Cutting Edge: DNAX Accessory Molecule 1–Deficient CD8⁺ T Cells Display Immunological Synapse Defects That Impair Antitumor Immunity Free

Cytotoxic CD8⁺ T cells form an integral part of the immune response against pathogens (1) and in the immunosurveillance of cancer (2). The activation of naive CD8⁺ T cells is dependent on TCR–MHC class I interactions with APCs, during which sustained contact between the T cell and APC leads to formation of an immunological synapse. The IS is a polarized structure that brings together the receptors, adhesion molecules, and adaptor proteins necessary for optimal signaling and downstream effector T cell functions such as cytokine secretion and cytotoxic activity (3). The activation and cytotoxic activity of CD8⁺ T cells is fine-tuned by a range of cosignaling molecules, such as members of the CD28 family (4). Another less characterized group of costimulatory molecules important for CD8⁺ T cell function are those of the Ig superfamily of receptors. These receptors bind to nectin and nectin-like proteins, and include DNAX accessory molecule 1 (DNAM-1; CD226, PTA-1) and others (5). DNAM-1 interacts with the polarity protein, Discs Large, and the actin binding protein, 4.1G, to mediate cytoskeletal changes important for cell adhesion (6). DNAM-1 also interacts with the integrin LFA-1 (7–9), an important mediator of stable conjugation and synapse formation between T cells and APCs or target cells (10).

Studies using DNAM-1 gene–deficient mice (DNAM-1^−/−) have highlighted the importance of DNAM-1 in both protection against infectious agents and control of tumor growth. DNAM-1^−/− mice take longer to clear lymphocytic choriomeningitis virus infection (11). DNAM-1 expression on NK cells promotes cytotoxic activity against DNAM-1 ligand-expressing tumor cells (5, 12), whereas DNAM-1^−/− mice show impaired antitumor responses in vivo (13). DNAM-1 ligands are frequently overexpressed by tumor cells (12, 14), and polymorphisms in the DNAM-1 gene are associated with the development of autoimmune diseases (15, 16). However, the mechanism by which DNAM-1 contributes to T cell activation remains unclear. DNAM-1 may be involved in direct T cell signaling pathways or the cell–cell interactions required for optimal T cell activation and effector function. We therefore investigated the response of CD8⁺ T cells from DNAM-1–deficient mice to Ag presentation and their interaction with tumor cells in vitro. Our results demonstrate that DNAM-1 is required for optimal activation of naive CD8⁺ T cells, and for efficient conjugation and synapse formation between effector CD8⁺ T cells and tumor cells.

The C57BL/6 DNAM-1^−/− mice were kindly supplied by Marco Colonna (Washington State University School of Medicine, St. Louis, MO) and backcrossed onto the T cell transgenic OT-1 mouse background (Peter MacCallum Cancer Centre). OT-1 DNAM-1^+/+ and OT-I DNAM-1^−/− littermates were used for all experiments. The Abs for flow cytometry were rat anti-Vα2, -CD62L, -CD25, -CD69, -CD44, -CD127, -CD107a, mouse anti–granzyme B, –IFN-γ and –TNF-α (BD Biosciences), and anti-DNAM-1 (Biolegend); for microscopy, they were purified rat anti–LFA-1 and -CD43 (BD Biosciences), phalloidin-rhodamine (Molecular Probes), rabbit anti-GM1 (Calbiochem, MERK Millipore, Darmstadt, Germany), and rabbit and rat anti-tubulin (Rockland). Secondary Abs conjugated to Alexa Fluorophores and ProLong antifade with DAPI were purchased from Molecular Probes.

Naive T cells were purified from splenocytes isolated from OT-1 DNAM-1^+/+ or DNAM-1^−/− mice by negative selection using an EasySep magnetic separation system (Stemcell Technologies, Vancouver, BC, Canada). Immature dendritic cells (DCs) were generated as previously described (17). For proliferation assays, isolated naive T cells were labeled with CFSE and incubated with 1 μM SIINKEL-pulsed DCs. The cells were harvested and analyzed for proliferation peaks by flow cytometry or for the concentration of cytokines in the supernatant using a Mouse Inflammation Cytometric Bead Array kit (BD Biosciences) after 72 h. Intracellular staining was performed using Cytofix/Cytoperm reagents (BD Biosciences). Cytotoxic activity of activated T cells was measured by a standard chromium release assay (18) using chromium-labeled MC38-OVA tumor cells or labeled/SIINFEKL-pulsed DCs as targets. For the conjugation assay, activated T cells (day 5) were labeled with CellTrace Violet Dye (Molecular Probes) and incubated with CFSE-labeled, SIINFEKL-pulsed EL4 tumor cells at 37°C. The cells were vortexed to separate any non-Ag-specific conjugates, and the cells were fixed in 4% paraformaldehyde and analyzed by flow cytometry for the presence of double-positive conjugates.

MC38-OVA tumor cells were adhered overnight onto eight-well chamber slides (Nalge Nunc, Rochester, NY). Activated T cells were overlaid for 1–2 h and nonadherent cells washed off. Cells were fixed and permeabilized and then labeled with primary Abs, followed by detection with Alexa Fluor–conjugated secondary Abs and mounted in Prolong antifade containing DAPI (Molecular Probes), as previously described (17). The slides were examined using a Fluoview FV1000 confocal microscope (Olympus, NY) mounted with a 60× oil immersion objective (NA 1.42). T cells selected for protein scoring had a single contact site with one tumor cell and polarized microtubule-organizing center (MTOC). Polarization to the synapse was determined by quantitating the amount of nonnuclear fluorescence in the proximal (P) region (first third of T cell closest to tumor cell) and distal (D) region (rest of T cell), and the ratio was determined by P-D/P+D where values approaching 1 represent strong polarization to the interface of the two cells (Supplemental Fig. 2C). Data were acquired using Matlab and TACTICS software (19).

Statistical significance was determined using an unpaired Student t test.

Because DNAM-1 has been implicated in impaired T cell responses to infection in vivo (11), we first determined whether naive T cells deficient for DNAM-1 responded normally to Ag presentation. We used naive CD8⁺ T cells from OT-I DNAM-1 wild-type (DNAM-1^+/+) or deficient (DNAM-1^−/−) mice to assess proliferation and cytokine production in an Ag-specific response to SIINFEKL-pulsed syngeneic DCs in vitro. Although naive DNAM-1^−/− T cells displayed a wild-type phenotype (CD62L^hiCD44^medCD25⁻CD69⁻; Supplemental Fig. 1A), they showed reduced proliferation, as measured by dilution of the division tracking dye CFSE, compared with DNAM-1^+/+ T cells when stimulated with peptide-pulsed DCs (Fig. 1A). We also detected lower amounts of two key CD8⁺ T cell effector cytokines, IFN-γ and TNF-α, in the supernatant of DNAM-1^−/− T cell–DC cocultures (Fig. 1B). This was a consequence of reduced T cell numbers, because DNAM-1^−/− and DNAM-1^+/+ T cells produced equivalent levels of intracellular IFN-γ and TNF-α upon stimulation with peptide-pulsed DCs (Supplemental Fig. 1C). This is consistent with previous data demonstrating that during acute lymphocytic choriomeningitis virus priming, CD8⁺ T cells from DNAM-1^−/− mice make equivalent levels of IFN-γ compared with DNAM^+/+ T cells (11). The reduced proliferation was not a result of TCR signaling defects, because anti-CD3/CD28 Ab stimulation resulted in proliferation and cytokine levels equivalent between DNAM-1^+/+ and DNAM-1^−/− T cells (Supplemental Fig. 1D, 1E). These results indicate that expression of DNAM-1 on CD8⁺ T cells contributes to optimal T cell activation and proliferation in response to Ag presentation by DCs.

Despite a reduction in proliferation and differentiation kinetics, a proportion of DNAM-1^−/− T cells were activated and proliferated in vitro. To determine whether this change in activation kinetics affected the phenotype or polarity of responding T cells, we first used flow cytometry to assess the cell-surface profile of the activated T cells. Activated DNAM-1^−/− T cells upregulated CD44, CD69, and CD25, and downregulated CD62L to equivalent levels as DNAM-1^+/+ T cells (Supplemental Fig. 1B). The activated DNAM-1^−/− T cells also displayed no detectable defects in polarization of actin and LFA-1 to the leading edge, CD43 to the uropod, and localization of the MTOC to the base of the uropod, as assessed by confocal microscopy (Fig. 1C). However, activated DNAM-1^−/− T cells did show impaired cytotoxic activity against peptide-pulsed DCs (Fig. 1D). Both DNAM-1^−/− and DNAM^+/+ T cells expressed equivalent levels of the cytotoxic molecule, Granzyme B, and showed no difference in degranulation after TCR stimulation (Supplemental Fig. 2A, 2B). Together, these data suggest that the DNAM-1 is critical for the cell–cell contact required for efficient killing of target cells by CD8⁺ T cells.

To explore potential reasons for the defect in killing shown earlier, we next determined whether activated DNAM-1^−/− T cells display normal cytotoxic activity against tumor cells that express the ligands for DNAM-1 (CD155 and CD112) (5). Mice deficient for DNAM-1 have impaired antitumor responses using models dependent on both NK and CD8⁺ T cell activity (5, 20). Furthermore, the association of DNAM-1 with LFA-1 and proteins that mediate cytoskeletal changes in T cells such as Dlg and 4.1G (6, 21) suggests a role for this accessory molecule in cell–cell interactions. We therefore assessed the ability of the DNAM-1^−/− T cells to form conjugates with tumor cells in vitro using a flow cytometry–based conjugation assay. We found that activated DNAM-1^−/− effector T cells formed significantly less conjugates than DNAM-1^+/+ T cells (Fig. 2A, quantitated on right), confirming that the absence of DNAM-1 impairs the stable interaction between T cells and tumor cells. Furthermore, consistent with our previous data showing that DNAM-1^−/− mice do not clear MC38-OVA tumors in vivo (22), the DNAM-1^−/− T cells showed reduced cytotoxic activity against MC38-OVA tumor cells in vitro, as determined by a chromium release assay (Fig. 2B). These results demonstrate that DNAM-1 expression on CD8⁺ T cells is critical for efficient conjugation and killing of Ag-presenting tumor cells.

To determine whether reduced cytotoxic activity of the DNAM-1^−/− T cells against MC38-OVA tumor cells was a result of impaired cytotoxic synapse formation between the T cells and tumor cells, we used immunofluorescent microscopy to dissect the polarization of synapse components during T cell–tumor cell interactions. Despite fewer conjugates being formed, in the DNAM-1^−/− T cells that attached to a tumor cell, we found the MTOC was recruited to the synapse, along with enrichment of actin at the interface of the two cells. However, we observed significantly impaired polarization of LFA-1 and lipid rafts (as determined by Ganglioside, GM-1, localization) to the interface between the OT-1 DNAM-1^−/− T cells and tumor cells (Fig. 2C and quantified in Fig. 2D).

Formation of a “suboptimal” cytotoxic synapse, and subsequent inability to secrete effector molecules in a directed manner, may therefore account for the reduced cytotoxic activity of DNAM-1^−/− T cells against tumor cells in vivo. Strong interactions between T cells and target cells is dependent on cell-surface molecules such as TCR and LFA-1, but is also controlled by cytoskeletal dynamics and the integrity of lipid rafts (23). The reduced polarization of both LFA-1 and lipid rafts containing costimulatory molecules such as DNAM-1 (9, 24) would therefore impact on conjugation efficiency and in the activation and cytotoxic activity of the T cells. Indeed, DNAM-1 is important for the LFA-1 costimulatory signal necessary for naive CD4⁺ T cell differentiation and proliferation (8). LFA-1–deficient mice show normal cytotoxic T cell responses to virus but fail to reject immunogenic tumors (25), suggesting that LFA-1 may also play a significant role in the interaction of CD8⁺ T cells with tumor cells.

Together, our results suggest that DNAM-1 deficiency affects T cell function in a 2-fold manner: 1) reduced activation of responding effector T cells, and 2) reduced ability of activated T cells to form an IS with target cells and deliver the cytotoxic payload necessary for elimination of target cells. Furthermore, our results demonstrate that these defects are manifested through suboptimal formation of a cytotoxic synapse involving reduced polarization of LFA-1 and lipid rafts. These results highlight the importance of DNAM-1 in the control of optimal synapse formation required for effective antitumor T cell responses.

We thank Marco Colonna (Washington State University School of Medicine) for the DNAM-1 mice and Nicole Haynes (Peter MacCallum Cancer Centre) for the MC38-OVA cell line.

The authors have no financial conflicts of interest.