Antigen Recognition Is Facilitated by Invadosome-like Protrusions Formed by Memory/Effector T Cells

Endothelial-presented Ag promotes activation of CD4 memory T cells

We wanted to elucidate the mechanisms by which memory/effector T cells probe for Ag using ECs as model APCs. Therefore, we isolated natural CD4 memory-like lymphocytes (nT_mem) and generated induced/expanded CD4 memory-like lymphocytes (iT_mem) from human peripheral blood. The isolated nT_mem were CD4⁺, CD45RO⁺, and contained both central memory and effector memory T cells (∼78 and ∼22%, respectively) based on CD62L staining (Supplemental Fig. 1A). The iT_mem were slightly more polarized toward the effector memory T cell subtype (∼69 versus ∼31%; Supplemental Fig. 1B). Monolayers of HLMVECs, HDMVECs, and HUVECs were used as models for the vasculature. In vivo endothelial MHC-II expression is dependent on IFN-γ (26). This was recapitulated in vitro by culturing ECs with exogenous IFN-γ (Supplemental Fig. 1C). Endothelium was additionally activated with TNF-α to promote an inflamed phenotype.

Next, we set up live-cell analysis to concomitantly monitor T cell activation (i.e., calcium flux) and migration on endothelium pulsed with Ag (i.e., bacterial superantigen, a widely used model Ag) (38, 40, 41). In the absence of Ag, nT_mem and iT_mem fluxed little calcium (Fig. 1A), remained polarized (Fig. 1B), and underwent continuous lateral (Fig. 1C) and transendothelial migration (i.e., “diapedesis”; Fig. 1D). On HLMVECs pulsed with Ag (1 μg/ml SEB and TSST), T cells rapidly fluxed calcium, lost polarity, and arrested migration (Fig. 1A–C). Similar results were found with alternate Ag (e.g., SEE and MAM) presented by HDMVECs or HUVECs (Supplemental Fig. 1D and data not shown). In addition, in the absence of Ag, the majority of iT_mem transmigrated within 30 min, whereas in the presence of Ag, diapedesis was delayed by ∼30–60 min (Fig. 1D). Thus, the migratory stop signal was transient. IT_mem also exhibited Ag-dependent transient migration arrest under conditions of laminar fluid shear flow conditions similar to those found in microvasculature in vivo (2 dyne/cm²; Fig. 1E–G, Supplemental Video 1).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Human microvascular ECs are capable of Ag-specific stimulation of CD4 memory-like T cells. (A) Human CD4 nT_mem (circles) or iT_mem (triangles) were loaded with Fura-2 and incubated on Ag-pulsed HLMVECs. Calcium flux at 5 min (left) or at maximal value over a 30-min duration (right) were plotted. (B and C) iT_mem were incubated on HLMVEC pulsed with Ag and T cell polarity (length/width ratio) at 5 min, and velocity was calculated over a 30-min duration. (D) iT_mem transendothelial migration was scored on Ag pulsed ECs as described in Materials and Methods. (E) iT_mem were infused over Ag-pulsed HLMVEC at 2 dyne/cm² and imaged. Depicted is an end-point image with migration tracks of apical (blue) and transmigrated (red) lymphocytes. See also Supplemental Video 1. (F and G) Lateral migration velocities for T cells from (E) during both prediapedesis and postdiapedesis phases of migration over 30 min (F) and percentage of iTmem transmigrated were measured (G). (H) Representative images of nT_mem incubated on Ag-pulsed HLMVEC and staining for actin (green), NFAT (red), and nucleus (blue). (I) nT_mem were incubated on activated HLMVEC. Nuclear NFAT translocation was scored according to line scan profiles (left). (J) Samples were acquired as in (H), and representative transmigrating or transmigrated (‘Under’) lymphocytes are shown. Data represent mean ± SD (B, C) or SEM (D, F, G, I). Scale bars, 5 μm. *p < 0.05, **p < 0.005, ***p < 0.0005.

During T cell activation, the transcription factor NFAT translocates from the cytoplasm to the nucleus. Incubation of nT_mem on Ag-pulsed ECs promoted robust nuclear translocation of NFAT (Fig. 1H, 1I). Significantly, transmigrating nT_mem and iT_mem retained nuclear NFAT, demonstrating that these cells remained activated (Fig. 1J). NT_mem on endothelium that was pulsed with Ag but not pretreated with IFN-γ (and therefore lacked strong MHC-II expression) exhibited little NFAT translocation (Fig. 1I), confirming that the responses to Ag-pulsed endothelium were MHC-II dependent.

To show that T cells achieved complete activation, we conducted studies using a transwell system, in which T cells loaded into an upper chamber migrated across an EC monolayer to reach the lower chamber. Flow cytometry revealed that after migrating across Ag-pulsed ECs, iT_mem showed increased surface expression of CD69 and a reduction in CD62L (Supplemental Fig. 1E). In addition, in these nonpolarizing conditions, a subset of the iT_mem upregulated IFN-γ expression (Supplemental Fig. 1F, 1G) and increased proliferation (Supplemental Fig. 1H). Collectively, these results demonstrate that endothelium is able to present model Ag in an MHC-II–dependent manner to CD4 memory/effectorlymphocytes, which induces a transient delay in diapedesis and T cell activation.

ILP arrays dominate the T cell–EC interface during activation

Next, we investigated the cellular and molecular basis for T cell activation in our endothelial APC model. Previously, we used fluorescent membrane markers (mem-YFP or mem-DsRed) expressed in the ECs to detect topological changes in the plasma membrane (34). In this way, we demonstrated that ∼0.5-μm fluorescent rings that formed dynamically on the endothelium under migrating T cells corresponded to cylinder-shaped cell surface invaginations (i.e., “podo-prints”) induced by lymphocyte ILPs. Thus, podo-prints formed on the EC surface served as an indirect but sensitive readout for T cell ILPs (34). In this study, we similarly observed in control settings that podo-prints formed and disappeared continuously (with lifetimes of tens of seconds) under the leading edge lamellipodia of migrating iT_mem (Fig. 2A, 2C, Supplemental Video 2). Strikingly, when ECs were pulsed with Ag, iT_mem rapidly formed dense arrays of podo-prints largely localized at the periphery of the T cell–EC interface under symmetrical lamellipodia (Fig. 2B, example 1, and Supplemental Video 3). A minority (∼20–30%) of the iT_mem cells formed rosette-type podo-print arrays that lacked bias for the cell periphery (Fig. 2B, example 2, and Supplemental Video 4). In both cases, podo-prints/ILPs were significantly stabilized (lifetimes of ∼18 min; Fig. 2C) and exhibited limited lateral translocation (0.31 ± 1.5 μm; Fig. 2D–F). After ∼30 min of contact, ILPs began to disappear and T cells initiated transmigration (Supplemental Video 4 and data not shown).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

CD4 memory-like T cells form stabilized arrays of ILPs on Ag-presenting endothelium. (A and B) HLMVECs were transfected with mem-YFP or mem-DsRed, activated, and pulsed without (A) or with (B) Ag (TSST/SEB). iT_mem were imaged live on addition to ECs. Individual frames at selected time intervals are shown. Arrows indicate fluorescent rings formed on endothelium under adherent lymphocytes. The corresponding control (A) and Ag (b1 and b2) videos are provided as Supplemental Videos 2–4, respectively. (C) Podo-print/ILP lifetimes from imaging studies as in (A) and (B). Data represent mean ± SEM of at least 80 ILPs formed by at least 25 cells from ≥3 separate experiments. (D) Schematic representation of podo-print/ILP lateral translocation analysis. Gray “T cell outline” represents a T cell–EC contact area with the “T cell centroid” indicated as a blue circle and individual podo-prints indicated as black rings. The linear distance between the centroid and podo-print was measured at both the first and last appearance of an individual podo-print/ILP during its lifetime. (E and F) Podo-print/ILP distances were measured as in (D). Distances from the cell centroid at time of formation and disappearance were plotted for individual podo-prints (E). Data from (E) was further processed to report change in distance (F). Scale bars, 5 μm. ***p < 0.0005.

As confirmation that the earlier observations reflected three-dimensional podo-prints in response to T cell ILPs, we cotransfected endothelium with mem-DsRed together with soluble GFP as a marker for cytoplasmic volume (34). This showed individual fluorescent membrane rings of each podo-print, in fact, represented cytosol-displacing invaginations into the EC surface (Fig. 3A). Furthermore, confocal imaging showed that podo-prints were ICAM-1–enriched cylindrical EC invaginations into which T cell ILPs extended (Fig. 3B, Supplemental Video 5). ILPs were enriched in LFA-1 (Fig. 3B), F-actin (Fig. 3C), and talin (Fig. 3D), similarly to invadosomes (35, 36). Analogous structures formed under physiologic shear flow (Supplemental Fig. 2A), with diverse ECs (including HLMVECs [Fig. 3A, 3C, 3D], HUVECs [Fig. 3B], HDMVECs [data not shown]), with nT_mem (Supplemental Fig. 2B), and with alternate Ag (Fig. 3B, Supplemental Fig. 2D). Similar structures also formed when previously activated murine OT-II CD4⁺ T cells were incubated on heart microvascular ECs pulsed with OVA 323–339 peptide Ag (Supplemental Fig. 2E).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

Ag-stabilized ILPs exhibit a discrete three-dimensional architecture. (A) Ag-stabilized T cell ILPs protrude into the EC surface. Imaging was conducted as in Fig. 2B on HLMVECs coexpressing soluble, cytoplasmic GFP (green) and mem-DsRed (red). (B) Three-dimensional reconstruction from confocal imaging of podo-prints and ILPs. iT_mem were incubated for 20 min on activated Ag-pulsed (SEE) HUVECs and then fixed, stained for ICAM-1 (green) and LFA-1(red), and imaged by confocal microscopy. Sections were digitally reconstructed and projected as three-dimensional renderings. (b) Magnified view of the ILP arrays. (b′) Orthogonal cross section. See also Supplemental Video 5. (C and D) Ag-stabilized ILPs are enriched in actin and talin. Mem-YFP–transfected (C; green) or mem-DsRed–transfected (D; red) HLMVECs were pulsed with Ag (TSST/SEB) and incubated with iT_mem for 5 min and stained for F-actin (C; red) or talin (D; green). (E) Quantitation of ultrastructural depth and width of T cell ILPs. Samples as in (C) were imaged by electron microscopy and ILPs were measured. Data represent mean ± SEM from at least 100 ILPs from at least 20 representative micrographs per condition. (F and G) ILPs enforce close T cell–EC membrane apposition in the absence and presence of Ag. T cell and endothelial nuclei are indicated with red and green overlays, respectively. Arrows and (a) highlight regions of extremely close lymphocyte–EC membrane apposition enforced at the ILP tips. Scale bars, 5 μm (A–D); 500 nm (F, G). *p < 0.05. ns, Not significant.

To assess the T cell–EC interaction in greater detail, we used transmission electron microscopy. In the absence of Ag, ILPs averaged 430 ± 34 nm in depth and 348 ± 33 nm in width (Fig. 3E, 3F). In the presence of Ag, ILPs were similar in morphology and size (depth = 437 nm, width = 277 nm), but tended to form in denser clusters (Fig. 3E–G). An important feature of the ILPs, whether in the absence or presence of Ag, was the existence of zones of extremely close T cell–EC apposition, typically at the tips of the ILPs as if driven by ILP extension (Fig. 3F, 3G, arrows). Although not exclusive to these locations, quantitative analysis revealed that intercellular contacts of <20 nm were 9-fold greater at ILP tips compared with other regions. The idea that ILPs can exert significant force is supported by their ability to both drive transcellular diapedesis (34) and to displace/deform organelles such as the nucleus (Fig. 3F).

Ag-stabilized ILPs share features of traditional TCR signaling microclusters

To address how these ILP-dominated cell–cell interfaces relate to TCR signaling, we stained for traditional IS markers (15, 20, 21). Confocal microscopy revealed that podo-prints on endothelium were modestly, but consistently, enriched in MHC-II compared with MHC-I or mem-YFP (Fig. 4A, 4B, Supplemental Fig. 3A), whereas T cell ILPs were enriched in CD3 (Fig. 4C, Supplemental Fig. 3B). Comparable enrichment formed under physiologic shear flow (Fig. 4I, 4J). In <5% of T cells, CD3 was distributed into a central supramolecular activation cluster-like cluster rather than in peripheral ILPs (Supplemental Fig. 3C).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

Ag-stabilized ILPs are foci for immune signaling. IT_mem cells were incubated with activated, Ag-pulsed (TSST/SEB) HLMVEC for 5 min, fixed, and stained as indicated and imaged by confocal microscopy. (A) Ag-stabilized ILPs (F-actin, green) protrude into MHC-II (red)–enriched podo-prints. Arrows indicate MHC-II–enriched podo-prints. (B) Samples as in (A) were stained for MHC-II (HLA-DR/DP/DQ; red) and MHC-I (HLA A/B/C; green). Schematic of gated regions of interest is shown on the left and included a region outside of the IS (I), the ILP-rich region of the IS (II), and the central region of the IS (III). Right panels show pixel fluorescence intensity histograms for regions I–III. (C) ILPs (F-actin, blue) protruding into podo-prints (mem-YFP, green) are enriched in CD3 (red). Within individual ILPs, CD3 is predominantly focused at the tip (b1) and, to a lesser extent, the edge (b2). (D) PKC-θ (red) is enriched with CD3 (green) in ILPs. (E–G) ILPs (F-actin, blue) colocalize with ZAP70 (E; red), phosphotyrosine (F; red), and HS1 (G; red). (H) HS1 (red) is highly enriched in ILPs. Cross-sectional views from serial-section confocal microscopy are shown of a lymphocyte adherent to the endothelium-presenting Ag. See three-dimensional rotation in Supplemental Video 6. (I and J) ILP arrays were allowed to form as above with the additional presence of physiologic laminar fluid shear flow (2.0 dyne/cm²; arrow indicates direction). (I) MHC-II (red) and F-actin (green) are shown. (J) CD3 (red) and ICAM-1 (green) are shown. Scale bars, 5 μm.

Molecules implicated in TCR signaling, including PKC-θ (Fig. 4D, Supplemental Fig. 3D), ZAP70 (Fig. 4E), and phosphotyrosine (Fig. 4F) were also enriched in ILP cores, suggesting active signaling at these sites. The cortactin homolog HS1, a known regulator of both podosomes and ISs (38, 42–44), showed particularly strong enrichment (Fig. 4G, 4H, Supplemental Video 6). Alternatively, the inhibitory molecules CD43 and CD45 localized primarily to the region outside of the T cell–EC interface (Supplemental Fig. 3B). These observations suggest that ILPs may serve as discrete loci for TCR signaling with general features similar to TCR signaling microclusters and “multifocal” synapses (15, 21, 22, 45–47).

Collectively, these experiments show that CD4 iT_mem activation caused by EC MHC-II/Ag is coupled to cell–cell interfaces dominated by arrays of ILPs that are enriched in TCR signaling molecules. We term this previously undescribed architecture, which shares features of both podosome belts/rosettes and multifocal ISs, a “podo-synapse.”

ILPs support efficient Ag recognition and sustained signaling

To determine the functional relationship between ILP formation and T cell activation, we concomitantly monitored calcium flux and ILP dynamics. In the presence of Ag, calcium flux was evident shortly after the first appearance of ILPs. It then peaked several minutes later and gradually decayed over the following 5–60 min (Fig. 5A, 5B, Supplemental Video 7). Calcium flux always occurred subsequent to (or in the same 10-s interval as) the appearance of at least one ILP, with an average offset time of ∼25 s (Fig. 5C). In turn, stabilized ILP arrays and symmetric lamellipodia became evident in the ∼30–60 s following the increase in calcium (Fig. 5A, 5B, Supplemental Video 7).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5.

T cell ILP formation precedes and supports efficient Ag recognition. (A) IT_mem were labeled with Fura-2 and imaged live (at a maximal temporal resolution of 10 s) during migration on mem-DsRed–transfected, Ag-pulsed HLMVECs. Upper panels show mem-DsRed. Arrows indicate initial ILP formation. Middle panels indicate calcium flux values on a rainbow scale. Lower panels provide a schematic representation of newly formed (green) and previously formed (in relation to previous field; red) podo-prints. (a) Frame of initial ILP/podo-print. (b) Frame when calcium flux rises above background. (c) Frame when the peripheral ILP array is stabilized. Note this correlates with peak calcium flux. See also Supplemental Video 7. (B) Graphical representation of calcium flux with frames a–c noted. (C) Quantitation of ILP-calcium flux offset time. Live-cell imaging was as in (A) and offset time (time from when first ILP forms until calcium flux rises above background) was calculated. Data were binned into 10-s intervals, and average ± SEM is shown for 35 individual T cells from 3 separate experiments. (D) Imaging was performed as in (A) with additional pretreatment of T cells with latrunculin A before addition to Ag-pulsed EC monolayers. (E) Podo-print/ILP index (average number of podo-prints/ILPs per cell) and average calcium flux at 5 min was calculated. Both analyses are pooled mean ± SEM from three separate experiments. (F) IT_mem were labeled with Fluo-4 and imaged live with anti-CD3/CD28 cross-linking with or without latrunculin A pretreatment. Left panels show resting T cells. Right panels show activated T cells imaged 60 s after addition of cross-linking Abs. Arrows indicate de novo formation of micron-scale T cell protrusions. (G) Average calcium flux was calculated from three separate experiments as in (F). Scale bars, 5 μm. Data represent mean ± SEM. *p < 0.05, ***p < 0.0005.

These observations suggested that the close intercellular contacts driven by ILPs may promote initial Ag recognition and TCR triggering. Testing this hypothesis is challenging because both ILPs and TCR signaling are fundamentally dependent on F-actin assembly and many of the same actin regulatory pathways (34, 35, 38, 42–44, 48). Thus, we simply compared effects of F-actin inhibition (via latrunculin A) on T cell activation through an Ag-pulsed APC versus direct TCR cross-linking. When stimulated by Ag-pulsed ECs, iT_mem pretreatment with latrunculin A caused 100% blockade of both ILP formation and calcium flux during the first 5 min of coincubation (Fig. 5D, 5E). By contrast, when TCRs were stimulated directly by anti-CD3/CD28 cross-linking, significant (although attenuated by ∼60%), immediate calcium flux was elicited in the presence of latrunculin A (Fig. 5F, 5G), as shown previously (49). From this we conclude that the total latrunculin A-induced blockade of early response to Ag-pulsed ECs reflects a defect in initial Ag recognition/TCR triggering (Fig. 5F, 5G). We speculate that ILP formation may represent the latrunculin A-sensitive process behind this defect.

Calcium flux is necessary and sufficient for Ag-mediated ILP stabilization

Next, we investigated the transition from dynamic ILP probing to formation of stable ILP arrays. Given the correlation between increase in calcium and appearance of ILP arrays, we hypothesized that calcium may be key for stabilizing ILPs. To test this, we pretreated T cells with the calcium chelator BAPTA and the CRAC channel inhibitor BTP2. On Ag-pulsed ECs, this greatly inhibited calcium flux in iT_mem, which correlated with strong reduction in the number of stabilized ILPs (Fig. 6A–C). To determine whether calcium flux was sufficient to stabilize ILPs, iT_mem were incubated on ECs in the absence of Ag, followed by application of thapsigargin to directly increase intracellular calcium. This caused iT_mem already engaged in dynamic ILP probing to arrest migration and stabilize ILP clusters (Fig. 6D–G). Thus, whereas dynamic ILP probing of ECs by iT_mem seems to facilitate initial Ag recognition, the resulting increase in calcium is both necessary and sufficient to drive accumulation/stabilization of ILPs into podo-synapses.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 6.

Calcium flux is necessary and sufficient for ILP stabilization. (A) IT_mem were Fura-2 labeled, pretreated with the calcium chelator BAPTA and the CRAC channel inhibitor BTP2, and imaged live on Ag-pulsed ECs. Arrows indicate a few sporadically formed podo-prints. (B and C) Podo-print/ILP index (B) and average calcium flux (C) were calculated at 5 min for experiments as in (A). (D) Live-cell imaging was conducted in the absence of Ag before and after addition of the calcium ionophore thapsigargin (at time = ‘0:00’). Arrows indicate ILPs/podo-prints. (E) Correlation of calcium flux with ILP number after addition of thapsigargin. (F) Quantitation of the number of newly formed ILPs in the 2 min before and after addition of thapsigargin. (G) Quantitation of the lifetime of ILPs with or without addition of thapsigargin. (H and I) Imaging and analysis was performed as in (A)–(C) except T cells were pretreated with the CAMKII inhibitor CK59. Data represent mean ± SEM. Scale bars, 5 μm. **p < 0.005, ***p < 0.0005.

The earlier findings point to a positive feedback loop for sustaining TCR signaling. To break this loop, we sought to block proximal signaling downstream of calcium flux, specifically targeting CAMKII via the inhibitor CK59 (50). Treatment with CK59 attenuated Ag-mediated ILP stabilization by ∼50% (Fig. 6H), which was coupled to a proportional decrease in calcium flux (Fig. 6I). This result supports the interdependence between ILPs and calcium flux.

CD8 CTLs use ILPs to probe for Ag

The endothelium has been shown to initiate MHC-I/Ag-dependent activation of CD8 CTLs leading to direct killing of ECs (31, 32). To investigate these responses, we incubated previously activated murine OT-I CD8 T cells on murine heart microvascular ECs pulsed with SIINFEKL peptide Ag. Ag-pulsed ECs initiated a rapid calcium flux (Fig. 7A) that was coupled to progressive specific lysis of ECs over a 4-h duration (∼50% lysis ∼2 h; Fig. 7B). Live-cell imaging showed that in the absence of Ag, CTLs avidly probed the endothelium with dynamic ILPs while migrating without fluxing calcium (Fig. 7C, Supplemental Video 8). In the presence of Ag, initial ILPs were rapidly followed by calcium flux (offset time = 32.8 ± 6.6 s), which then led to migrational arrest and formation of peripheral ILP arrays similar to those formed by CD4 effector T cells (Fig. 7D, 7E). Thus, T cell ILPs seem to be a general feature of Ag sensing on ECs.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 7.

Murine CD8 CTLs use ILPs to sense Ag-bound MHC-I on endothelium. (A) Previously activated murine OTI CD8 T cells (CTLs) were imaged on mem-YFP–transfected, mouse heart microvascular ECs pulsed with SIINFEKL. Individual ratiometric calcium flux values at 5 min were plotted for ±Ag conditions. (B) CTLs were added to mixed cultures of Ag-pulsed/unpulsed ECs, and percentage of specific lysis is plotted at indicated time points. (C) CTLs were imaged during EC probing in the absence of Ag. Differential interference contrast and mem-YFP are shown in the upper and lower panels, respectively. Dashed line represents the outline of the migrating CTL under which transient ILPs are continuously formed. See also corresponding Supplemental Video 8. (D) Imaging was conducted as in (C) on ECs pulsed with SIINFEKL. Red arrow indicates initial ILP formation preceding calcium flux. (E) Average offset time for calcium flux relative to initial ILP formation. Data represent mean ± SEM. Scale bars, 5 μm.

A planar-coated cell model for Ag recognition

We next considered whether ILPs either represent unique features of T cell-endothelial Ag recognition or may be more broadly relevant and uniquely revealed by endothelium. We speculated that the rigidity of coated substrate models may frustrate/mask ILP formation, whereas orientation/resolution issues might obscure detection of ILPs with traditional cellular APCs (16–18).

First, we simply asked whether evidence consistent with ILPs could be detected in a coated glass model (Fig. 8A). IT_mem plated on glass coated with ICAM-1 and anti-CD3 (but not anti-CD43/CD45) Ab rapidly fluxed calcium (Supplemental Fig. 4A–C). IRM revealed that initial “microcontacts” (∼0.2- to 1-μm dots consistent with T cell ILPs and/or microvilli) always preceded calcium flux (Supplemental Fig. 4B, 4C; offset time = 21.4 s). New microcontacts also continued to form after T cell spreading (Fig. 8B, Supplemental Fig. 4D, 4E, Supplemental Video 9).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 8.

ILPs facilitate Ag recognition on professional APCs. (A) Schematic of the “coated-glass model” using ICAM-1– and Ab-coated glass. (b) Magnified view of the interaction surface. (B) iT_mem were added to an ICAM-1–Fc– and anti-CD3–coated glass chamber and imaged by differential interference contrast and IRM. Arrows indicate IRM-detected “microcontacts.” (C) Schematic of the “planar coated-cell APC model.” CHO-K1 cells expressing ICAM-1–GFP and soluble DsRed were surface “coated” with Abs against CD3, CD43, or CD45 using a biotin/SA capture approach (see also Supplemental Fig. 4F). (d) Magnified view of the interaction surface. (D) Cells prepared as in (C) coated with anti-CD3 Ab were imaged live on addition of Fura-2–labeled iT_mem. Panels demonstrate relatively early and later phases of interaction as indicated. Arrows indicate podo-prints/ILPs. See also Supplemental Video 10. (E and F) Imaging experiments were conducted as in (D), but T cells were pretreated with latrunculin A. Podo-print/ILP index (E) and calcium flux (F) were calculated. (G) IT_mem were incubated with Ag-pulsed Priess B cells for 5 min and stained for LFA-1 and actin. Image represents a three-dimensional projection of T cell settled on top of the B cells with the IS forming more nearly parallel to the x–y imaging plane. See also Supplemental Video 11. (H) Transmission electron micrograph of T cell–B cell IS as in (G). The T cell is identified by a red overlay. Arrows indicate close intercellular contacts formed by T cell ILPs. (I) Schematic (upper left panel) shows both lateral (a) and various en face (b) interfaces typically formed in vitro between T cells and the highly irregular surfaces of DCs. (1) and (2) show previously activated murine CD4 OT-II lymphocytes interacting with Ag-pulsed mem-YFP–transfected DCs (green, 2). Red outlines depict T cell perimeter based on DIC imaging (1). (3) is identical to (2) but shows three boxed regions (i–iii) expanded on the left. (3i) shows a lateral interaction in which the T cell has pushed small finger-like invagination into the side of the BMDC putatively formed by ILPs (arrows). (3ii) and (3iii) show en face interactions whereby ILPs seem to be extending normal to the imaging plane, giving rise to the typical ring-shaped podo-print appearance observed in endothelium (i.e., Fig. 2). (J) DCs were transfected with soluble monomeric DsRed and mem-YFP, pulsed with Ag, and CD4 OT-II T cells were added. Panels show cytoplasm displacing podo-prints/ILP (arrows), which precede calcium flux and are stabilized after calcium flux. (K) Transmission electron micrograph of T cell–DC IS as in (J) fixed after 5 min. The T cell is identified by a red overlay. Arrows indicate close intercellular contacts formed by T cell ILPs. (L) Calcium flux of previously activated CD4 OTII T cells 8 min after interaction with C57BL/6 (WT) or Ciita^−/− Ag-pulsed DCs with or without latrunculin A pretreatment. Data represent mean ± SEM. Scale bars, 5 μm. ***p < 0.0005.

We hypothesized that these microcontacts at least partly reflected ILP activity that was mechanically frustrated by the rigid substrate. To test this, we designed a “coated-cell APC” model whereby the earlier Ab-coated glass substrate was recapitulated on the surface of a deformable CHO-K1 cell (Fig. 8C, Supplemental Fig. 4F). In this setting, clear three-dimensional ILP arrays were readily detected that were coupled to calcium flux (Fig. 8D, Supplemental Video 10). As with iT_mem on ECs, initial ILPs always preceded (Supplemental Fig. 4G; offset time ∼56 s) and seemed to functionally support (Fig. 8E, 8F) initiation of calcium flux. These findings suggest a general tendency of memory/effector T cells to use ILPs to probe diverse substrates for recognition and response to Ag (as modeled in Fig. 9).

ILPs are involved in recognition of Ag presented by professional APCs

Next, we re-examined Ag recognition with professional APCs. Thus, iT_mem were incubated with Priess B cells pulsed with Ag and imaged by confocal microscopy. Equatorial cross sections recapitulate classic views (12) of the T cell–B cell IS whereby T cell LFA-1 shows enrichment in peripheral-supramolecular activation cluster–like peripheral membrane bulges (Supplemental Fig. 4H, inset 2, arrows). Cross sections taken at the edge of the IS additionally suggest presence of LFA-1–rich finger-like protrusions (Supplemental Fig. 4H, insets 1a–d) similar to ILPs formed on ECs (Fig. 3Bb). More compelling imaging of putative ILPs can be seen in occasional examples where the T cells settle on top of B cells such that the IS aligns with the optimal imaging plane (18) (Fig. 8G, Supplemental Video 11). Ultrastructural views, evident in a subset of micrographs, further support the presence of ILPs (Fig. 8H).

Finally, we investigated murine OTII CD4 T cells incubated with BMDCs coexpressing mem-YFP and soluble DsRed. T cells interacted with DCs both laterally (Fig. 8I, subset a) and to a lesser extent through en face contacts (Fig. 8I, subset b). In the former, side views of DC invaginations were readily evident that were similar to podo-prints/ILPs seen on endothelium (Fig. 8I, panel 3i). En face interactions also revealed discrete circular podo-prints/ILPs (Fig. 8I, panels 3ii, 3iii), although these were generally less obvious than those formed on endothelium because of the highly active and irregular surface topology of DCs. Dynamic imaging showed ILP probing preceded calcium flux, which, in turn, was coupled to peripheral ILP array formation (Fig. 8J). Ultrastructural confirmation of ILP formation could also be obtained in a subset of electron micrographs (Fig. 8K). Finally, treatment of T cells with latrunculin A strongly inhibited both ILPs and initial calcium responses to Ag presented by DCs (Fig. 8L).