Actin foci facilitate activation of the phospholipase C-γ in primary T lymphocytes via the WASP pathway

1Upon encountering antigen-presenting cells (APCs) displaying T cell receptor (TCR) ligands and adhesion molecules, T cells undergo a series of actin dependent shape changes and signaling steps. These ligands include MHC complexed with antigentic peptide (MHCp) for mouse monoclonal T cells, agonist anti-CD3 antibody for polyclonal mouse (2C11) or human (OKT3) T cells. All of these ligands when presented with ICAM1 in supported lipid bilayers (SLB) trigger a series of changes in T cell synaptic cytoskeleton and morphology. First, TCR signaling is initiated at F-actin-based protrusions of polarized T cells. Next, these early activation events lead to F-actin dependent formation of a broad immunological synapse (IS) containing T cell antigen receptor (TCR) microclusters (MCs). Synapse spreading is followed by myosin II dependent contraction to form supramolecular activation clusters (SMACs). TCR MCs form in the outer lamellipodia-like ‘distal’ (d)SMAC, after which signaling MCs move centripetally through the lamella like ‘peripheral’ (p)SMAC, driven by centripetal flow of F-actin to eventually reach the ‘central’ (c)SMAC. The cSMAC is an F-actin depleted zone, in which TCR is destined for down-regulation via extracellular vesicle formation.

2Actin polymerization and remodeling continues throughout the lifetime of the immunological synapse, and studies using a variety of T cell activation systems have identified a critical requirement for F-actin and its molecular effectors in optimal TCR signaling and T cell function. Pharmacological disruption of global F-actin or genetic manipulations that reduce synaptic F-actin polymerization result in defects in early signaling events such as cell adhesion, TCR MC formation, early TCR signaling, TCR MC transport, as well as the TCR-distal events including intracellular calcium rise, and store operated calcium ion entry. Since global actin perturbation impairs all the above steps, it has been a challenge to dissect the mechanistic role of F-actin in TCR signaling and to identify whether there exist functionally distinct F-actin networks within SMACs that may play distinct roles at various stages of the pathway.

3One way to investigate functional diversity within the subsynaptic F-actin network, during TCR signaling, is to utilize actin perturbations that dissociate early TCR signaling from late TCR signaling. One such context is provided by the loss of an actin effector protein—Wiscott Aldrich Syndrome Protein (WASP). F-actin polymerization relies on regulatory factors such as nucleation promoting factors (NPFs) and downstream nucleation factors such as Arp2/3 complex. WASP is a hematopoietic-cell-specific NPF activated downstream of TCR, originally identified as a target of loss-of-function mutations in the X-linked immunodeficiency Wiskott-Aldrich syndrome (WAS). WAS T cells exhibit a variety of T cell activation defects such as aberrant synapse morphology, impaired proliferation in response to TCR-activation stimuli, and impaired calcium ion signaling. Similar to WAS patients, T cells from Was−/− mice also display profound defects in antigen receptor-induced proliferation, IS stability, nuclear NFAT translocation and IL-2 production. T cells from Was−/− mice and human WAS T cells have apparently normal total F-actin levels as well as SMAC organization within the immunological synapse, while initial TCR–associated kinase signaling in response to MHC-peptide complexes in the context of adhesion ligands is also intact. Despite many years of study, the F-actin network to which WASP contributes, and the specific TCR-signaling steps in which it participates to regulate calcium signaling, remain unknown.

4How might WASP regulate T cell calcium ion responses without affecting total synaptic F-actin? As an NPF, WASP binds to Arp2/3 and G-actin, increasing the ability of Arp2/3 to nucleate actin branches from existing filaments. Moreover, WASP binds hematopoietic lineage cell-specific protein 1 (HS1) through its SH3 domain. HS1 is also activated in response to TCR stimulation and can weakly activate Arp2/3 complex, as well as stabilize branched F-actin filaments. HS1 deficient T cells show defects similar to WASP−/− T cells in TCR activation dependent calcium elevation, proliferation, IL-2 secretion and NFAT activation. It is therefore possible that a previously uncharacterized subclass of the synaptic F-actin network at the TCR MC that represent a small fraction of total synaptic F-actin, is generated by WASP and stabilized by HS1, supports calcium signaling. Alternatively, it has also been proposed that WASP is a modular scaffolding protein capable of interacting with other proteins of the TCR signalosome, independent of its role as an NPF. Although these two hypotheses are not mutually exclusive, an F-actin dependent role could be addressed by identifying the F-actin network in the immunological synapse to which WASP contributes, and independently targeting this network to investigate the role of the WASP-generated F-actin subpopulation in calcium signaling at the synapse. Thus, WASP can be utilized as a tool to probe for functionally distinct organizational categories of F-actin within the synapse.

5The signaling cascade leading up to calcium ion elevation in response to TCR engagement has been studied in much detail. TCR ligation triggers a molecular program that results in activation of phospholipase C-γ1 (PLCγ1), through phosphorylation on Y-783 by Itk. Once it has been activated, phospho-PLCγ1 catalyzes the conversion of phosphatidylinositol-4,5 bisphosphate (PIP2) to inositol trisphosphate (IP3) and diacylglycerol. IP3 then acts as a second messenger and facilitates release of calcium ions from intracellular stores. Following TCR activation, PLCγ1 recruitment at the synapse is primarily mediated via binding to linker of activated T cells (LAT). Additionally, recent studies using Jurkat T cells and thymocytes have reported a role for the cortical cytoskeleton in both promoting and inhibiting PLCγ1 activation. Although PLCγ1 binds F-actin in biochemical assays, and loss of F-actin dynamics led to reduced PLCγ1 phosphorylation in Jurkat T cells, the dependence of PLCγ1 activation on WASP activity has not been tested in primary T cells. We hypothesize that WASP and HS1 generate an F-actin network that maintains phosphorylation of PLCγ1 at the synapse, accounting for their role in calcium ion elevation.

6In this study, we tested these hypotheses by characterizing the F-actin microarchitecture at the immunological synapse that is selectively regulated by WASP, and evaluating its role in early signaling, HS1 and PLCγ1 dynamics, and calcium signaling at the immunological synapse. The results presented here identify and functionally characterize a WASP-dependent actin network at the immunological synapse that regulates phospho-PLCγ1 levels at TCR MC and calcium ion elevation in T cells. This network is visualized as F-actin foci that result from new F-actin actin polymerization at TCR MC. Disruption of these actin foci does not impair initial synapse formation or early TCR signaling. Importantly, disruption of the F-actin foci using a selective Arp2/3 complex inhibitor also spares early TCR signaling, but results in the same impairment in HS-1 and PLC-γ1 activation that has been observed in WASP deficient T cells. Defining a new F-actin network in the immunological synapse, and its molecular regulation, furthers our mechanistic understanding of the cytoskeletal regulation of T cell activation and its dysfunction in WAS pathology.

7Utilizing a planar surface functionalized with anti-mouse CD3 antibody and ICAM1 as an antigen-presenting surface, we first compared the F-actin cytoskeleton in polyclonal primary CD4 T cells from either wild type (WT) or WASP deficient C57BL/6J mice at the early spreading stage of immunological synapse formation. To resolve dynamic F-actin networks, we employed a labeling strategy designed to highlight growing filament ends. A 1 min pulse of Rhodamine-ATP-G-actin in live-permeabilized cells identified freshly incorporated subunits (Fresh F-actin), while subsequent fixation and Alexa488-phalloidin staining identified total F-actin. T cells synapses were visualized using total internal reflection fluorescence microscopy (TIRF) to selectively illuminate structures within 200 nm of the interface. WT T cells exhibited discrete F-actin rich features (Figure 1A, arrowheads). We refer to these structures as F-actin ‘foci’. To selectively extract and quantify F-actin foci from the lamellar actin background, we utilized a spatial frequency-filter based local background correction method (Figure 1A graph, Figure1—figure supplement 1A,2, ‘Materials and methods’). This foci extraction method is capable of reliably identifying foci on a variable background of lamellar F-actin (Figure 1—figure supplement 1B). Assessment of the amount of newly polymerized actin in foci vs surrounding lamellar areas using the above-mentioned labeling and quantification method revealed that the F-actin foci incorporate fresh actin subunits (Figure 1B). This indicates that the foci are dynamically polymerizing structures, rather than sites where F-actin accumulates due to rearrangement of preexisting filaments. Furthermore, the polymerization rate of individual filaments (Fresh/Total ratio) is similar between foci and surrounding areas in WT T cells (Figure 1B), indicating that the foci result from preferential local nucleation, rather than enhanced polymerization rate of existing filaments. When compared at the whole synapse scale, overall synaptic F-actin levels, as well as the rate of actin polymerization were similar in WT and WASP−/− T cells, as expected. However, there was a significant reduction in F-actin foci in WASP−/− cell synapse (Figure 1C), suggesting that dynamic F-actin foci are formed in a WASP dependent manner. The loss of foci in WASP−/− T cells is not due to non-specific developmental defects, since acute siRNA mediated knockdown of WASP in activated mouse CD4 T cells resulted in a similar phenotype to the WASP−/− T cells. No reduction in total F-actin and significant reductions in foci were observed in WASP siRNA transfected cells imaged on SLB with anti-CD3 and ICAM1 (Figure 1—figure supplement 3). This lack of effect on total F-actin in WASP deficient or depleted cells may be attributed to the fact that these foci contribute to <10% of the total synaptic F-actin intensity (Figure 1—figure supplement 1B top graph), or perhaps to a compensatory increase in the lamellar and lamellipodial actin when more ATP-G-actin is available for polymerization regulated by other NPFs such as WAVE2. These results indicate that WASP dependent F-actin foci form in the T cell synapse.

8We next assessed the role of other WASP-family proteins—NWASP and WAVE2—in generating F-actin foci in WT cells. NWASP (WASL) is a close homolog of WASP and is capable of compensating for WASP's function in T cell development. However, primary CD4 T cells from Wasl−/− mice (Figure 1D) formed immunological synapses with F-actin foci similar to T cells from WT mice. WAVE2 is required for immunological synapse formation itself, thus we could not examine synaptic foci in WAVE2 deficient T cells. Instead, we determined the localization of NWASP and WAVE2 in the immunological synapse in order to gain further insight into their possible role in foci formation. Activation of T cells with anti-CD3 and ICAM1 recruited NWASP and WAVE2 to the IS (Figure 1—figure supplement 4A), but the recruited proteins failed to co-localize with F-actin foci (Figure 1—figure supplement 4B, NWASP coloc. = 9.48% ± 0.76 n = 78; WAVE2 coloc. = 14.36% ± 1.25, n = 54). We were unable to identify suitable reagents for staining of endogenous WASP in the immunological synapse. These results demonstrate that NWASP does not contribute to F-actin foci and suggest that WAVE2 is unlikely to contribute directly to formation of F-actin foci.

9In addition to WASP, deficiency of HS1, a cortactin-related NPF, also leads to reduced calcium ion signaling in T cells. Therefore, we examined the role of HS1 in generation of F-actin foci using T cells from WT and HS1−/− mice. F-actin foci formation was only marginally impaired in HS1−/− T cells, this reduction was modest compared to T cells from mice lacking both HS1 and WASP (Figure 1E). A similar marginal reduction in F-actin foci intensity was obtained by siRNA-mediated knockdown of HS1 in activated mouse CD4 T cells (Figure 1—figure supplement 5A). HS1 is phosphorylated on Y397 (phospho-HS1) during T cell activation. We next determined if this phosphorylation is WASP dependent. WT mouse CD4 T cells activated with anti-CD3 and ICAM1 displayed foci of phospho-HS1 in the synapse (Figure 1—figure supplement 5B). These phospho-HS1 foci were significantly reduced in WASP−/− or WASP−/− NWASP−/− CD4 T cells, but not in CD4 T cells from NWASP deficient mice (Figure 1—figure supplement 5B). Thus, HS1 is recruited to the T cell synapse in a WASP dependent manner.

10To understand the mechanism underlying formation of the F-actin foci, we determined if they were associated with TCR MCs. In murine polyclonal CD4 T cells activated on anti-CD3 and ICAM1 containing bilayers, as in Figure 1, TCR MC localized with F-actin foci (Figure 2A, upper panels, arrows). Similar co-localization of F-actin foci with TCR MC was observed with monoclonal AND TCR transgenic T cells stimulated with MHCp and ICAM1 (Figure 2A, lower panels). The tracking of TCR with H57 Fab allowed accurate quantification of co-localization between TCR and F-actin. After 1 min of cell attachment, 40 ± 1.7% of TCR MCs co-localized with F-actin foci in these cells (Figure 2—figure supplement 1A, ‘Materials and methods’, Figure 2B), significantly above chance level co-localization (Figure 2—figure supplement 1B), whereas only 15 ± 0.07% of total ICAM1 exhibited association with actin foci (Figure 2B), which was not different from chance. When examined in Human primary CD4 T cells as well, F-actin foci formed in WASP-dependent manner (Figure 2—figure supplement 2A,B), and TCR MCs, but not ICAM1 MCs, were associated with F-actin foci (TCR co-localization with F-actin foci was 34.5% ± 2.0, n = 76, co-localization of ICAM1 with F-actin foci was 13.1% ± 0.8, n = 66) (Figure 2C,D). Furthermore, TCR-activation led to HS1 phosphorylation at the T cell synapse, as shown previously (Figure 2—figure supplement 3A), and the phospho-HS1 co-localized with F-actin foci in both TCR-activation stimuli (Figure 2—figure supplement 3B). This indicated that TCR-MCs associated foci exist in a variety of primary CD4 T cells including mouse and human primary T cells, stimulated with diverse TCR-triggering contexts, namely anti-CD3 and peptide-MHC complexes.

11Since a fraction of TCR MC did not localize with foci, we sought to examine this population. High magnification microscopy of T cell synapses formed by mouse CD4 T cells on anti-CD3 and ICAM1 SLB provided insight into both synaptic and non-synaptic F-actin organization (Figure 2—figure supplement 4, Videos 1–2). As expected, the F-actin signal progressively decreases toward the synapse center where central TCR MC are devoid of F-actin foci (Figure 2—figure supplement 4, asterisks in the ‘Merge’ panel). To eliminate the possibility that the lack of foci on TCR MC in the nascent cSMAC is a consequence of cell fixation methodology, we examined foci in live T cells by transfecting them with LifeAct-GFP, a peptide construct that selectively labels F-actin. Since primary mouse T cells showed poor survival after transfection, we used human primary CD4 T cells that exhibit higher viability. In live cell synapses, LifeAct-GFP was observed to form foci on TCR MCs, which continued to co-migrate until the delivery of MCs to the cSMAC (Figure 2—figure supplement 5, Video 3). Analysis of kymographs revealed that the average TCR MC speed was 5.66 ± 2.2 μm/min, average Foci speed was 6.63 ± 3.1 μm/min (p = 0.21, ns). The TCR signals persisted in the cSMAC, whereas the GFP foci were extinguished, consistent with continual nucleation of F-actin at the MC site (Figure 1A) until it reaches the cSMAC, where TCR signals are known to be terminated.

12We also note that the Jurkat T cell line stimulated using a similar anti-CD3 and ICAM1 activation system displayed only random co-localization of TCR MC and F-actin (9.0% ± 0.09 TCR colocalized with F-actin, n = 47, Figure 2—figure supplement 4, right panels, Video 4). This suggests that a distinct F-actin organization may exist in this commonly used leukemia derived cells line.

13Since F-actin foci are associated with TCR MC sites in pSMAC and dSMAC zones, we sought to investigate the requirement of TCR signals for their formation. AND T cells activated using ICAM1 alone displayed some interface F-actin (Figure 3A, top images); however, a significant induction of F-actin foci only occurred when AND TCR transgenic T cells were incubated with both ICAM1 and agonist peptide-MHC (Figure 3A, top graph). A similar requirement of TCR triggering for F-actin foci formation was observed in primary human CD4 T cells (Figure 3A, bottom graph), although it did not scale with the dose of TCR agonist; in T cells incubated with low or high concentrations of anti-CD3 antibody, we did not observe a corresponding increase in F-actin foci (Figure 3—figure supplement 1). These results show that TCR activation triggers robust foci formation above a threshold level of TCR engagement, in the presence of a constant level of ICAM1 mediated adhesion. The residual foci intensity observed in T cells in the ICAM1 alone case (Figure 3A graph) is primarily due to fluctuations in lamellipodial F-actin detected as foci by our analysis method (Figure 1—figure supplement 1,2).

14To further investigate the role of TCR signaling in foci formation, we examined the role of Src family kinases (SFK) signaling. TCR signaling proceeds by accumulation of active Lck, as indicated by phosphorylation of the activation loop, in MCs. Consistent with this, phospho-SFK localized to F-actin foci associated MC (Figure 3—figure supplement 2). Furthermore, inhibition of SFK phosphorylation using PP2 significantly reduced the number of F-actin foci (Figure 3B), indicating that TCR-driven SFK signaling is required for their formation. We conclude that F-actin foci correspond to sites of early TCR signaling and require active SFKs.

15The next step in TCR proximal signaling is the SFK mediated recruitment of Zap70 to the TCR cytoplasmic domains and the phosphorylation of Zap70 by SFK on sites including Y319 in the interdomain linker. TCR-induced Zap70 Y319 phosphorylation in MC and overall F-actin levels at the synapse were comparable between primary WT and WASP−/− cells (Figure 3C). This result demonstrates that WASP is dispensable for TCR-proximal signaling including Zap-70 recruitment and activation.

16Having found that TCR MC formation and proximal signaling is independent of WASP, we investigated further TCR-distal signaling events that would lead to WASP-dependent calcium ion flux, as reported previously in human and mouse T cells. PLCγ1 is one of the key effector molecules regulating calcium ion flux downstream of TCR-activation, which mediates inositol-1,3,4-trisphosphate synthesis in the plasmamembrane, thereby facilitating calcium ion release from endoplasmic reticulum stores. In addition, there is evidence that PLCγ1 can bind to F-actin. We therefore reasoned that F-actin foci may locally enrich PLCγ1 at the TCR MC, supporting PLCγ1 interaction with TCR signalosome effectors such as Itk and LAT, thereby promoting its phosphorylation and activation. We therefore assessed the role of WASP in PLCγ1 activation in WT and WASP−/− T cells. WASP−/− T cells exhibited significantly reduced phospho-PLCγ1 levels at the synapse (Figure 4A). The reduction in phospho-PLCγ1 levels was not due to a general decrease in total PLCγ1 levels in the WASP−/− T cells, as the levels of total cellular PLCγ1 were comparable in WT and WASP−/− whole T cell lysates (Figure 4B). Thus, a loss of WASP-driven F-actin foci is correlated with impairment of PLCγ1 activation at the synapse.

17We next asked whether the well-established role of WASP in T cell calcium ion signaling pathway could be accounted for by F-actin foci, or by other pathways involving WASP-dependent protein–protein interactions. WASP is a scaffolding protein and interacts with a variety of proteins at MCs. To distinguish between two possible roles of WASP, activating Arp2/3 complex to generate F-actin foci, or acting as a direct molecular scaffold, in the regulation of PLCγ1 activation at the synapse, we overexpressed mutant forms of human WASP, including an Arp2/3-activation deficient form (WASP Δ473–480 or WASPΔC) and SFK scaffolding deficient form (WASPY291F) in human CD4 T cells. WASP binds with Arp2/3 via its C-terminal verprolin-homology, central, acidic (VCA) domain, while a stretch of hydrophobic residues in the central domain is reported to facilitate Arp2/3 activation without affecting the binding of Arp2/3 to the acidic domain. WASPΔC lacks this Arp2/3 activation sequence within the C-region (Figure 4C, schematic). Indeed, T cells expressing GFP-WASPΔC showed normal levels of Arp2/3 complex recruitment (data not shown), as well as total F-actin at the synapse, and synaptic recruitment of WASPΔC was comparable to WT WASP (Figure 4C,D). However, WASPΔC expressing cells were severely defective in F-actin foci generation, as well as in PLCγ1 phosphorylation at the synapse (Figure 4C,D). This result indicates that WASP-dependent Arp2/3 activation is essential for foci genesis and activation of calcium ion signaling effector PLCγ1 at the synapse. Overexpression of WASP Y291F, a mutant form defective in phosphorylation dependent interaction with SH2 domain of SFKs, had no effect on foci formation or PLCγ1 activation. This result suggests that WASP's scaffolding of SFKs via Y291 does not play a role in foci formation or PLCγ1 phosphorylation (Figure 4D). Together these results demonstrate a critical requirement of WASP's Arp2/3 activation and foci nucleation ability for PLCγ1 activation. Thus, decoupling of WASP's interactions in MC from the ability to activate the Arp2/3 complex, but not SFKs, results in a strong disruption of F-actin foci.

18Next, we more directly addressed the role of Arp2/3 complex, which is known to nucleate actin filaments de novo downstream of WASP, in foci induction. A previous study utilizing shRNA-mediated silencing of the Arp2/3 complex subunits in Jurkat T cells had demonstrated that actin polymerization at the synapse can occur independent of the Arp2/3 complex. However, the high-resolution subsynaptic distribution of Arp2/3 complex at the primary T cell synapse remains unknown. We first assessed the association of Arp2/3 complex with TCR MCs. A significant fraction of endogenous Arp2/3 complex is recruited at the TCR MC sites (Figure 5A, Figure 5—figure supplement 1A). In T cells treated with CK666, a small molecule inhibitor affecting conformational change in Arp2/3 complex and thus actin nucleation, the association of both Arpc2 and 3 with TCR MCs was reduced (Figure 5A, Figure 5—figure supplement 1A). The same treatment conditions also led to loss of F-actin foci (Figure 5B). As a control, CK689, an inactive isomer of CK666 (Figure 5—figure supplement 1B), and the inhibitor for another class of actin nucleation factors, formins, or inhibition of myosin II using blebbistatin, did not inhibit F-actin foci (Figure 5B, Figure 5—figure supplement 1C), indicating that these foci specifically require Arp2/3 complex for nucleation, and myosin-mediated actin rearrangement is not required for their formation.

19Since CK666 treatment can rapidly eliminate foci, we utilized CK666 to probe the kinetics of foci formation at individual TCR MCs. Given the small size and large lamellar actin background at the primary T cell synapse as well as highly mobile nature of MCs, this analysis would be extremely challenging to perform in the bilayer system. Therefore, we activated T cells on micron-scale anti-CD3 dots adsorbed on a glass surface where the remaining space was coated with ICAM1 in an attempt to restrict TCR signaling to a known, micron scale location in an artificial immunological synapse. Due to their aforementioned performance in time-lapse studies after transfection, human CD4 T cells were utilized for LifeAct-GFP transfection and live imaging on anti-CD3 microdots. Incubation with the patterned surface triggered rapid actin polymerization at the anti-CD3 microdot, reflected in LifeAct-GFP accumulation (Figure 5C–E). This enrichment of LifeAct at the microdot was not due to the topography of the cell contact interface, since a global membrane labeling dye, DiI, and integrin LFA1 were not enriched at these sites (Figure 5—figure supplement 2A,B). Within 5 s of CK666 treatment, pre-existing LifeAct enrichment was lost from the microdot site (Figure 5C–E Video 5) and fresh microdot contact events failed to elicit LifeAct-GFP enrichment in the presence of CK666 (Video 5). These data confirm that the TCR engagement sites dictate Arp2/3 complex-dependent F-actin foci formation at the synapse, de novo. Furthermore, ligand microdots can spatially pattern F-actin foci by predefining the sites of TCR engagement, and thus provide a suitable platform for spatially isolating TCR dependent mechanisms.

20We have shown above that WASP's foci forming ability is required for PLCγ1 activation (Figure 4B). In order to test whether the independent elimination of foci by perturbation of Arp2/3 complex would phenocopy this effect, we chose to deplete foci using CK666 and monitor the consequences of this treatment on early TCR signaling, PLCγ1 activation and subsequent calcium ion signaling. CK666 treatment, which resulted in severe inhibition of F-actin foci (Figure 5B, Figure 6—figure supplement 1B), did not diminish TCR-proximal activation events. While fewer T cells formed synapses after CK666 treatment (37% reduction), upon adhesion, F-actin dependent processes including formation of TCR MC and exclusion of CD45 phosphatase from MCs (Figure 6A), and the phosphorylation of Zap70 on Y319 (Figure 6B) proceeded normally in CK666-treated cells. In addition, other TCR-activation dependent signaling events leading to calcium flux, such as the phosphorylation of LAT on Y171 and SLP76 on Y145 were not altered in CK666-treated cells in comparison to control treated cells (data not shown). However, a severe defect was observed in the levels of TCR-induced PLCγ1 phosphorylation on Y783 after CK666 treatment (Figure 6C,D), where a drop was observed in both the number and intensity of phospho-PLCγ1 clusters at the synapse (Figure 6C). Furthermore, while phospho-PLCγ1 localized with F-actin foci in control cells, treatment with CK666 led to loss of foci-associated phospho-PLCγ1 (Figure 6E, Figure 6—figure supplement 1A). Upon attenuation of F-actin foci, total PLCγ1 levels were also reduced at the synapse, matching the decline in phospho-PLCγ1 levels (Figure 6F) and indicating that the impaired levels of phospho-PLCγ1 at the whole cell level may be accounted for by the reduction in PLCγ1 enrichment at the TCR signalosome. When assaying signaling events downstream of PLCγ1 activation, such as TCR-induced calcium ion elevation in the cytoplasm and nuclear mobilization of NFAT1, CK666-treated cells showed significant reduction in these processes (Figure 6G,H). Although a reduction in lamellipodial and lamellar actin was observed in CK666 treated cells (mean total F-actin 42% reduced), this reduction was relatively lower than foci depletion (mean 67% reduced) or under global F-actin depletion (CytoD, mean F-actin 90% reduced), and did not affect early TCR signaling (Figure 6—figure supplement 1B). We thus conclude that WASP and Arp2/3 dependent foci play a non-redundant role in PLCγ1 activation and calcium ion signaling at the synapse. In line with this result, CK666-treated cells exhibited a loss of phospho-HS1, as well as HS1, at the synapse (Figure 6—figure supplement 1C), similar to that observed in WASP−/− T cells (Figure 1E). Similar to the TCR-distal signaling defects observed in CK666 treated mouse CD4 T cells described above, CK666-treated human CD4 T cells also displayed a reduction in phospho-PLCγ1, but not in phospho-Zap70 (Figure 6—figure supplement 2), indicating that foci-dependent PLCγ1 activation cascade is conserved in T cells from diverse origins.

21To directly visualize the involvement of F-actin foci in PLCγ1 recruitment and stabilization at the TCR signalosome in live cells, we transiently overexpressed PLCγ1-YFP in human CD4 T cells, and examined its distribution at the synapse on SLB presenting anti-CD3 and ICAM1 in real time. PLCγ1 localized with TCR microclusters in live cell synapse, and this distribution was lost upon treatment of cells with CK666 (Figure 6—figure supplement 3). As a control, the association of human Zap70-GFP with MC was maintained after CK666 treatment (Figure 6—figure supplement 3). However, due the mobile nature of MC, it was challenging to follow individual cluster before and after CK666 treatment in this setting. We thus chose to utilize anti-CD3 microdots to activate human T cells. Human CD4 T cells expressing PLCγ1-YFP were incubated with micron-size immobile anti-CD3 dots, and visualized using live cell imaging immediately after cell attachment. Similar to F-actin foci assembly at the microdot, PLCγ1 enrichment was also visible at the microdot sites. CK666 treatment of cells led to a reduction of PLCγ1 enrichment at microdots (Figure 6I,J). These data show that TCR MC associated F-actin foci assist in sustained accumulation of PLCγ1 at the TCR MC.

22While SLB and solid-phase adsorbed antigen presentation systems allowed for superior optical resolution of the synapse, it is important to further test the validity of predictions from these reductionist models in more physiological T cell-APC interactions. Thus we utilized an activated cultured endothelial cells (EC) based planar APC system, where a flat synaptic interface is formed that is ideal for en-face visualization. Using this system, F-actin and HS1 rich ‘invadopod-like-protrusions’ (ILPs) are observed at TCR MC-like features in the periphery of the synapse. We found that ILPs were associated with the Y397 phosphorylated form of HS1, similar to F-actin foci (Figure 7A). Furthermore, when pre-formed T cell-EC conjugates were treated with CK666, there was concomitant loss of ILPs and phospho-HS1, indicating that these structures also require Arp2/3-dependent continuous polymerization of F-actin, and support local phospho-HS1 levels (Figure 7—figure supplement 1A–C). Since CK666-treatment also led to a reduction in total synaptic F-actin, which could be a consequence of inhibition of the EC cytoskeleton (Figure 7—figure supplement 1A–C), we utilized specific shRNA-mediated reduction in WASP levels in T cells, and examined foci (ILPs), total F-actin and phospho-HS1 levels at the T cell-EC conjugate interface (Figure 7A,B). WASP silenced T cells display selective loss of both ILPs as well as phospho-HS1, while maintaining total F-actin levels (Figure 7A–C). Therefore, the F-actin foci that we have primarily been characterized using SLB in this study, are the counterparts of ILPs in the cell conjugate system, and are necessary for efficient T cell activation. Thus, F-actin foci may represent a cytoskeletal module that functions to optimally support TCR distal signaling (Figure 7C) in diverse antigen presenting contexts.

23Compartmentalization of signaling proteins is a well-established strategy for spatial regulation of signaling in mammalian cells. At the immunological synapse, the compartmentalization of TCR signalosome proteins is regulated for optimal function. It is therefore surprising that cytoskeletal behavior is routinely measured as an all-or-none response of total polymerized F-actin, effectively ignoring different F-actin network subpopulations that could have distinct contributions to specific steps of the TCR signaling process. The WASP-dependent F-actin microarchitecture at the TCR MC that we report here is an important example. These F-actin ‘foci’ facilitate specific downstream TCR signaling steps, including PLCγ1 activation and calcium ion elevation, but are not required for TCR MC formation and other proximal signaling events such as Zap70 phosphorylation. Our study has shown that, when F-actin foci are depleted by WASP deficiency or CK666 treatment, the remaining lamellipodial and lamellar F-actin networks are not able to support optimal PLCγ1 activation, indicating that actin foci are indeed necessary for this later signaling step. Furthermore, we found that the molecular machinery associated with ‘flat’ TCR MC formed on planar substrates is the same as that in the distinctly three-dimensional ILP structures formed in T cell interfaces with antigen presenting endothelial cells. Others have noted the three dimensional nature of T-APC synapses that appear to have a classical immunological synapse/SMAC architecture. Our study thus defines a common F-actin rich module that bridges planar models and more complex, but more physiological, T-APC systems.

24F-actin ‘foci’ have been noted in prior studies, but they have not been studied in any detail. Ours is the first study, to our knowledge, to have performed a detailed characterization of foci as structures that are polymerized de novo via a WASP dependent mechanism. Previously, it was proposed that similar TCR associated F-actin foci were formed through a buildup of retrogradely flowing F-actin moving over artificially immobilized TCR on chrome micro-barriers in both Jurkat and primary T cells. Further characterization of this buildup revealed that it is due to accumulation of pre-existing filaments. We avoided this pitfall here by utilizing a system in which TCR MC remain mobile or are pre-established in micron scale spots that are segregated and are less likely to be traversed by F-actin flows, preventing an accumulation of filaments.

25T cell ILP display molecular similarities to ‘podosomal’ actin structures that are a common feature of leukocyte contact with substrates. Composed of dynamic F-actin, such podosomes are rich in HS1 related cortactin and are formed in a WASP-dependent manner at the adhesion interface of leukocytes. We speculate that the TCR utilizes similar WASP-HS1 based molecular machinery to generate F-actin foci/ILP that then serve to support calcium ion signaling through stabilization of PLCγ recruitment and phosphorylation. In the immunoreceptor driven synapses of non-T cells, this foci-like organization of actin microfilaments has not been observed thus far. F-actin architecture at the NK cell synapse appears to be uniform and the deep lamellipodial network dominates the synaptic interface. Similarly, in B cells, a thick cortical actin meshwork that is inhibitory for early B cell receptor signaling has been reported, but evidence of any ‘foci-like’ organization is missing. In such cell types, WASP could serve other roles, and may not promote immunoreceptor signaling via foci generation at the synapse.

26In the case of both CD4 and CD8 primary T cells, foci are robustly triggered following TCR ligation. In activated mouse T cell blasts, antigen-dependent induction of foci was observed as well; however, a few low intensity actin features were visible in a fraction of T cell blasts activated on ICAM1 in SLB (Figure 4A). We do not understand the origin of these low-intensity foci. It is possible that foci are generated at a basal level throughout the active signaling state of T cell blasts after initial priming, and antigen re-stimulation is a potent trigger for their formation. Alternatively, it is possible that integrins such as LFA-1 are capable of forming a few transient actin-rich features in activated lymphocytes in a specific chemokine environment, when its ligand ICAM1 is highly expressed. In contrast to the primary T cell systems, we have not detected foci in the Jurkat T cell line activated using the SLB system. A signaling network linking TCR to WASP has been outlined in the Jurkat cells, so it is likely this signaling pathway is intact. However, any WASP-dependent foci that may exist in these cells are either smaller than the resolution limit, or are transient in nature and rapidly dissipate into the lamellar network before they can be observed. Interestingly, in Jurkat T cells, the actin polymerizing function of WASP is not required for its role in calcium ion signaling, and Arp2/3 silencing does not prevent PLCγ1 activation. Thus, it is also possible that these cells utilize an alternative WASP-independent cytoskeletal pathway to support TCR-induced calcium ion elevation, and therefore do not need to utilize foci for this purpose. For example, the lack of negative regulators like PTEN may allow Jurkat cells to sustain calcium ion signals through alterations in lipid metabolism, without F-actin foci. A detailed comparison of the differential organization and function of F-actin modalities between Jurkat cells and the primary T cells will provide fundamental insights into the operation of compensatory cytoskeletal pathways that assist TCR distal signaling and may be important at different developmental stages in the T cell lineage or part of the failure of growth control in malignancy.

27Our study revealed at least two functional classes of F-actin at the synapse. The first component is critical for very early signaling events encompassing TCR triggering and the proximal signaling cascade. This fraction is represented by a slow turnover CK666-resistant cortical F-actin pool, which most likely relies on filament remodeling proteins other than actin branching for its dynamics. The second component, the WASP and Arp2/3-dependent F-actin foci network is responsible for assisting PLCγ1 activation and calcium ion signaling. Furthermore, in addition to these two networks, there is evidence for a third, CK666-sensitive network that we have not investigated in this study. This network is made up of a fraction of lamellar and lamellipodial F-actin that is rapidly eliminated under even short duration treatments of CK666 based on the reduction in ‘total F-actin’ (Figure 6—figure supplement 1B). This observation indicates that this third sub-population is composed of dynamic F-actin. This network is not required for the early TCR signaling steps that we have tested, indicating that it may play other roles at the synapse that are known to be dependent on F-actin. For example, it may be equivalent to the WAVE2 dependent network that is not required for early TCR signaling or PLCγ1-dependent calcium ion efflux from endoplasmic reticulum, but is necessary for CRAC channel mediated calcium ion entry from extracellular medium.

28Considering that all the above-mentioned networks are made up of the same building blocks of G-actin, how are functional boundaries between them set up and regulated, and how is this related to the organizational differences between these networks? Given the small size of T cells, the highly dynamic nature of their F-actin structures and the pleotropic action of pharmacological inhibitors on all F-actin organizations, it is challenging to dissect the specific roles of these sub-populations during TCR signaling. Thus, further interrogation of these networks will require a combination of genetic tools enabling conditional perturbation, as well as rescue, of Arp2/3 complex and NPFs, in combination with super-resolution microscopy in primary T cells. The results from these experiments will reveal how these networks combine to generate an emergent multifunctional synaptic F-actin network during T cell activation.

29The primary recruitment of PLCγ1 from cytosol to the MC has been well characterized, and it involves its interaction with TCR signalosome proteins LAT and SLP76, and subsequent phosphorylation at PLCγ1 residue Y783 by membrane-bound Itk. Loss of F-actin foci resulted in reduced phospho-Y783 PLCγ1 levels at the synapse (Figure 6C–D). The microdot enrichment assays and the SLB experiments both indicated that this was largely due to defects in PLCγ1 accumulation at the MCs (Figure 6). Accumulation of PLCγ1 at actin foci could occur via its direct association with actin filaments, as reported earlier, as well as with foci-recruited HS1. Both these mechanisms could simultaneously contribute to accumulation of PLCγ1 at TCR MCs.

30The precise manner by which accumulation at actin foci promotes PLCγ1 activation is not clear. There are several possible mechanisms that could underlie this process. First, it is probable that the continual nucleation of F-actin filaments at foci provides a sustained cytoskeletal microenvironment for the enrichment of PLCγ1 around the TCR signalosome. As the TCR MC traverses regions with highly variable levels of lamellar F-actin en route to the cSMAC, the foci would provide for the continuous presence of high levels of F-actin at the TCR microcluster, which may in turn ensure continuous local enrichment of PLCγ1 at the MC. Second, foci F-actin growing in juxtaposition to the plasma membrane may facilitate the interaction of F-actin bound PLCγ1 with its plasma-membrane anchored effectors such as Itk, thereby enhancing its phosphorylation. In this model, the growth of foci would exert force on and push the plasma membrane in its vicinity. Consistent with this mechanism, we find that foci zones mark areas of high membrane tension, observed as the foci-dependent localization and increased phosphorylation of mechanosensory protein CasL at the MC-foci sites (Santos et al., manuscript under review). This is also consistent with the manifestation as ILPs on the softer endothelial cell substrates. This observation also implies a possible role of F-actin foci in supporting synapse stability. Previously we have shown that WASP−/− T cell synapses form normally with normal pSMAC and dSMAC actin organizations, however they lost stability faster than those of WT cells. It will therefore be interesting to investigate whether the synapse stabilizing role of WASP is mediated by actin foci, and if foci influence pSMAC oscillations, as reported earlier in WASP−/− cells.

31Interestingly, in CK666-treated cells, there is a modest initial rise in intracellular calcium ion concentration, but final calcium ion concentrations are significantly lower than in DMSO-treated control cells and rapidly decline thereafter (Figure 6G). The factors contributing to residual calcium ion elevation may be either residual phopspho-PLCγ1 in CK666-treated cells, or an additional mechanism of PLCγ1 activation that is insensitive to this inhibition. In accordance with this, we do not observe a complete loss of PLCγ1 at microdot sites in CK666 treated cells. In contrast to the above mentioned processes that augment calcium ion signaling in mature T cells, there are alternative actin cytoskeletal pathways in immature thymocytes that impinge upon PLCγ1, and total F-actin is inhibitory for PLCγ1 activation and calcium ion flux. Detailed examination of F-actin networks in T cells during different developmental stages will clarify how the actin cytoskeleton can generate different activation outcomes at the level of PLCγ1.

32We find a specific requirement for WASP in the generation of F-actin foci. WASP−/− T cells show normal levels of total F-actin at the synapse, but no TCR MC associated F-actin foci. This lack of effect on total F-actin in WASP deficient or depleted cells may be attributed to the fact that the foci contribute to <10% of the total synaptic F-actin intensity (Figure 1—figure supplement 1B), and perhaps to a compensatory increase in the background actin networks when more ATP-G-actin is available for polymerization regulated by other NPFs such as WAVE2. We have shown that WASP's role in generating F-actin foci cannot be substituted for by NWASP. This selective requirement for WASP could be due to its specific interactions with upstream activators at the TCR signalosome. For example, a conventional mode of WASP and NWASP activation is through interaction with Cdc42 GTPase, However, WASP recruitment can occur independently of Cdc42 at the T cell synapse. In agreement with this, in preliminary experiments, Cdc42−/− T cells showed normal F-actin foci (data not shown) indicating that these foci form without the aid of Cdc42. Additionally, Cdc42-dependent phosphorylation of WASP at Y291 by SFK is known to increase its affinity for Arp2/3 complex. Overexpression of a phospho-deficient mutant (WASP Y291F) did not perturb F-actin foci formation or PLCγ1 phosphorylation, indicating that the WASP activation pathway utilized for foci formation may be distinct from the upstream regulatory pathways involving Cdc42 activation, or SFK binding at phospho-Y291. While it was not possible to dominantly uncouple TCR MC from PLCγ1 activation by over-expressing WASP that is able to activate Arp2/3, but does not interact with SFKs, it is possible that other regulatory phosphorylation sites can scaffold WASP's interaction partners, thereby playing an ancillary role in PLCγ1 activation. These results, and the failure of compensation by NWASP, suggest that WASP is activated and employed in a specialized manner in T cells to generate F-actin foci. The dimerization property of WASP and positive feedback with Arp2/3 complex, as well as its role in polarizing micron-scale ‘lipid rafts’ at the synapse may contribute towards a specific enrichment and involvement of WASP at foci. At this stage, the mechanistic basis for WASP vs NWASP utilization is not understood.

33To summarize, we have characterized a subsynaptic F-actin organization in T cells that is functionally distinct from larger scale lamellipodial and lamellar networks. These actin foci provide a signaling scaffold that is required for TCR-distal signaling and play a critical regulatory role in calcium ion signaling. There are significant questions that remain to be answered. For example, we do not yet know the MC-intrinsic properties that temporally regulate F-actin foci formation. We find that ∼40% of TCR MCs are associated with F-actin foci at steady-state. While a large fraction of remaining clusters are in the central cSMAC region of the cell, previously characterized to be an actin depleted zone containing signaling incompetent TCR, a smaller fraction of foci-unassociated MC resides in the lamellar zone as well. At lamellar zone TCR MCs, it is unclear which MC-intrinsic attributes dictate foci induction, maintenance and subsequent termination of actin nucleation on entry into the cSMAC, and if integrin signaling assists in any of these steps. Further investigation will not only provide a deeper understanding of subsynaptic F-actin dynamics, but will also illuminate other specialized roles that F-actin subpopulations play during the lifetime of a signalosome. The results will also provide deeper insights into signaling compartmentalization of the T cell synapse that is regulated by the actin cytoskeleton and its alteration in WAS pathology.

34AND Mouse T cells used in this study were isolated from 2–3 months old AND B10.Br mouse spleen and lymph nodes. For experiments involving activation with bilayer containing MHC-peptide, AND CD4 T cell blasts were used. Briefly, total splenocytes and lymph node cells were expanded in vitro, in the presence of 2 μM MCC peptide and 10 units/ml IL-2 for 6 days in Kanagawa's modified Dulbecco's modified Eagle Medium (KDMEM) supplemented with FBS, and were subsequently used for experiments. In other experiments freshly isolated CD4 T cells were used, as indicated in figure legends. For experiments with mouse CD4 T cells, spleens and lymph nodes were harvested, and CD4 T cells were purified using mouse CD4 isolation kit (Miltenyi Biotec, Auburn, CA). These CD4 T cells were used immediately after isolation for the experiments. For WASP single knockout experiments, Was−/− C57BL/6J mice (stock number 019458) were purchased from Jackson Laboratory (Bar Harbor, Maine), and age-matched WT C57BL/6J mice were also purchased as controls (Stock number 000664). For WASP and HS1 single knockout, and Was−/− Hcls1−/− double knockout studies, C57BL/6J, Hcls1−/−, Was−/− and Hcls1−/− Was−/− mice were used, and CD4 T cells purified as described above. For experiments with WASP−/−, NWASP−/− and WASP−/− N-WASP−/− CD4 T cells, cells were purified from spleen and lymph nodes of WT 129 or knockout 129 mice (obtained from Snapper laboratory), as described above. For experiments with human CD4 T cells, cells were purified from fresh peripheral blood samples (New York Blood Center, NY) using Rosette Sep kit (Stemcell Technologies, Vancouver, BC, Canada) according to the manufacturer's protocol. Human CD4 T cells were rested overnight in RPMI supplemented with FBS, prior to their use in experiments.

35N-WASP (H-100) and WASP (H-250) antibodies were purchased from Santa Cruz Biotechnology (Dallas, TX). HS1 antibody (D5A9), Phospho-Y397 HS-1 antibody (D12C1), phospho-Y319 Zap70/Y352 Syk (affinity purified antisera #2704), PLCγ1 (D9H10), phospho-Y783 PLCγ1 (#2821), NFAT1 (D43B1), phospho-Y416 SFK (#6943), phospho-Y171 LAT (#3581) and WAVE2 (D2C8) antibodies were obtained from Cell Signaling Technology (Beverly, MA). Anti-Arp3 monoclonal antibody (FMS338) and anti-MyH9 (Myosin IIA, #SAB2101542) antibody were purchased from Sigma Chemicals (St. Louis, MO). Rabbit p34-Arc (anti-Arpc2) antibody was purchased from EMD Millipore and phospho-Y145 SLP76 (#NBP140511) was purchased from Novus Biologicals. Celltracker-CM-DiI was purchased from Life Technologies. Anti-rabbit secondary antibodies conjugated with Dylight647 and Dylight564 dyes or anti-mouse Fc-specific Alexa568 secondary antibodies were procured from Jackson Immunoresearch (West grove, PA) and tested for lack of cross-reactivity prior to their use in assays. For anti-TCR antibodies-mediated T cell activation on bilayers, 145-2C11 (#16-0031, for mouse cells) and OKT3 (#16-0037 for human cells) clones were purchased (eBioscience, San Diego, CA), mono-biotinylated and labeled with fluorophores, and tested for mono-biotinylation prior to use in experiments. For experiments involving assessment of PLCγ1 activation using western blots, T cells were activated with anti-CD3/CD28 coated Dynabeads (Life Technology, Grand Island, NY, #11452D) according to manufacturer's instructions. A plasmid encoding Lifeact-GFP was a kind gift of B Geiger (Weizmann Institute, Rehovot, Israel), and GFP-tagged WASP constructs were a kind gift of Dianne Cox (Albert Einstein School of Medicine, New York). Blebbistatin, Wiskostatin and PP2 were obtained from Sigma chemicals; CK666 and CK689 were obtained from Calbiochem (EMD Millipore, Billerica, MA). Alexa488-phalloidin and Alexa568-phalloidin were purchased from Invitrogen (Life Technologies). Other fluorescent probes were obtained from Invitrogen and were conjugated with desired proteins according to the manufacturer's protocols. I-Ek-6His and mouse ICAM-1-12His were produced in the S2 insect cell line. I-Ek-6His was either loaded with moth cytochrome C peptide 91–102 or in some cases was the same MCC peptide was covalently coupled to the β chain. The density I-Ek-6His and ICAM1-12His was 20–50 molecules/μm2 and 200 molecules/μm2, as described previously, unless otherwise noted.

36For siRNA experiments, siRNA were obtained from Dharmacon (Thermo Fisher Life Sciences, Pittsburgh PA). Mouse WASP siRNA (on-target plus, Cat no-LU-046528-01-0005), Mouse HCSL-1 (HS1, on-target plus, Cat no-LU046134-01-0002) were purchased as a set of four. A mixture of these four siRNAs was used at a concentration of ∼600 nM for 5 million cells, for WASP, HS1 and non-specific (control) siRNA (D-001810-10-20) cases. Cells were electroporated using Amaxa Nucleofector (Lonza, Basel, Switzerland) following the instructions from the manufacturer, and grown in K-DMEM for 72–80 hr before experiments. For lentiviral transduction experiments, GIPZ lentiviral expression system (Thermo Scientific, Pittsburgh PA) was used. Briefly, the HEK cells were transfected with GIPZ-WASP shRNA plasmids (RHS4287-EG7454, a mixture if three sequences-5′ TTCTTATCAGCTGGGCTAG 3′ (V2LHS_92677), 5′ TTGCTGATCTTCTTCTTCC 3′ (V2LHS_92678), and 5′ TCATCTTCATCGCCAGCCT 3′ (V3LHS_352271)), or non-specific shRNA plasmid (RHS4346) along with the trans-lentiviral packaging mix (RHS4287-EG7454), using the calcium phosphate method. The culture supernatant containing viral particles was isolated after 68 hr of transfection, and utilized to transduce CD4 T cells according to manufacturer's instructions. The transfected T cells were identified by GFP expression in both control and WASP shRNA transduction cases.

37WASP depletion was assessed using western blots. Approximately 2 million cells from the control and siRNA samples were harvested, lysed using RIPA buffer supplemented with protease inhibitor, and the resulting lysates were resolved in reducing conditions on a polyacrylamide gradient gel (Biorad, Hercules, CA). After transfer to the nitrocellulose membrane, protein levels were assessed by staining with primary (overnight at 4°C) and IR-dye conjugated secondary antibodies, and subsequently imaged using an Odyssey infrared imager (LI-COR Biosciences Lincoln, NB).

38CD4 T cells were incubated with bilayer reconstituted with MHCp (I-Ek-MCC) and ICAM1, or 2C11 and ICAM1 (AND mouse CD4 T cells and polyclonal C57BL/6J T cells), or with bilayer reconstituted with OKT3 and ICAM1 (human polyclonal CD4 T cells) for the indicated time duration in Flow chambers (Bioptech Inc., Butler, PA) or Ibidi microscopy chambers (Ibidi, Verona, WI). Bilayer deposition and ligand density determination was performed as described previously. For immunostaining, cells were fixed using PHEMO buffer (10 mM EGTA, 2 mM MgCI2, 60 mM Pipes, 25 mM HEPES, pH 6.9) supplemented with 3.7% paraformaldehyde at room temperature for 20 min and permeabilized using 0.1% Triton X-100 for 2 min. Cells were then blocked using 5% casein for 30 min, washed and incubated with primary antibody overnight at 4°C and appropriate secondary antibodies for 1 hr, followed by washing and visualization using the indicated microscopy method.

39For all of the colocalization experiments, mouse AND T cell blasts activated on MHCp bilayers were utilized. For WASP knockout T cell experiments, freshly isolated T cells were utilized. For all live imaging experiments, peripheral human CD4 T cells were utilized, unless otherwise noted.

40AND CD4 T cell blasts were loaded for 30 min with 2 µM of Fluo4 (Life Technologies) in HEPES buffered saline (HBS) which was supplemented with 1% HAS and pluronic acid. Cells were then treated with DMSO or 10 µM CK666 for 1 min, and then incubated in glass chambers (Nunc Lab-Tek, Thermo Scientific, Rochester, NY) containing 2C11 and ICAM1. Cells were imaged for the indicated time, using a confocal microscope with fully open pinhole. The Fluo4 fluorescence obtained for each cell was normalized to baseline fluorescence, and mean values from individual cells were then plotted on the graph ±standard error of the mean (SEM).

41Patterns of OKT3 antibody were created by micro-contact printing onto glass coverslips. First, topological masters were fabricated by electron beam using a 1 µm PMMA layer, and spun-coated onto silicon wafers. Then, a combination of a thin layer of hard PDMS and a thick layer of Sylgard184 was spun-coated onto these wafers, resulting in hydrophobic stamps for patterning. Stamps were incubated with 1:1 ratio of Alexa647-OKT3 and unlabeled isotype antibodies, maintaining a total concentration of 25 µg/ml. Following coating, stamps were rinsed with PBS, PBS + 0.05% Tween-20, and deionized (DI) water, dried with N2 gas, and then placed in contact with plasma-cleaned coverslips for 2 min. Coverslips were washed and incubated with 1 µg/ml ICAM1 for 40 min, after which they were washed again and then used for the experiment.

42Barbed end labeling of F-actin was carried out as previously described. Briefly, T cells were incubated with glass coverslip coated with ICAM1 and anti-CD3 for 5 min, and then incubated with 0.0125% saponin dissolved in Rinse buffer (20 mM HEPES, 138 mM KCl, 4 mM MgCl2, 3 mM EGTA, pH 7.5) along with freshly dialyzed Rhodamine labeled G-actin (0.5 µM in chilled G-actin buffer-4 mM TRIS, 0.2 mM CaCl2, 0.2 mM ATP, 1.0 mM DTT) for 1 min at 37°C. They were then rapidly fixed using 0.5% glutaraldehyde dissolved in cytoskeletal buffer (10 mM MES, 3 mM MgCl2, 138 mM KCl, 2 mM EGTA) for 10 min. Following this, cells were washed and permeabilized using 0.5% X-100 dissolved in cytoskeletal buffer for 5 min, and then incubated with sodium borohydride (100 mM made fresh in PBS) for 10 min. Subsequently, cells were processed for total F-actin staining, using Alexa488-phalloidin, mounted using Fluoromount, and visualized using TIRF microscopy.

43As described previously, human dermal microvascular endothelial cells (HDMVECs) were cultured on fibronectin (10 mg/ml) in complete EBM-2 MV media (Lonza, MA). ECs were transfected with palmitoylated-YFP DNA construct (‘membrane-YFP’, from Life Technologies, CA) using Amaxa electroporation according to manufacturer's instructions. Prior to the experiments, ECs were activated for 48 hr with 100 ng/ml IFNγ and 12 hr with 50 ng/ml TNFα. ECs were eventually pulsed with 1 mg/ml SEB and TSST for 60 min (sAg, Toxin technologies, FL), incubated with human peripheral T cells (sAg expanded T cells, cultured in presence of IL-15) for the relevant time duration, and imaged using either an upright laser scanning confocal microscope (Zeiss) or spinning disc confocal microscopy (Andor technology CT).

44For TIRF microscopy, imaging was carried out using a Nikon Eclipse Ti microscope with a 100× 1.49 NA objective, and an AndorDU897 back illuminated EMCCD camera. For spinning disc confocal microscopy, an Olympus IX70 microscope with a 150× 1.45 NA objective, CSU (Solamere Technology, Salt Lake City, UT) and a Hamamatsu Orca ER camera was used. Solid-state lasers (Coherent, CA) provided illumination at 488, 561 and 641 nm, and narrow pass filters (Chroma Technology, Burlington, VT) were used for detection.

45All image processing was carried out using ImageJ and Metamorph software. For colocalization analysis, raw images were subjected to local background subtraction as follows. Raw images were subjected to a local 50% intensity rank (median) filter (1.6 × 1.6 microns box size, see Figure 1—figure supplement 1,) across the whole image in order to generate a ‘background image’. This background image was then subtracted from the raw image. The resultant high-pass filtered image was thresholded, and used to assess co-localization using similar methodology to assess pixel overlap across overlaid thresholded images as described before, via ‘co-localization class’ plugins in ImageJ. The intensity of overlapping pixels for each cell were normalized to total intensity of the same cell in the thresholded images, and were plotted as ‘fraction co-localized’ per cell (Figure 2—figure supplement 1).

46To assess F-actin foci alone, raw images were subjected to 50% rank filtering (1.6 × 1.6 microns box size, subsample ratio 1) and resultant image was subtracted from original image (Figure 1—figure supplement 1). All the samples were subjected to identical filter settings prior to analysis. Cell intensities from the subtracted image were then calculated by marking regions around the cells, and using ‘regional measurements’ features of MetaMorph Software. In most graphs ‘intensity’ or ‘levels’ on Y-axis refers to integrated intensities in the cell synapse, unless otherwise indicated.

47The images are displayed at identical contrast settings, and all the images presented here are raw images, unless otherwise mentioned. All the experiments were repeated at least thrice unless otherwise mentioned. For the graphs showing F-actin foci intensity, the datasets were normalized to the average value from ‘control’ data for that experiment, and plotted as ‘normalized intensity’ unless otherwise mentioned. For statistical analysis, the Mann–Whitney unpaired t-test was performed, unless otherwise mentioned. The bars and error bars in all graphs represent mean ± SEM.