CD2 molecules redistribute to the uropod during T cell scanning: Implications for cellular activation and immune surveillance

Abstract

Dynamic binding between CD2 and CD58 counter-receptors on opposing cells optimizes immune recognition through stabilization of cell–cell contact and juxtaposition of surface membranes at a distance suitable for T cell receptor–ligand interaction. Digitized time-lapse differential interference contrast and immunofluorescence microscopy on living cells now show that this binding also induces T cell polarization. Moreover, CD2 can facilitate motility of T cells along antigen-presenting cells via a movement referred to as scanning. Both activated CD4 and CD8 T cells are able to scan antigen-presenting cells surfaces in the absence of cognate antigen. Scanning is critically dependent on T cell β-integrin function, as well as myosin light chain kinase. More importantly, surface CD2 molecules rapidly redistribute on interaction with a cellular substratum, resulting in a 100-fold greater CD2 density in the uropod versus the leading edge. In contrast, no redistribution is observed for CD11a/CD18 or CD45. Molecular compartmentalization of CD2, T cell receptor, and lipid rafts within the uropod prearranges the cellular activation machinery for subsequent immune recognition. This “presynapse” formation on primed T cells will likely facilitate the antigen-dependent recognition capability required for efficient immune surveillance.

Arrival of antigen-laden dendritic cells into the lymph node results in the activation of naive T cells, leading to their proliferation and acquisition of effector functions (1). The subsequent migration of activated T cells into pathogen-invaded body tissues is a key feature of immunity and immune surveillance. Diapedesis of T cells through the vascular endothelium, as well as their crawling movement on the surface of antigen presenting cells (APCs), parenchymal cells and extracellular matrix (ECM), requires T cell polarization (2). Although the molecular basis of this process is not well understood, certain features have been identified. Cell polarization is characterized by the formation of a leading edge at the front of the cell and a uropod at its back, allowing conversion of mechanical forces generated through adhesion to a substratum into net cellular locomotion (3). While the actin cytoskeleton redistributes at the leading edge, the microtubule organizing center (MTOC) localizes within the uropod allowing a greater deformability of the migrating cells (4–6). Given that the MTOC and the Golgi apparatus colocalize, the secretion of soluble mediators is directed toward the ligated cell surface (7). Moreover, chemokine receptors responsible for the directionality of the migration along a gradient of chemoattractants are concentrated within the leading edge (8). In contrast, the uropod mediates important adhesion functions, as suggested by accumulation of a notable number of adhesion molecules including ICAM-1, CD44, CD43, P-selective glycoprotein ligand 1 (PSGL-1), and l-selectin during migration (9, 10). The uropod therefore appears to anchor the cell to the ECM, APCs, or endothelial cells while the leading edge propels the cell forward. It has also been suggested that adhesion molecules within the uropod support binding to other bystander leukocytes, thereby enhancing inflammatory cell recruitment (9).

Human CD2 is a transmembrane cell surface glycoprotein expressed in a restricted manner on T cells, thymocytes, and NK cells, whereas its counter-receptor CD58 is expressed on a diverse array of nucleated and non-nucleated cells including APCs and stromal cells (reviewed in refs. 11 and 12). CD2 functions in both T cell adhesion and activation processes (13). Of note, the weak affinity of the CD2-CD58 interaction (K_d ≈ 1 μM) is associated with remarkably fast on and off rates that foster rapid and extensive exchange between CD2 and CD58 partners on opposing cell surfaces (14–16). These biophysical characteristics are reminiscent of the selectin–ligand interactions involved in the rolling processes along endothelial cells before extravasation (17). Recent x-ray crystallographic analysis of the human CD2-CD58 complex explains this behavior by demonstrating a very favorable charge complementarity at the molecular interface in the presence of few hydrophobic contacts (18). The cytoplasmic tail of CD2 is linked to the cytoskeleton via a variety of adapters that influence cytoskeletal polarization, adhesion, and activation, raising the possibility that CD2 may regulate T cell polarization in a key manner (19–21). Furthermore, given the clustering of CD2-CD58 counter-receptor pairs in the T cell–APC contact zone, which positions cell membranes at a distance of 15 nm suitable for subsequent T cell receptor (TCR)–peptide MHC interaction, it is not surprising that CD2-CD58 counter-receptors enhance the efficiency of immune recognition (20, 22–24).

Here we use time-lapse video microscopy and image analysis to examine the role of CD2 in T cell motility, polarization, and immune surveillance. We demonstrate that CD2 ectodomain crosslinking results in T cell polarization with attendant redistribution of CD2 to the uropod along with the TCR and lipid raft microdomains. This highly specified redistribution process prearranges the activation machinery for subsequent antigen recognition. The distinctly different behavior of CD11a/CD18 on these same T cells implies that the critical role of β2 integrins in T cell adhesion, migration, and immune activation are nonredundant with those of CD2.

Materials and Methods

Cell Purification and Culture.

Human peripheral blood mononuclear cells (PBMC) were purified from leukopaks by using a Ficoll gradient. T cells were isolated using the nylon wool method, cultured in RPMI medium 1640 supplemented with 10% human AB serum (Nabi, Miami), 2 mM l-glutamine (GIBCO), and penicillin/streptomycin (GIBCO; complete medium), and activated for 72 h in the presence of 25 ng/ml PMA (Sigma) and 1:200 anti-CD3 mAb (2Ad2, IgM ascites). CD4 and CD8 T cell subsets were isolated by negative selection with resulting purity >95%. Human monocytes were isolated from purified PBMC by using magnetic beads coated with anti-CD14 (Miltenyi Biotec, Auburn, CA; ref. 25). Immature dendritic cells were obtained by culturing CD14⁺ monocytes with granulocyte/macrophage colony-stimulating factor (GM-CSF) and IL-4 at 2 ng/ml for 9 days. J77, a clone of Jurkat cells, was maintained in RPMI medium 1640 supplemented with 10% FCS (Sigma), 2 mM l-glutamine, and penicillin/streptomycin (GIBCO). CHO58 and additional Chinese hamster ovary (CHO) transfectants were produced and maintained as described (26).

Preparation of the APCs and Time-Lapse Experiments.

The APCs were plated on sterile circular cover slips (Fisher Scientific) in 6-well plates at ≈1 × 10⁶ cells per well and cultured overnight at 37°C to obtain a 95% confluency. The cell-coated coverslips constituted the base of a chamber (Micro Video Instruments, Avon, MA) connected to a source of 10% CO₂ balanced air (BOC Gases, Hingham, MA) and a brass flow meter (BOC Gases) to maintain the pressure at 130 mm. The chamber was placed on a 37°C heating stage. Activated T cells (3 × 10⁵) suspended in 3 ml of complete medium were inserted into the chamber and allowed to settle for ≈10 min before the initiation of recording. The frames were captured using a SPOT RT camera (Diagnostic Instruments, Sterling Heights, MI) connected to a Nikon Diaphot 300 fluoro-microscope (Micro Video Instruments). A user-designed program allowed capture of frames of the selected field in differential interference contrast (DIC) or in fluorescence with an interval of 30 s and over a period of 1 h. All of the images were processed using the public domain nih image program (National Institutes of Health) and the iplab software (Scanalytics, Fairfax, VA). In blocking experiments, the APCs were preincubated with anti-CD58 (TS2/9, 200 μg/ml) for 30 min at 4°C, whereas T cells were preincubated with anti-CD11a (TS1/22, 50 μg/ml, IgG1; Endogen, Cambridge, MA), anti-CD18 (TS1/18, 30 μg/ml, IgG1; Endogen), and anti-CD3 (RW2–8C8, 10 μg/ml) for 30 min at 4°C. T cells were also preincubated with the kinase inhibitors 50 μM PD98059, 10 μM SB203520, and 10 μM ML-7 (Calbiochem).

Transfection of Murine L Cell Fibroblasts.

Murine L cell fibroblasts were transfected with human GPI-linked CD58 (27) in pSH-SX vector and human ICAM-1 (CD1.8 plasmid was kindly provided by T. Springer, Harvard Medical School) subcloned into pIRES2-EGFP vector (CLONTECH). Cells (10⁶) were plated the day before the transfection and 20 μg of each individual plasmid were transfected using the calcium phosphate method. After 48 h, the cells were put under selection in the presence of 1 mg/ml G418 (GIBCO) and/or 1 mg/ml Hygromycin B (Roche Diagnostics GmbH) for 4 days. The positive cells were sorted on the MoFlo (Cytomation, Fort Collins, MO) and maintained at 0.2 mg/ml G418 and/or 0.1 mg/ml hygromycin.

Immunofluorescence of T Cells and Analysis of Fluorescence Intensity.

Activated T cells (2–3 × 10⁶) were suspended in 50 μl of PBS + 2% FCS and stained for CD2 (T11–2 Texas red mAb, 1.6 μg), for CD45 [KC56 (T-200)-FITC (Coulter) 8 μl], for CD11a/CD18 (TS2/4 Texas red, 2.4 μg), for CD3 (RW2–8C8 Texas red mAb, 4.8 μg), and for Raft GM-1 ganglioside [Cholera toxin B subunit (CTx)-biotin (Sigma), 4 μg, detected by Streptavidin Texas red (Jackson Immunoresearch lab 1.8 μg] for 30 min at 4°C. After washing, 3 × 10⁵ stained cells suspended in 3 ml of complete medium were incubated in the chamber on a monolayer of CHO58 cells. Scanning T cells were followed over a period of 15 min to 1 h, recording DIC and fluorescence images at intervals of 30 s.

To evaluate different T cell compartments, regions of 225 pixels (2) were selected on the DIC image of the uropod or of the leading edge of a given cell and overlayed on the corresponding fluorescent image. Five areas surrounding the cells of interest were measured and the intensity values averaged to determine the mean intensity value of the background [I(bkg)]. For each of the 12 cells analyzed per molecule, the I(bkg) value was subtracted from the fluorescence intensity value of the leading edge and of the uropod. The values obtained were summed and the uropod/leading edge ratio calculated as ∑ [I(uropod) − I(bkg)]_n/∑ [I(leading edge) − I(bkg)]_n.

Polarization and Change of Morphology of T Cells On CD2 Crosslinking.

Glass cover slips (Clay Adams) were incubated overnight at 4°C with the indicated mAbs at 5 μg/ml in PBS and then washed with PBS to remove the excess mAbs. Activated T cells (5 × 10⁶) were resuspended in complete medium and plated on the mAbs-coated cover slips. After incubating at 37°C for 1–5 h, the coverslips were washed with PBS and mounted on microscope slides (Fisher). To examine the Golgi apparatus distribution, Jurkat 77 cells at 4 × 10⁶ per ml in PBS + 5% FCS were incubated for 30 min at 37°C on the mAb-coated cover slips, washed twice, and then fixed with 3.7% paraformaldehyde in PBS for 1 h at 4°C. Following further washing, the cells were permeabilized with 0.1% saponin/1% BSA in PBS for 30 min at 4°C and stained with 0.5 μM BODIPY TR ceramide (Molecular Probes) for 1 h at 4°C. Subsequently, the cells were washed twice in permeabilization buffer and mounted on slides (Fisher) for analysis.

Results

Polarization of T Cells by CD2 Crosslinking.

Previous analysis of T cell–APC interaction showed that CD2 plays an important role in immune recognition (23, 28). CD2–CD58 interaction contributes directly to stabilized conjugate formation between the T cell and CD58-bearing APC, thereby reducing the molar requirement for foreign nominal peptide antigen by >1 order of magnitude (23). As shown in Fig. 1, conjugate formation between human T cells and hCD58-expressing CHO cells (CHO58) is antigen-independent and results in redistribution of CD2 to the interface of the cell conjugate. In three T cells at the bottom of Fig. 1B and in the two T cells interacting with the upper CHO cell, CD2 is redistributed almost exclusively along the CHO58–T cell junction. In contrast, CD2 maintains its uniform distribution on the T cell not contacting the APC (upper left). This result is comparable to that reported previously between human T cells and CD58-transfected murine fibroblasts (23). In the present study, we address whether the redistributed CD2 is localized to a specific subcellular compartment and whether this reorganization is linked to cell polarization and, perhaps more importantly, to T cell motility.

Figure 1.

CD2 redistributes to the interface of T cell/APC conjugates. (A) DIC image of the conjugates between CD58 transfected CHO cells (labeled CHO) and T cells (unlabeled). (B) Immunofluorescence staining of CD2 on the same conjugates as in A. (C–F) Polarization of activated T cells induced by mAbs over 1–5 h: (C) anti-MHC class I mAb (W6/32; ref. 54), 1 h, 37°C; (D) anti-CD3ɛ mAb (RW2–8C8; ref. 55), 1 h, 37°C; (E) anti-CD2 (anti-T11₂ + anti-T11₃; ref. 13), 1 h, 37°C; (F) anti-CD2 (anti-T11₂ + anti-T11₃), 5 h, 37°C. (Insets) Staining of the Golgi apparatus on Jurkat 77 cells plated on the above mAbs coated cover slips. (Scale bar, 20 μm.)

To assess whether T cell morphological changes are induced on CD2 crosslinking consistent with a polarized phenotype, activated T cells were incubated on cover slips coated with anti-CD2 or unrelated mAbs at 37°C for 1–5 h. As shown in Fig. 1E, activated human T cells undergo a dramatic change in cell shape on incubation with anti-CD2 mAbs for 1 h at 37°C. As described below, this phenotype is associated with increased locomotion/migration. After 5 h of incubation, polarization becomes even more striking, as demonstrated by formation of a very elongated uropod (trailing edge) that can extend more than 60 μm in length (Fig. 1F). Polarization appears to be specifically induced by CD2 crosslinking because neither anti-MHC class I nor anti-CD3ɛ mAb-coated surfaces induce a similar effect at 1 h (Fig. 1, C and D, respectively) or any time interval examined (data not shown). To independently evaluate the ability of CD2 clustering to induce T cell polarization, we further examined the distribution of the Golgi apparatus in T cells after mAb-mediated crosslinking. For clarity of observation, the larger-size, cytoplasm-rich Jurkat J77 T cells were used. As shown in the Insets in Fig. 1, anti-CD2 but not anti-MHC class I or anti-CD3ɛ mAb crosslinking results in Golgi reorganization/polarization. Hence, staining of the Golgi apparatus of J77 cells after interaction with anti-CD2 mAb-coated cover slips shows a clear Golgi polarization toward the center of the cell (Fig. 1E Inset) compared with its random position in J77 cells bound to anti-MHC-class I or anti-CD3ɛ mAb-coated surfaces (Fig. 1, C and D Insets, respectively). Both the cell migration and Golgi polarization are observed when cells are incubated with anti-T11₂ mAb alone. This result rules out the possibility that the effects are mediated by a mitogenic stimulus; anti-T11₂ alone unlike the anti-T11₂ + anti-T11₃ combination does not induce T cell proliferation, as judged by ³H-TdR incorporation (13).

T Cell Scanning on a Cellular Substratum.

We next performed time-lapse video microscopy analysis to examine the role of the CD2–CD58 interaction on T cell migration on a cellular monolayer consisting of CHO58 cells. Several observations are apparent. First, ≈50% of substrate-bound T cells scan CHO58, whereas very little or no scanning is observed on a substratum of CHO cells alone or CHO cells transfected with mutant CD58 molecules defective in CD2 binding (see below). Second, individual cell migration behavior on CHO58 cells is heterogeneous; some T cells scan continuously along the APC surface, whereas others stop and start, often reversing direction. In Movies 1 and 2 (which are published as supporting information on the PNAS web site, www.pnas.org), we highlight the migration of several cells to document this point. Third, although not shown, both purified activated CD4 and CD8 T cells demonstrate scanning at comparable levels, as do CD45RA and CD45RO subpopulations (29). Fourth, resting T cells failed to mediate scanning activity, indicating a clear requirement for cellular activation as might occur in secondary lymphoid organs. Cell motility is a dynamic process involving lamellipodia formation at the leading edge, attachment to the substratum (30, 31), generation of contractile forces and traction (3), and ultimate release of the uropod (6).

In this paper we define “scanning” as the migration of T cells on the surface of a cellular monolayer substratum. As examples, we show the trajectory covered by two T cells (Fig. 2; see Movies 3 and 4, which are published as supporting information on the PNAS web site). In Fig. 2A (Movies 3), the cell enters from the bottom of the field as it scans the edges of three CHO58 cells while continuing toward the right of the field. The cell then stops, inverts its direction, and proceeds to scan two other CHO58 cells during the time of observation interval. In Fig. 2B (Movie 4), the T cell arrives at the top of the field. During its subsequent scanning activity, the uropod remains attached to one CHO58 cell, thereby tethering the T cell and pulling it back; later, the T cell continues the scanning toward the right of the field. Note how the T cells invariably round up before altering their direction of migration.

Figure 2.

Representative trajectories of T cells scanning the surface of a cellular substratum. A and B delineate the pathway of two individual T cells along the CD58 CHO epithelial surface. Yellow dots represent the relative position of the analyzed cell at each 2 min interval and arrows show the direction of scanning. Accompanying digitized movies are included (Movie 3 and 4). C and D are graphical representations of the surface area (blue, μm²) and distance traversed (pink axis, μm) per unit of time (min) during the scanning function of cells in A and B, respectively. The analysis is performed using nih image software.

Analysis of the same two T cells in terms of distance traversed per unit time is given in Fig. 2 C and D. The graphs show that, even in the event of continuous cell motion, the distance covered by the centroid of the T cell per unit time varies within a range of 1.5–8 μm/0.5 min, suggesting abrupt acceleration and deceleration intervals rather than a constant velocity. Furthermore, we measured the surface area of the T cells during scanning and observed a positive correlation between the surface area of the cells and the distance traversed. This finding suggests that the polarized phenotype is primarily associated with locomotion. In the absence of movement or on trajectory inversion, redistribution of intracellular components, including the actin cytoskeleton, occurs with attendant reduction in cell surface area and loss of traction forces required to maintain polarity. The pattern shown for these two cells is representative of 20 other cells including both continuous vs. intermittent scanning types (Movie 1) and “top” vs. “under” scanners (Movie 2). The process of cells navigating under cellular substratum has been termed pseudoemperipolesis (32).

Role of CD2 and CD11a/CD18 in T Cell Scanning.

To directly investigate the role of CD2–CD58 interaction in scanning, we adopted two different approaches. The first was to preincubate the APC monolayer with anti-CD58 mAb and then plate T cells onto the pretreated CHO58 substratum. One of the parameters affected by this treatment is the percentage of scanning T cells bound to the CHO58 monolayers. In Fig. 3A, we show that the percentage of scanning T cells significantly decreases from 50% [n (number of cells analyzed) = 282] on the untreated CHO58 substratum (Movie 5) to 30% (n = 88) on anti-CD58 mAb-treated CHO58 cells. The second approach was antibody-independent, using two mutant forms of human CD58 (K34A and K87A) with reduced CD2 binding activity (33). The analysis of the percentage of T cell scanning shows a reduction to 23% on K34A (n = 148) and to 11% on K87A mutants (n = 27; Fig. 3A). Note how CHO transfectants of CHO58 variants, E76A and K50A, involving mutations located outside the CD2 binding site (18), have no influence on scanning activity [E76A (53%, n = 57); K50A (57%, n = 21)]. Collectively, these findings show that blocking CD2–CD58 interaction significantly impairs the percentage of scanning T cells without affecting the adhesion to the CHO58 cells via the uropod, implicating additional molecules in local adhesion.

Figure 3.

T cell scanning on all APC surfaces is critically dependent on β2 integrin function. Three cellular substrata were examined: CHO (A), human dendritic cells (B), and murine L cell transfectants (C). The percentage of T cells scanning over the total number of APC bound T cells are designated on the y axis. In A, treated or untreated T cells were plated on CHO58, mAb treated CHO58, or mutant CHO cells as described in Materials and Methods. In B, DC were untreated (CTR) or pretreated with anti-CD58 or with anti-CD18 mAb. In C, the scanning substratum consists of untransfected, hCD58-transfected, or hICAM-1-transfected L cells. **, P ≤ 0.001; *, P < 0.05; as calculated by one-sided χ² test and by Fisher's exact test.

CD11a/CD18 is the most abundant β2 integrin on the T lymphocyte surface and binds to ICAM-1 and -2 on stromal cells (34, 35). Integrins play an important role in the adhesion processes between leukocytes and other cells, as well as to extracellular matrix (ECM) proteins (12). The strong and stable interactions between integrins and their ligands are primarily responsible for the termination of leukocyte rolling on endothelial cell surfaces before extravasation (36). Blocking experiments using T cells pretreated with anti-CD18 and anti-CD11a mAbs demonstrate that β2 integrins and, in particular, CD11a/CD18 are indispensable for scanning activity. As shown in Fig. 3A, the percentage of T cells scanning is reduced to 7% on anti-CD18 mAb treatment (n = 106) and to ≈10% on anti-CD11a mAb treatment (n = 39). In contrast, following pretreatment with anti-CD3ɛ mAb, the percentage of scanning T cells is comparable to that of the untreated control, implying that TCR complex crosslinking does not affect scanning function.

Myosin Light Chain Kinase (MLCK) and T Cell Motility.

An important aspect of cell motility is the myosin-dependent cell contraction (37). The modulation of myosin light chain (MLC) phosphorylation through the regulation of MLCK and MLC phosphatase mediated by GTPases is a key event in the generation of contractile forces. A role for mitogen-activated protein kinase (MAPK) in the regulation of MLCK and in cell migration has been suggested (38). As CD2 ectodomain clustering on peripheral T lymphocytes activates MAPK (39), we investigated the involvement of the ERK pathway in the migration of T cells on the CHO58 surface. Inhibition of MAPK kinase (MEK) upstream of ERK by using an inhibitor (PD98059; ref. 40) reduces the percentage of T cell scanning to 32% (n = 38). Because PD98059 pretreatment blocks CD2-mediated activation of the MAPK without effecting CD2-induced adhesion (41), we assume that this effect is a consequence of impairment in concomitant integrin signaling (42). Moreover, blocking MLCK downstream of ERK by using a specific inhibitor (ML-7) reduces the percentage of T cells scanning to ≈18% (n = 61; Fig. 3A). This result is consistent with the hypothesis that the phosphorylation of MLC is essential for the assembly of the contractile migration machinery. We also investigated the involvement of p38 MAPK. On treatment with a specific inhibitor (SB203580; ref. 43), we observe a reduction in the percentage of T cells scanning (29%, n = 34) comparable to that obtained with the MEK inhibitor (Fig. 3A). From these data we conclude that the ERK and p38 MAPK pathways are linked to T cell motility.

CD18 Is Essential for Scanning of APCs by T Cells.

The ability of anti-CD18 mAb to affect T cell scanning of CHO58 more than anti-CD58 mAb blockade suggests that β2-integrins may be essential for cell locomotion on other cellular monolayers. We therefore performed time-lapse experiments using human dendritic cells as APCs. As shown in Fig. 3B, 76% (n = 104) of T cells are able to migrate on dendritic cell monolayers with an average speed of 8 μm/min (see Movie 6, which is published as supporting information on the PNAS web site). While the preincubation of the APCs with anti-CD58 mAb only modestly (but with statistical significance) affects the percentage of scanning T cells (63%, n = 82), the preincubation of the T cells with anti-CD18 mAb markedly reduces the percentage of scanning to 4% (n = 135; see Movie 7, which is published as supporting information on the PNAS web site). Furthermore, anti-CD58 preincubation does not significantly alter the speed of T cell migration (7 μm/min). As an independent confirmation of the importance of β2-integrin function in T cell scanning, we analyzed T cell movement on monolayers consisting of untransfected, CD58- or hICAM-1-transfected L cells. In Fig. 3C we show that the percentage of T cells scanning on a monolayer of hICAM-1-transfected L cells is ≈46% (n = 648; see Movie 8, which is published as supporting information on the PNAS web site), whereas there is no significant difference between the percentage of T cells scanning on CD58-transfected L cells (8%, n = 181; see Movie 9, which is published as supporting information on the PNAS web site) and on untransfected L cells (6%, n = 53). Note that FACS analysis on L cells shows that neither murine ICAM-1 nor murine CD48 molecules are expressed (data not shown). From these results, we conclude that a productive CD11a/CD18-ICAM-1 interaction is critical for migration/scanning function of T cells on cellular monolayers, whereas the CD2–CD58 interaction is important in a more restricted way—i.e., only being key on certain APC substratum.

Differential Redistribution of Molecules to the Uropod During T Cell Scanning: Formation of a Preactivation Complex.

The requirement for productive CD2–CD58 interaction during cognate antigen-triggered T cell activation is well established (23, 24, 27). The limited role for CD2 in T cell scanning of certain APC, therefore, suggests that a linked, yet distinct, CD2 function may be critical for its participation in immune recognition. To this end, we investigated the distribution of surface molecules on T cells during scanning, analyzing the disposition of CD2, CD3, and lipid rafts, as well as the tyrosine phosphatase, CD45 (44), and the integrin CD11a/CD18. In Movie 10, a single T cell is followed over a period of 30 min. As shown, CD2 molecules become exclusively localized to the uropod. Fig. 4A (a–d) represents still images derived from the time-lapse sequence. Fig. 4A (e–h) shows the distribution of the CD45 molecules on various T cells at different stages of polarization. In contrast with the discrete localization of CD2 in the uropod, the CD45 is present in a more uniform way over the cell surface and, hence, detected at the leading edge as well as at the uropod. The same diffuse pattern is shown for CD11a/CD18 (Fig. 4A, i–l). This system also allowed us to investigate the distribution of molecules important in T cell activation, including the TCR complex itself and lipid microdomains (lipid rafts) before TCR engagement. In a fashion similar to CD2, CD3 molecules are redistributed primarily in the uropod (see Movie 11, which is published as supporting information on the PNAS web site, and Fig. 4A, m–p). Moreover, the same uropod-skewed distribution is observed for lipid rafts themselves as revealed by CTx-biotin/streptavidin TR (see Movie 12, which is published as supporting information on the PNAS web site, and Fig. 4A, q–t). From these observations, we conclude that the essential activation machinery represented by the TCR complex, those lipid microdomains binding CTx and associated signaling molecules as well as CD2, are recruited to the same locale before TCR engagement. Fig. 4B offers a quantitative representation of the ratio of the fluorescence intensity in the uropod and in the leading edge for these various structures. CD2 and CD3 molecules concentrate 10–100-fold in the uropod as determined with specific mAb probes. Likewise, rafts are concentrated ≥10-fold based on CTx localization. In contrast, CD11a/CD18 and CD45 show no quantitative difference.

Figure 4.

Redistribution of CD2 molecules to the uropod during scanning and formation of a “preactivation” complex. (A) Cellular localization of surface molecules on T cells plated on a monolayer of CHO58 cells. Activated T cells were stained for CD2, CD45, CD11a/CD18, CD3, or Raft GM-1, plated on a monolayer of CHO58 cells and images of T cells at different stages of polarization were recorded. (a–d) Sequential images at the indicated times in minutes following T cell binding to the cellular substratum. Other panels represent independent cells. (B) Graphical representation of the ratio between the fluorescence in the uropod and in the leading edge calculated for each of the indicated molecules. **, P < 0.001, as calculated by two-sided Wilcoxon rank sum.

Discussion

An important feature of CD2 ectodomain ligation noted in the current study is the induction of T cell polarization and motility. Thus, on anti-CD2 mAb binding or on contact with a CD58-expressing cell monolayer, a T lymphocyte transforms itself from a round cell into an elongated motile counterpart with a clearly delineated leading edge and uropod. Both activated CD4 and CD8 T cells demonstrate this CD2-inducible morphological transformation and manifest accompanying scanning activity. In contrast, polarization/scanning activity is absent from CD2-triggered resting T cells, indicating that primed but not naive T cells are capable of such directed motile behavior. Consistent with these findings, prior analysis of the interaction between rat CD2 Jurkat transfectants and rodent CD48 on glass-supported lipid bilayers showed that cytoskeletal polarity was linked to CD2 ligation, as evidenced by microtubule organizing center (MTOC) reorganization (19). Cytoskeletal polarity depended on the CD2 cytoplasmic tail and involved CD2AP, one of several recently identified CD2 cytoplasmic tail adaptor proteins (19–21). The reorganization of the Golgi apparatus triggered by CD2 crosslinking herein independently confirms the role of CD2 in cytoskeletal polarization and links CD2 to specialized T cell motility.

Using various monolayers (CHO58 cells, murine L cell transfectants, and human dendritic cells), we were able to demonstrate that CD2 molecules specifically redistribute to the uropod together with the TCR- and CTx-binding lipid rafts. In contrast, CD45 and CD11a/CD18 molecules maintain a uniform distribution on the T cell surface. Independent analysis of human T cell interaction with human umbilical vein endothelial cells is also consistent with differential localization of adhesion molecules (10). Moreover, CD11a/CD18 interaction with ICAM-1 is required for T cell scanning on all monolayers examined, whereas CD2–CD58 participation is selectively relevant on CHO58 and, to a lesser extent, on DC. Analysis of CD58 and hICAM-1 fibroblasts confirms the notion that CD2 participation in scanning per se is not essential. Whether this result implies that hCD2–CD58 interaction is more critical for certain cellular conjugates because of a differential expression and/or engagement of adhesion receptors pairs remains to be determined. The difference in subcellular distribution and scanning requirement unequivocally distinguishes CD2 and CD11a/CD18 molecules and provides additional evidence for their nonredundant function (45).

A role for CD2 in positioning the T cell membrane at a distance suitable for TCR recognition of pMHC (≈130 Å) has been previously suggested and confirmed by x-ray crystallographic analysis (18). Consistent with this notion, interaction of CD2 with an elongated form of CD48 significantly inhibits immune recognition in the rodent system (46). In this regard, the 100-fold enrichment of CD2 in the uropod of APC-conjugated T cells with attendant concentration of TCR molecules in the same subcellular locale preconfigures the T cell surface in an idealized mode for subsequent pMHC recognition. The rapid on- and off-rates between CD2 and its counter-receptor (14) favor an appropriate opposing membrane distance geometry to support TCR–pMHC interaction without impeding TCR diffusion into the critical cell–cell contact surface area as the uropod scans the opposing APC in search of MHC-bound peptides. Furthermore, the elongated shape of the uropod affords a larger surface contact area to optimize the likelihood of productive immune recognition. Only a fraction of the lamellipodia preserves contacts with the APC, by contrast, although the importance of such initial cell–cell contact via the leading edge is not to be underestimated.

Recently, we demonstrated through subcellular fractionation that a significant number of human CD2 molecules is inducibly recruited into lipid rafts on CD2 ectodomain ligation (47). This translocation is independent of the TCR and requires no tyrosine phosphorylation. The uropod-specific relocalization of CD2 and lipid rafts observed microscopically appears to be the morphological counterpart of that inducible CD2-raft association analyzed biochemically. This redistribution is important because numerous signaling molecules localize in lipid rafts. These include Src family protein-tyrosine kinases, heterotrimeric and monomeric Ras-like G proteins, molecules involved in Ca²⁺ flux, and the small signaling molecule phosphatidylinositol bisphosphate (reviewed in ref. 48).

Analysis of T cell interaction with planar bilayers containing fluorescently labeled pMHC and ICAM-1 molecules has identified a specialized junction between T lymphocytes and “APC” referred to as the immunological synapse (49). This site has been shown to consist of a central cluster of TCRs surrounded by adhesion molecules, most notably integrins. Dynamic transport of TCR–pMHC complexes into the center of the synapse has been observed. Moreover, studies by Krummel et al. (50) demonstrated that the initiation of T cell activation correlates with formation of small yet unstable clusters of CD3 and CD4 driven by specific pMHC/TCR oligomerization, independent of the cytoskeleton. Yet other studies suggest an important role for CD28 within the immunologic synapse as well (51). The present analysis using an entirely cell-based system in the absence of specific cognate antigen defines a specialized subcellular compartmentalization in the uropod that precedes MHC-restricted peptide recognition. We suggest that this geometric configuration may be properly referred to as the “presynapse” because it is likely that such preorganization precedes pMHC-dependent TCR engagement and fosters subsequent productive interactions particularly important for T cell proliferation and cytokine production. To our knowledge, this represents the first molecular description using time-lapse methodology to correlate T cell surface immune recognition molecules and their subcellular locale during cell–cell interaction. Nevertheless, as similar T cell motility has been shown for both human and mouse lymphocytes during their interaction with dendritic cells in a three-dimensional collagen matrix (52), we suspect that surface T cell molecule preconfiguration during T cell movement noted herein is generally applicable to most, if not all, immune recognition events.

Previously, a human CD2 cytoplasmic tail-binding protein termed CD2BP1 was identified and shown to interact via its SH3 domain with aa 300–309 of the CD2 tail (20). CD2BP1 binding to CD2 was dramatically augmented as a result of CD2 ectodomain crosslinking by anti-CD2 mAb or conjugation to CD58-expressing cells. Hence, it follows that CD2BP1 association with CD2 should be maximal in the CD2-enriched T cell uropod during scanning. CD2BP1 also independently binds a second intracellular ligand, the tyrosine phosphatase PTP-PEST, at a site separate from CD2. Because PTP-PEST dephosphorylates p130^cas, it is probable that focal adhesion can be readily down-modulated by this means (30, 53). Transfection analysis in COS cells and Jurkat cells vis-à-vis CD2-dependent adhesion function are consistent with this view (20). Thus, an important role of focal CD2 redistribution will be to regulate adhesive interactions within the uropod, including but not restricted to CD2. In this way, the uropod is able to detach from the cellular substratum without fracture from the cell body because of the adhesive forces applied to the T lymphocyte during scanning. The emerging picture of T cell adhesion in general, and CD2-mediated function in particular, is that of a highly dynamic, regulated state of complex molecular and spatial interactions. The biology of T cell movement, membrane reorganization, and immune recognition are tightly interrelated and worthy of detailed exploration. Such analysis will lead to a molecular understanding of the processes involved in immune surveillance.

Abbreviations

APC

antigen-presenting cell

CTx

Cholera toxin

DIC

differential interference contrast

MHC

major histocompatibility complex

MLC

myosin light chain

pMHC

peptide complexed MHC

TCR

T cell receptor

CHO

Chinese hamster ovary