IFN-γ responses in CD82^−/− T cells stimulated with ConA were indistinguishable from WT (data not shown). Moreover, in vitro stimulation with mAbs against CD3 and CD28 revealed that T cells from CD82^−/− mice were hyperproliferative (Fig. 1B–E). This finding is consistent with previous findings that T cells deficient in other tetraspanins are also hyperproliferative (33), but does not explain the poor in vivo cellular immunity observed in CD82^−/− mice.

The impaired T cell responses observed in CD82^−/− mice (Fig. 1) are reminiscent of mice deficient in the tetraspanin CD37, the closest phylogenetic relative to CD82 (25). In this article, the poor T cell responses induced by CD37 ablation were attributed to defects in DC biology, because CD37 both restrains Ag presentation and promotes DC migration (14, 24). We therefore hypothesized that CD82 also had an important role in DCs, and thus sought to compare CD37 and CD82 function in DCs. We first compared the expression of CD82 and CD37 in WT mouse BMDCs, before and after activation, by quantitative PCR, because Abs are not available for murine CD82 and CD37. After 1 h of activation, CD37 expression is moderately upregulated and CD82 moderately downregulated. However, 18 h later, CD82 was markedly upregulated, whereas CD37 was downregulated and comparable with basal levels (Supplemental Fig. 1A). The data are broadly consistent with published flow cytometry that confirmed a downregulation of CD37 protein in human monocyte-derived DCs (MoDCs) upon LPS activation (23) and with flow cytometry and confocal microscopy where we observed that CD82 protein is upregulated upon human MoDC activation (Supplemental Fig. 1B, 1C).

Given that CD37 and CD82 molecularly interact with MHC in APCs (16–18, 22, 28), we next investigated whether CD82 has a role in regulating the priming of naive T cells by DCs. WT, CD82^−/−, and CD37^−/− BMDCs were pulsed with either CD4⁺- or CD8⁺-specific peptides and cocultured with Ag-specific T cells. In contrast with CD37^−/− BMDCs, which showed a hyperstimulatory phenotype (Fig. 2A, 2B, 2D, 2E), in both temporal and dose responses, CD82^−/− BMDCs showed a clear impairment in the ability to present Ag to either class I–restricted CD8⁺ OT-I T cells (Fig. 2A, 2B) or class II–restricted CD4⁺ OT-II T cells (Fig. 2D, 2E). We observed a similar impairment of T cell activation in Ag-pulsed splenic CD82^−/− DCs (Fig. 2C, 2F). We conclude that, unlike CD37, which restrains Ag presentation, CD82 expression in DCs is essential for T cell activation, consistent with the poor cellular immunity observed in CD82^−/− mice caused by a failure of CD82^−/− DCs to adequately stimulate naive T cells.

To understand the mechanism by which CD82 in DCs regulates T cell activation, we used classical assays of Ag presentation to T cell hybridomas, the upregulation of activation markers after stimulation, costimulatory activity, and cytokine production. Even though we confirmed that CD37^−/− BMDCs were hyperstimulatory to T cell hybridomas (24), we could detect no differences between WT and CD82^−/− BMDCs (Supplemental Fig. 2A–E). CD82 is found in intracellular compartments, including phagosomes, and MHC II compartments together with MHC II and HLA-DM and HLA-DO (16–18). Therefore, we next considered the possibility that CD82 plays a role in Ag uptake, processing, and MHC maturation. Even though CD82 ablation did not affect the phagocytosis of fluorescent beads (Supplemental Fig. 2F, 2G), pulse chase analyses of MHC II maturation did show a small (7%) but reproducible and significant decrease in the formation of mature MHC/peptide complexes relative to WT cells (Fig. 2G, 2H). From these data, we conclude that CD82 does regulate MHC II maturation and that this regulation might contribute to the impairment of T cell activation induced by Ag-pulsed CD82^−/− DCs.

Because CD82 is involved in the regulation of cellular adhesion in nonimmune cells (26), we reasoned that CD82 might regulate cell–cell adhesion between DCs and T cells; therefore, we examined whether CD82^−/− DCs were able to form stable conjugates with T cells. BMDCs from WT or CD82^−/− mice were stained with CT670 and pulsed with increasing concentrations of OVA helper peptide. BMDCs were then cocultured for 15 min with CFSE-labeled OT-II T cells and analyzed by flow cytometry to enumerate DC–T cell conjugates defined as viable double-color–positive events (Fig. 3A, top right quadrant). Representative flow cytometry data show a clear defect in the ability of CD82^−/− BMDCs to form stable conjugates with Ag-specific T cells (Fig. 3A). Quantitative analyses demonstrate that increasing peptide concentration significantly increased conjugate formation for all strains, but CD82^−/− BMDCs formed significantly fewer conjugates with T cells than WT BMDCs (or CD37^−/− BMDCs). Indeed, conjugate formation was attenuated by up to 40% at high peptide concentrations (Fig. 3B, 3C). The same phenotype was observed in primary CD82^−/− splenic DCs (Fig. 3D, 3E). To examine the kinetics of conjugate formation, we used a live cell imaging approach where BMDCs were labeled with either CT670 or CFSE, and cocultured for 18 h with LPS and 2 h with peptide before being imaged for up to 1 h with either OT-I or OT-II T cells stained with Cell Tracker Orange. An example (shown in Fig. 3F) demonstrates that although WT BMDCs interacted with T cells for several minutes (see blue DCs interacting with orange T cell from 5 to 15 min in Fig. 3F), the interactions between CD82^−/− BMDCs and T cells were transient (see green DCs interacting with T cell from 2 to 5 min, but by 7.5 min the interaction is not maintained in Fig. 3F; see also Supplemental Video 1 for another example). Quantification of the interactions revealed that the average duration of the interactions between CD82^−/− BMDCs and T cells was shorter than WT counterparts for both MHC class I– and class II–restricted Ags (Fig. 3G, 3H), with the CD82^−/− BMDCs having a much larger proportion of short-lived interactions relative to WT (Fig. 3I). Moreover, when assessing the nature of the interaction, we noted that the proportion of interactions between DC and T cells that were static (as opposed to interactions where the T cell crawled over the DC surface) was >2-fold higher in WT BMDCs compared with CD82^−/− BMDCs (Fig. 3J). We therefore conclude that CD82^−/− DCs are poor stimulators of T cells mainly because of their inability to form long, stable, static, cellular interactions with T cells.

We have recently published that CD37 is important for optimal DC migration from the periphery to the draining lymph node (14). Given the extensive data from cancer cell biology documenting a role for CD82 in suppressing migration and metastases (27), we hypothesized that CD82 might also regulate DC migration. Therefore, we assessed migration of WT, CD82^−/−, and CD37^−/− DCs using a series of in vivo and in vitro assays. First, in a model where skin DCs migrate out of mouse ear explants in response to the chemokine CCL19, we confirmed our previous observations that CD37^−/− DCs showed impaired migration (14). By contrast, CD82^−/− DCs showed a striking hypermigratory phenotype (Fig. 4A). This finding could not be attributed to a DC developmental defect, because the total number of CD11c⁺ DCs in CD82^−/− ear tissue, enumerated by enzymatic digestion and release, was comparable with WT mice (Fig. 4B). To determine whether the dysregulated migration induced by CD82 ablation was intrinsic to DCs, or might be explained by defects in CD82^−/− microanatomy, BMDCs were loaded into the upper chamber of transwells with or without CCL19 in the lower chamber. CD82^−/− BMDCs migrated toward CCL19 in much greater numbers than WT BMDCs, and in contrast with the hypomigratory CD37^−/− BMDCs (Fig. 4C). To determine whether the dysregulated migration induced by CD82 ablation affected the ability of DCs to traffic to draining lymph nodes, we differentially labeled BMDCs from WT, CD82^−/−, or CD37^−/− mice with CFSE or CT670 and coinjected intradermally. Forty-eight hours later, draining lymph nodes were analyzed by flow cytometry. In these experiments, CD82^−/− BMDCs displayed a markedly increased ability to migrate to draining lymph nodes (Fig. 4D). This was confirmed by determining the migration index calculated relative to an internal CT670-stained WT control. CD82^−/− BMDCs showed, on average, 6-fold greater numbers arriving in the draining lymph nodes than WT BMDCs (Fig. 4E). To determine how this hypermigratory phenotype affected the efficiency of CD82^−/− DCs (which are poor presenters of Ag to T cells; Figs. 2, 3) to initiate T cell activation in vivo, WT and CD82^−/− BMDCs were fluorescently labeled, pulsed with SIINFEKL, and injected intradermally into WT recipients, where both the migration of BMDCs to draining lymph nodes and the proliferation of adoptively transferred, CFSE-labeled OT-I T cells were measured and expressed as a ratio. Although the total amount of T cell divisions induced by CD82^−/− BMDCs did not differ from their WT counterparts (data not shown), the efficiency of in vivo Ag presentation per migrated CD82^−/− BMDCs was significantly impaired (Fig. 4F). Finally, the expression of cell-surface molecules that are involved in DC migration and adhesion, such as the CCL19 receptor CCR7 and integrins, were assessed by flow cytometry; we could detect no significant differences between WT and CD82^−/− BMDCs (Fig. 4G). We conclude that in contrast with CD37, which is required for optimal DC migration, CD82 negatively regulates DC migration.

Tetraspanin CD82 negatively regulates DC migration yet promotes Ag presentation. Conversely, its closest phylogenetic relative among the tetraspanins, CD37, opposes CD82 as it promotes DC migration but negatively regulates Ag presentation. Given that adhesion-dependent and -independent DC migration are both reliant on cytoskeletal rearrangement, and that the cytoskeleton plays a key role in organizing the immunological synapse and promoting Ag presentation in DCs (34, 35), we reasoned that ablation of the tetraspanins CD37 and CD82 in DCs might induce a dysregulation in cytoskeletal function. To assess cytoskeletal function in BMDCs, we first analyzed their ability to spread upon adhesion to the integrin ligand fibronectin. The morphology of BMDCs in all three strains was heterogeneous, and three types of cell morphologies were observed: 1) a small unspread morphology that lacked cellular projections, 2) a large spread cell that lacked cellular projections, and 3) a classical dendritic morphology (Fig. 5A). Quantitative assessment of spreading confirms our previous findings (14) that CD37^−/− BMDCs tend to be of cell type 1, in that they spread poorly on fibronectin and have a lack of cellular projections. CD82^−/− BMDCs also lacked projections, but by contrast were predominantly of cell type 2, large spread cells without projections (Fig. 5B). This was confirmed by measurement of surface area of BMDC spreading on fibronectin, which was exaggerated in the absence of CD82 (Fig. 5C). Fibronectin-adherent BMDCs were permeabilized, costained with phalloidin (FITC) and Abs against β-tubulin (red), and analyzed by confocal microscopy to further analyze the cytoskeleton. Representative images depicting the cytoskeletal network in WT BMDCs, large spread CD82^−/− BMDCs, and small unspread CD37^−/− BMDCs are depicted (Fig. 5D–F). To quantify the tubulin and polymerized networks, we measured the mean fluorescence intensity of phalloidin and β-tubulin. The data show a defect in the amount of polymerized actin measured in both CD82^−/− and CD37^−/− BMDCs (Fig. 5H), whereas a poor expression of β-tubulin was evident in the CD37^−/− BMDCs (the increased expression of β-tubulin in CD82^−/− BMDCs did not reach statistical significance, p = 0.1; Fig. 5G).

To understand how tetraspanin deficiency could lead to a dysregulation of cytoskeletal rearrangement, we assessed whether CD82 or CD37 influenced the activation of the key regulators of the cytoskeleton, the Rho GTPases. This was done by measuring Rho GTPase activation in adherent BMDCs in response to the small GTPase-activating toxin CN04. The data show a clear dysregulation in the activity of these signaling molecules in tetraspanin-deficient BMDCs (Fig. 5I–K). After activation with CN04, CD82- and CD37-deficient DCs both showed significant elevations in the amount of active Cdc42 relative to WT. In contrast, RhoA activation was significantly elevated only in CD82^−/− BMDC lysates, whereas the activation of Rac-1 was impaired only in the absence of CD37 (Fig. 5I–K). We conclude that CD82 and CD37 control DC biology through their opposing regulation of cytoskeletal dynamics. CD82 is a negative regulator of RhoA, whereas CD37 promotes Rac-1 activation.

The mechanisms that DCs use to coordinate the simultaneous modulation of Ag processing and presentation, and cellular migration, during their activation are not fully understood. An important article by Faure-André et al. (36) identified the invariant chain as a key molecular link between the Ag presentation machinery and the cytoskeleton. The invariant chain molecularly interacts with MHC II and plays an important role in the trafficking and maturation of MHC II complexes. However, in cell migration, it acts as a negative regulator via its ability to modulate the phosphorylation of myosin L chain and inhibit actinomyosin contraction. In this study, we identify the tetraspanins CD82 and CD37 as novel proteins that also molecularly interact with MHC and play a role in DC biology in linking and coordinating Ag processing and presentation with cell migration.

CD82 expression in mouse BMDCs and human MoDCs is upregulated upon activation (Supplemental Fig. 1); these data agree with published microarray analyses that show that CD82 expression is upregulated upon activation of both mouse splenic DCs (37) and human MoDCs (38) (data are accessible at National Center for Biotechnology Information Gene Expression Omnibus database (39), accession numbers GDS352 and GDS2221). In this study, the increased CD82 expression modifies DC biology in two ways: first, CD82 inhibits DC migration; and second, CD82 is required for DCs to activate naive T cells. In a tissue explant model of skin DC migration and in in vitro chemotaxis and an in vivo migration of BMDCs from the periphery to the draining lymph node, CD82^−/− BMDCs were notably hypermigratory (Fig. 4). This finding is in accordance with the well-established role of CD82 as a metastasis suppressor gene in nonimmune cancers (26, 27). A dysregulation of the Rho GTPases is likely to be a key mechanism that drives this enhancement in DC migration. We observed increased RhoA and Cdc42 phosphorylation in assays of activated CD82^−/− BMDCs, and cell-spreading assays demonstrated an increase in cytoskeletal rearrangement. CD82^−/− BMDCs spread to a greater extent than WT BMDCs and have a relative inability to produce membrane protrusions (Fig. 5). This observation agrees with studies of prostate cancer that also showed a negative regulation of RhoA by CD82 (40), and with signaling studies of T cells that show that CD82 cross-linking leads to RhoA activation (41).

The inability of Ag-pulsed CD82^−/− splenic DCs and BMDCs to stimulate T cells suggests that CD82 expression in DC promotes Ag presentation and productive interactions with Ag-specific T cells. In this study, we observed that CD82 plays an important role in both the processing of MHC II (Fig. 2G, 2H) and particularly the promotion of physical interaction of DCs and T cells (Fig. 3). CD82 molecularly interacts with MHC II both at the cell surface and in MHC II compartments (15–17, 28), and although we can detect no differences in cell-surface expression of MHC in CD82^−/− DCs, pulse chase analyses show a significant, albeit small, defect in the maturation of MHC/peptide complexes (Fig. 2G, 2H). The latter change might reflect the association between CD82 with HLA-DM and HLA-DO (16, 28). The impaired ability of CD82^−/− DCs to physically interact with T cells, like the dysregulation in cell migration, is also likely driven by the dysregulation of cytoskeletal rearrangement and aberrant RhoA and Cdc42 activation. Upon specific interactions between the TCR and peptide-MHC, activated DCs rapidly polarize their cytoskeleton, a process that is essential for immunological synapse formation and the optimal activation of T cells (34, 35, 42). The failure of CD82^−/− BMDCs to form an adequate number of stable conjugates with T cells (Fig. 3), their increased spreading on fibronectin, and their dysregulation of Rho GTPase signaling (Fig. 5) all demonstrate a critical role for CD82 in mobilizing and regulating cytoskeletal rearrangements in DCs during interactions with T cells. That this impairment in cytoskeletal dynamics affects immune responses is supported by the poor cell-mediated immunity induced in CD82^−/− mice (Fig. 1A). Indeed, this poor cell-mediated immunity is consistent with our observations that CD82 ablation affects not only the duration (Fig. 3G–I) of the T cell/DC interaction but also whether the interaction is static or kinetic (Fig. 3J). Static conjugate formation is required to form an initial immunological synapse, which within the OVA system must be at least 30 s (43). A stable, static conjugate influences asymmetry within the T cell, and in turn fate, including differentiation into Th1, Th2, or Th17 for naive CD4⁺ cells, or memory or effectors for naive CD8⁺ cells. Interactions between DCs and T cells, which are transient (44), or where the T cells crawl over the DC surface may lead to tolerance and anergy (45).

A major question in the tetraspanin field concerns functional overlap between family members. Given that tetraspanins often molecularly interact with one another within TEMs, it is not surprising that often tetraspanins share overlapping functions, a classical example being the ability of at least five tetraspanins to negatively regulate T cell hyperproliferation (33). In this article, we show that, at least in splenic DCs and BMDCs, the closely related tetraspanins CD37 and CD82, rather than displaying functional overlap, counteract one another in function. Unlike CD82, which is upregulated upon DC activation, CD37 is expressed predominantly in immature DCs (Supplemental Fig. 1) (23) [see also published microarray analyses that show that CD37 expression is downregulated upon activation in both mouse splenic DCs (37) and human MoDCs (38); data accessible at National Center for Biotechnology Information Gene Expression Omnibus database (39), accession numbers GDS352 and GDS2221]. Analyses of CD37^−/− BMDCs suggest that CD37 expression in immature DCs impairs Ag presentation (Fig. 2) (24) but promotes cell migration (Fig. 4) (14). Although both tetraspanins molecularly interact with MHC (16, 22), the mechanisms by which CD37 regulates Ag presentation differs from that of CD82. Although CD37 clearly regulates the cytoskeleton, as evidenced by an impairment in cell spreading and the formation of membrane protrusions observed in CD37^−/− BMDCs (Fig. 5), this does not affect stable adhesion with T cells because CD37^−/− DCs are able to form conjugates normally (Fig. 3). In contrast, CD37^−/− DCs have a dysregulation in the presentation of peptide/MHC as evidenced by their hyperstimulation of T cell hybridomas, possibly through the display of MHC clusters (24) (Supplemental Fig. 2A). With regard to cell migration and cytoskeletal signaling, in adherent BMDCs, CD37 is essential for the adequate phosphorylation and activation of Rac-1 (Fig. 5J). The impaired cell migration and dendrite formation in the CD37^−/− BMDCs is similar to observations of Rac-deficient DCs (3). However, the inability of Rac-1^−/− DCs to activate naive T cells is in stark contrast with the hyperstimulatory CD37^−/− DCs and suggests that the regulation of Ag presentation by CD37 is either Rac independent or more complex than a mere abrogation of Rac activation. In this study, it is notable that CD37 also negatively regulates Cdc42 (Fig. 5K). CD37 ablation induces in DCs a similar phenotype to that observed when the tetraspanin CD81 has been knocked down through inhibitory RNA (46). Both CD37^−/− and CD81^kd DCs are poor migrators in two dimensions and have a similar inability to form membrane protrusions and activate Rac; indeed, a direct molecular interaction between Rac and CD81 has been reported (11). However, a key difference would appear to be in migration in three dimensions: CD81 deficiency has no effect in in vivo systems (11), in contrast with CD37 ablation that dramatically reduces the ability of DCs to migrate from tissues to draining lymph nodes (14).

Thus, when considering the difference in biology in DCs before and after they are activated by danger signals, the expression of the MHC-interacting tetraspanins CD37 and CD82 is of critical importance. Our analyses of BMDCs show that an early activated DC expresses low levels of CD82, a negative regulator of RhoA activation, but high levels of CD37, a positive regulator of Rac-1, resulting in a highly motile cell with a limited ability to activate naive T cells. By contrast, a late activated DC expresses high levels of CD82 and low levels of CD37 and, therefore, has modified its migratory, cytoskeletal, and Ag presentation machinery to become a cell superbly adapted to activating naive T cells. The next challenge is to determine how strongly the controlled expression of CD37 and CD82 contributes to the coordinated regulation of migration and Ag presentation in classical migratory DCs.

We thank the staff at Clayton and Alfred Medical Research and Education Precinct animal houses for animal care, and Stephen Cody and the staff of Monash Micro Imaging for assistance with setup of live cell microscopy. We thank Dr. Greg Moseley and Aaron Brice for advice on tubulin staining, and Dr. Katrina Binger for critical reading of the manuscript.

The authors have no financial conflicts of interest.