LFA-1 and kindlin-3 enable the collaborative transport of SLP-76 microclusters by myosin and dynein motors

T cells are activated via the interaction of T cell receptors (TCR) with antigenic peptides presented by major histocompatibility complex (MHC) proteins. This occurs in the context of a transient (kinapse) or stable (synapse) adhesive junction that links the T cell to an antigen-presenting cell (APC). Persistent immune synapses (ISs) are subdivided into concentric subdomains, or supramolecular activation clusters (SMACs) (Monks et al., 1998). The outermost region, the distal SMAC (dSMAC), is analogous to a circumferential lamellipodium (Freiberg et al., 2002; Varma et al., 2006; Sims et al., 2007; Dustin, 2008). Internal to the dSMAC, a lamellar domain known as the peripheral SMAC (pSMAC) is enriched in the active form of the integrin LFA-1 (αLβ2) (Monks et al., 1997, 1998; Dustin, 2008). The innermost region, the central SMAC (cSMAC), can be subdivided into regions engaged in co-stimulatory signaling, signal termination and secretion (Varma et al., 2006; Stinchcombe and Griffiths, 2007; Yokosuka et al., 2008; Saito et al., 2010; Vardhana et al., 2010). The kinapses formed by migratory T cells contain similar domains, oriented asymmetrically, with the leading edge corresponding to the dSMAC.

In response to ligand recognition, TCRs are assembled into microclusters and phosphorylated by Src kinases, leading to the recruitment of the tyrosine kinase ZAP-70 (Bunnell et al., 2002; Choudhuri et al., 2005; Varma et al., 2006; Dushek et al., 2012). TCR microclusters form in the dSMAC, at contact sites generated via the dynamic polymerization of actin (Valitutti et al., 1995; Varma et al., 2006). When the TCR ligands are laterally mobile, these microclusters traverse the immune synapse and accumulate in a central region. Movement through the dSMAC is driven by Arp2/3-dependent actin flows, while movement through the pSMAC is governed by the contraction of formin-dependent actomyosin arcs (Ilani et al., 2009; Babich et al., 2012; Beemiller et al., 2012; Yi et al., 2012; Murugesan et al., 2016). Although a sparse network of actin filaments is present in the cSMAC, the last step in the centralization of TCR microclusters requires dynein motors, which concentrate TCR microclusters near the microtubule-organizing center (MTOC) (Hashimoto-Tane et al., 2011; Murugesan et al., 2016; Fritzsche et al., 2017). Signaling microclusters are terminated by endocytosis or exocytosis from the central region (Lee et al., 2003; Vardhana et al., 2010; Choudhuri et al., 2014). Nevertheless, the myosin-dependent forces that drive the peripheral movements of TCR microclusters favor the phosphorylation of Src kinases, ZAP-70, LAT and CasL (NEDD9) and have been linked to the quality of the stimulatory ligand, implying that contractile forces contribute to T cell activation (Yu et al., 2012; Murugesan et al., 2016; Hong et al., 2017).

The triggering of the TCR also leads to the assembly of microclusters containing the adaptor SLP-76 (LCP2), which is crucial for T cell development and activation. These structures form at distal sites adjacent to TCR microclusters and are normally co-transported with TCR microclusters. However, when the triggering ligands are immobile, SLP-76 microclusters separate from TCR microclusters and are independently transported towards the center of the synapse (Bunnell et al., 2002; Yokosuka et al., 2005; Nguyen et al., 2008; Yi et al., 2019). The persistence and movement of SLP-76 microclusters require cooperative interactions among their constituents and with surrounding actin filaments (Barda-Saad et al., 2005; Braiman et al., 2006; Bunnell et al., 2006; Houtman et al., 2006; Balagopalan et al., 2007; Barda-Saad et al., 2010; Pauker et al., 2011; Sylvain et al., 2011; Pauker et al., 2012; Coussens et al., 2013; Ophir et al., 2013; Lewis et al., 2018). Within these small ‘signalosomes’, SLP-76 enters signaling complexes nucleated by another critical adaptor, LAT, and partitions, in an actin-dependent manner, into a ring that surrounds the LAT-containing core (Sherman et al., 2011; Barr et al., 2016; Sherman et al., 2016). In primary cells, similar SLP-76 microclusters are surrounded and stabilized by ‘micro-adhesion rings’ containing actomyosin filaments, LFA-1 (αLβ2 integrin), and several integrin-associated proteins (Hashimoto-Tane et al., 2016). The elimination of the SLP-76- and actin-binding adaptor ADAP (FYB1) destabilizes SLP-76 microclusters, suggesting that microcluster cohesion requires the coupling of SLP-76 to these enclosures (Lewis et al., 2018).

The interactions among the cytoskeletal systems in the immune synapse and the mechanisms by which these systems influence SLP-76 microcluster centralization and TCR signaling remain incompletely understood. SLP-76-associated proteins, such as ADAP, Nck (NCK1 and NCK2) and WASP (WAS), interact with various cytoskeletal systems and may provide the motive force for the transport of SLP-76 microclusters (Barda-Saad et al., 2005; Nguyen et al., 2008; Pauker et al., 2011; Lewis et al., 2018; Ditlev et al., 2019). In contrast, in the presence of immobile ligands, the integrin VLA-4 (α4β1) slows actin flows and immobilizes SLP-76 microclusters in the dSMAC (Nguyen et al., 2008; Jankowska et al., 2018). Immobile LFA-1 ligands also slow actin flows, and may inhibit the movement of SLP-76 microclusters in this manner (Murugesan et al., 2016). Conversely, integrins might promote the centripetal movement of SLP-76 microclusters by tethering these structures to cytoskeletal systems via the adaptors ADAP and SKAP55 (SKAP1), which regulate integrin activation and integrin signaling via SLP-76 (Hunter et al., 2000; Griffiths et al., 2001; Peterson et al., 2001; Kliche et al., 2006; Baker et al., 2009; Ophir et al., 2013; Lewis et al., 2018). In support of this model, SKAP55 promotes the translocation of the integrin-activating protein talin-1 from the margin of the immune synapse to the center of the contact, where it accumulates in proximity to SLP-76 (Ophir et al., 2013).

We now show that contractile actomyosin filaments, microtubule-dependent dynein motors and SLP-76 microclusters reciprocally regulate one another, and that the centralization of SLP-76 microclusters and contractile filaments are co-regulated by myosin II and dynein motors. These processes are coordinated by SKAP55, kindlin-3, β2 integrins and several integrin-associated cytoskeletal proteins, but do not involve β1 integrins or the integrin-activating protein talin-1. The factors that regulate centripetal transport processes govern the TCR-induced upregulation of CD69, which is unaffected by the loss of talin-1. These findings emphasize the importance of cytoskeletal function in T cell activation, clarify the distinct and specialized roles played by integrins and their associated proteins, and demonstrate that LFA-1 supports critical signaling functions, even in the absence of an extracellular ligand.

SLP-76 microclusters display position-dependent transitions in cluster velocity similar to those occurring with TCR microclusters (Nguyen et al., 2008). To address the involvement of contractile actomyosin systems in the transport of SLP-76 microclusters, we visualized SLP-76, myosin IIA (heavy chain MYH9) and the activated form of the regulatory myosin light chain (pMLC) in J14.SY cells stimulated on anti-CD3ε coated glass substrates. The pMLC antibody targets an S19-phosphorylated epitope thet is perfectly conserved between MRLC1/MLC-9 (MYL9) and MRLC2/MLC-12B (MYL12B). The J14.SY cell line, previously derived from SLP-76-deficient J14 Jurkat T cells, stably expresses a SLP-76.YFP chimera. In a serum-independent manner, TCR-induced SLP-76 microclusters move from the periphery into a compact domain at the center of the contact (Fig. 1A; Fig. S1A). Immunofluorescence staining reveals that a subset of endogenous myosin IIA accumulated in a dense structure at the center of the contact, overlapping with SLP-76, while thinner filaments of myosin extended from this structure to the periphery of the contact and terminated near to SLP-76 microclusters (Fig. 1A, white arrows emphasize radial filaments). These structures were even more prominent in the pMLC channel, indicating that they are contracting (Fig. 1A). Because this distribution differs from previous reports, we also evaluated the patterning of myosin IIA in wild-type Jurkat E6.1 cells and in Jurkat E6.1 cells expressing SLP-76.YFP (E6.SY cells). Whereas Jurkat E6.1 cells generate circumferential arcs of myosin IIA, exogenous SLP-76 at about three times endogenous levels reproduced the centripetal accumulation of myosin IIA and pMLC observed in the J14.SY cells, without affecting TCR expression (Fig. 1B; Fig. S1B,C). In all three lines, and in primary human T cells stimulated on analogous substrates, most myosin IIA accumulated in a ring 7–9 µm from the center of the contact (Fig. S1D). Increased SLP-76 expression shifted the average radial positions of myosin IIA and pMLC towards the center of the contact and created a small, but distinct, central pool that was enriched in pMLC (Fig. S1D).

We derived from J14.SY cells a cytoskeletal reporter line (J14.SY-CRL) that expresses fluorescent chimeras of the filamentous actin-binding protein F-Tractin and the human regulatory myosin light chain encoded by MYL9 (MLC-9). These chimeras did not alter the centralization of SLP-76 microclusters and the distribution of MLC-9 mirrored that of pMLC (Fig. 1C). In this line, the SLP-76 microclusters formed in response to weak stimuli do not centralize and MLC-9 is not recruited into distinctive structures (Fig. 1D). Intermediate ligand densities drive the appearance of filamentous MLC-9 structures in the cell periphery but do not elicit microcluster centralization. Further increases in ligand density shift MLC-9 filaments from the cell periphery first to a pSMAC-like ring, and then to a compact central cluster linked via radial spokes to the cell periphery (Fig. 1E). Concurrently, SLP-76 microclusters display partial, and then full centralization, suggesting that the subcellular distribution of contractile myosin is linked to that of SLP-76, and that both processes are influenced by TCR signal strength.

Background-corrected versions of the images in Fig. 1B revealed that myosin IIA and pMLC form a network of closed cells, or rings, of ∼1–2 µm in diameter (Fig. 1F,G). These units are most prominent in the periphery of the contact, while more loosely interconnected myosin IIA structures are present in the center of the contact. SLP-76 microclusters are frequently found within these closed cells (yellow brackets) or alongside myosin filaments (white arrows). In primary human CD4⁺ and CD8⁺ T cell blasts stimulated on analogous substrates, myosin IIA forms circumferential arcs similar to those observed here and in earlier TIRF-SIM studies (Fig. S2A) (Murugesan et al., 2016; Hong et al., 2017). Most of these cells also develop a ‘purse-string’-like structure (red arrowheads) at the inner boundary of the myosin IIA arcs. This contractile structure often separates from distal myosin structures (gaps denoted by red brackets). Links between the myosin IIA arcs give rise to cells resembling those observed in Jurkat T cells (Fig. S2B). By algorithmically identifying the centers of voids within the myosin IIA meshwork (Fig. S2C, top row), we were able to determine that the peaks of myosin IIA intensity, as function of radial distance, did not differ in Jurkat E6.1 cells, E6.SY cells, J14.SY cells, primary human CD4⁺ T cells, primary human CD8⁺ T cells, or E6.1 cells expressing a tagged form of myosin IIA (Fig. S2D). The diameters of the ring-like structures defined in this manner averaged 1.74±0.06 µm (mean±s.d.), which is comparable to the size of the previously reported micro-adhesion rings (Hashimoto-Tane et al., 2016). SLP-76 and pMLC are also reduced in the centers of these voids and are enriched at a similar distance to that of myosin IIA. However, SLP-76 does not colocalize extensively with either myosin IIA or pMLC, and SLP-76 microclusters are typically associated with the inner or outer margins of the myosin rings (Fig. S2C, bottom row; Fig. S2E). This is consistent with the intensities of myosin IIA and pMLC respectively peaking ∼0.26 µm and ∼0.45 µm from the centers of SLP-76 microclusters (Fig. S2D, right).

To examine the role of myosin IIA in the transport of SLP-76 microclusters, we pretreated J14.SY cells for 30 min with the myosin ATPase inhibitor blebbistatin or with H1152 or Y27632, which are inhibitors of the Rho kinase (ROCK) subfamily of myosin light chain kinases. Kymographs are used to illustrate the rates of microcluster movement within the indicated regions (Fig. 2A,B). All three compounds inhibit the centripetal movement of SLP-76 microclusters, demonstrating that Rho kinase-dependent contractility is a driver of SLP-76 centralization. To verify the impacts of blebbistatin and H1152 on myosin filament organization, we treated Jurkat E6.1 cells expressing an EGFP-tagged myosin IIA chimera with blebbistatin or H1152 (Fig. 2C). Blebbistatin retards the continuous transport of myosin IIA filaments from the cell periphery without disrupting these filaments or perturbing the ‘cells’ defined by myosin IIA. In contrast, H1152 prevents the appearance of myosin filaments, consistent with its mode of action. However, we were unable to suppress the centripetal movement of SLP-76 microclusters by knocking down myosin IIA (data not shown). A parallel effort to characterize the SLP-76 microcluster proteome provided new insight into this longstanding problem (Hammer and Burkhardt, 2013; Le Floc'h and Huse, 2015). These studies involved the stimulation of J14.SY cells using beads coated with anti-CD3ε antibody and recombinant human VCAM-1, the fixation and sonication of the cells, and the magnetic recovery of bead-bound signaling complexes (Fig. S3A). This approach activates cells and enriches TCR-associated signaling molecules (Fig. S3B,C). Mass spectrometry of individual bands yields proteins with molecular masses appropriate to their positions within the original gel (Fig. S3D,E). The resulting hits include proteins involved in cytoskeletal function, membrane transport and signal transduction (Fig. S3F, see Table S2 for full dataset). Myosin IIA (MYH9) is captured by stimulatory beads, but not by beads coated with a pro-adhesive and non-stimulatory antibody (Fig. S3G,H). Peptides unique to myosin IIB (MYH10) are found in these complexes, even though primary murine T cells only express myosin IIA (Fig. S3F) (Jacobelli et al., 2004). When transfected into J14.ST cells (analogous to the J14.SY cell line but expressing SLP-76.TRT), EGFP-tagged variants of myosin IIA and IIB assemble into filaments that associate with SLP-76 microclusters (Fig. 2D). Transitive western blots employing specific antibodies confirm that myosin IIB is expressed in Jurkat T cells, although at lower levels than myosin IIA (Fig. S3I). Endogenous myosin IIB is also assembled into filaments after TCR ligation, and these myosin IIB filaments display a sensitivity to SLP-76 expression that is identical to myosin IIA (Fig. S3J). Finally, we were able to show that the suppression of myosin IIA and IIB impairs the centripetal transport and central accumulation of SLP-76 microclusters (Fig. 2E–H).

To determine whether Arp2/3-dependent actin flows, formin-dependent actomyosin filaments and microtubule-dependent dynein motors contribute to the movement of SLP-76 and myosin, we visualized the dynamic redistribution of SLP-76 and MLC-9 (MYL9) in the presence or absence of inhibitors of these systems (Bunnell et al., 2002; Hashimoto-Tane et al., 2011). Maximum-over-time (MOT) projections (Fig. 2I) and kymographs (Fig. S4A) show the direction, extent and speed of movement. The Arp2/3 inhibitor CK666 abolishes T cell spreading and precludes the formation of SLP-76 microclusters (data not shown). A photostable inhibitor of myosin ATPase function, para-amino-blebbistatin (paBB), suppresses microcluster centralization (Kolega, 2004; Varkuti et al., 2016). The formin inhibitor SMIFH2 prevents SLP-76 microcluster centralization and disrupts MLC-9-labeled filaments, leaving small specks of MLC-9 (Fig. 2I, white arrows versus red arrowhead). The microtubule-disrupting agent colchicine impairs the movement of SLP-76 microclusters, but also prevents the central accumulation of MLC-9 (Fig. S4B). Finally, the dynein ATPase inhibitor EHNA also prevents the accumulation of SLP-76 and MLC-9 at the center of the contact (Fig. S4B). Collectively, these findings suggest that the centralized myosin filaments observed here are analogous to the actomyosin arcs that drive the transport of TCR microclusters and extend this model by demonstrating that dynein-dependent interactions between microtubules and peripheral actomyosin fibers dictate the final arrangement of actomyosin and the ultimate position of SLP-76 microclusters. We also note that paBB impairs the extension of microtubules into the periphery of the contact and impedes the polarization of the MTOC towards the contact interface (Fig. S4C). Thus, dynein motors embedded in the actomyosin system may reciprocally regulate MTOC polarization and centripetal transport processes (Nath et al., 2016; Maskalenko et al., 2020).

Immobilized VLA-4 ligands prevent the centripetal movement of TCR-induced SLP-76 microclusters and slow actin flows (Nguyen et al., 2008). In response, myosin IIA and pMLC shift from radial arrays into peripheral, circumferential structures (Fig. 3A, left-hand panels, see enlarged views for greater detail). However, myosin filaments continue to incorporate SLP-76 microclusters and to terminate at or near to peripheral SLP-76 microclusters. Using an objective with a higher numerical aperture, these myosin IIA filaments can be resolved as regularly spaced, sarcomere-like puncta (Fig. 3A, right-hand main panels and magnified views). SLP-76 microclusters interdigitate between these puncta in the same manner as TCR microclusters (Murugesan et al., 2016). Immobilized VCAM-1 also shifts MLC-9 into a distal ring and retains SLP-76 microclusters outside of this ring (Fig. 3B,C). LFA-1 ligation by immobilized ICAM-1 also shifts MLC-9 into a circumferential ring, but this ring is displaced inward, and SLP-76 microclusters are arrested inside the pSMAC (Fig. 3B,C). Plots of SLP-76 and MLC-9 intensity as a function of distance from the outer edge of the cell confirm these effects (Fig. S5A). Thus, integrin ligation inhibits the centripetal flow of myosin filaments and hinders the centripetal movement of SLP-76 microclusters, but VLA-4 and LFA-1 act at distinct sites.

Centralized SLP-76 microclusters interact with β1 integrin and talin, indicating that integrins could engage SLP-76 microclusters and direct their centripetal transport (Ophir et al., 2013). In TCR-stimulated J14.SY cells, SLP-76 microclusters either align with radial structures containing α4 integrin and talin or are found within α4 ‘rings’ that are similar in size to actomyosin rings (Fig. 3D, see yellow brackets in magnified views). VLA-4 ligation does not alter the segregation of SLP-76 from α4 integrin and talin but favors more uniform distributions of α4 integrin and talin. When cells are sheared from substrates lacking VCAM-1, SLP-76 microclusters remain attached, but very little α4 is retained (Fig. 3E). In contrast, VCAM-1 promotes the retention of a reticular network of α4 integrin and traps SLP-76 microclusters in the cavities of this network, implying that SLP-76 is tightly coupled to VLA-4. Finally, in J14.ST cells expressing tagged β1 and β2 integrins, the ligation of either VLA-4 or LFA-1 immobilizes SLP-76 microclusters, even though these tagged integrins are excluded from SLP-76 microclusters (Fig. 3F). In sharpened images, β1 and β2 appear to surround and overlap minimally with SLP-76 microclusters, as observed for α4 integrin (Fig. 3G). The radial positions α4 and β2 integrins relative to SLP-76 microclusters, derived using raw images, confirm this effect and indicate that integrins and SLP-76 microclusters are enriched in distinct but adjacent subdomains (Fig. S5B).

Because our model system does not require integrin co-ligation for T cell attachment, it is ideally suited to the analysis of how integrins contribute to microcluster transport. In J14.SY control cells, TCR-induced SLP-76 microclusters accumulate in the center of the synapse (Fig. 4A; Movie 1). In A1 cells expressing SLP-76.TRT (A1.ST), SLP-76 microclusters centralize normally (Fig. 4A; Movie 2) (Romzek et al., 1998). By contrast, cells lacking β2 (see Fig. 4B) cannot accumulate SLP-76 microclusters in the center of the synapse, even though the suppression of β2 does not eliminate microcluster movement in the distal, lamellipodial region of the cell (Fig. 4A; Movie 3). These cells were also scored for microcluster localization using four categories: central (SLP-76 microclusters are absent from the center of the contact); partial (SLP-76 microclusters adopt a multifocal and/or asymmetric pattern); peripheral (SLP-76 microclusters are absent from the center of the contact); and dispersed (SLP-76 microclusters are uniformly distributed). As noted, the loss of β1 integrin does not alter the behavior of SLP-76 microclusters (Fig. 4C). In contrast, the suppression of β2 integrin decreases the proportion of cells scored as ‘central’ and increases the fractions of cells scored as ‘peripheral’ or ‘dispersed’. This effect is eliminated by reconstitution with αL.mCFP and β2.mYFP (Fig. 4D,E). The depletion of integrins or integrin associated proteins does not alter the expression of SLP-76.YFP or the TCR, β1 or β2 (Fig. S1C,D). Using a similar scoring system to assess a β2-deficient version of our cytoskeletal reporter cell line (Fig. 4F), we confirmed that the suppression of β2 decreases the central accumulation of MLC-9 and SLP-76 microclusters (Fig. 4G,H). Finally, we examined the distribution of LFA-1 in the contacts formed by Jurkat T cells. High-affinity mAb24-reactive LFA-1 forms a prominent ring ∼7 µm from the center of the contact (Fig. S5C, upper). Radial analyses reveal that a subset of high-affinity LFA-1 is recruited to the center of the synapse, with SLP-76. When superimposed on earlier data, the distribution of open LFA-1 follows that of myosin IIA (Fig. S5C, lower). Thus, active β2, but not β1, integrins mediate the centralization of SLP-76 microclusters and the transport of contractile myosin filaments to the center of the synapse.

The integrin-activating adapters SKAP55 and ADAP enhance the persistence and movement of SLP-76 microclusters (Ophir et al., 2013; Lewis et al., 2018). The N-terminal dimerization motif in SKAP55 recruits several LFA-1-activating proteins, including RAPL (RASSF5), MST1 (STK4), RIAM (APBB1IP) and talin-1, suggesting that SKAP55 could link SLP-76 to LFA-1 (Kliche et al., 2006; Menasche et al., 2007; Lee et al., 2009; Raab et al., 2010; Kliche et al., 2012). J14.SY cells lacking SKAP55 (i.e. JSKAP.SY cells) are unable to form TCR-induced central clusters of myosin IIA and pMLC, and are less able to centralize SLP-76 microclusters, whereas reconstituted JSKAP.SY cells behave like the parental J14.SY line (Fig. S6A,B). Previously, we showed that a SKAP55-derived tandem dimer (TD) of SH3 domains that lacks a critical talin-1-binding and integrin-activating domain supports the persistence and centralization of SLP-76 microclusters (Ophir et al., 2013). This chimera also restores the centralization of MLC-9, implying that talin-1 is dispensable for both forms of movement (Fig. S6B).

The acquisition of an extended, high-affinity conformation by integrins requires the integrin-activating proteins, talin-1 and kindlin-3, which bind, respectively, to the proximal and distal NPxF/Y motifs in the tails of β-integrins (Morse et al., 2014). The near elimination of talin-1 has virtually no effect on the centralization of SLP-76 microclusters (Fig. 5A–C; Movie 4). By contrast, the partial suppression of kindlin-3 impairs microcluster centralization, yielding effects comparable to the suppression of β2 (Fig. 5A–C; Movie 5). Similarly, the suppression of kindlin-3, but not of talin-1, prevents the co-accumulation of active myosin with SLP-76 microclusters (Fig. 5D,E).

To verify that these centripetal movements depend on kindlin-3 but not talin-1, we transfected a J14.ST β2 knockdown (β2-KD) line with wild-type (WT) αL.mCFP and with alanine-replacement NPxF motif-mutant versions of β2.mYFP (Fig. 6A). As expected, the suppression of β2 alters the patterning of SLP-76 and pMLC in J14.ST cells (Fig. 6B,C). Reconstitution with wild-type αL.mCFP and β2.mYFP reproduces the behavior of the parental line. A β2 chimera that lacks the proximal talin-binding NPLF motif also restores the parental patterns of myosin and SLP-76. In contrast, β2 chimeras that lack the distal kindlin-3-binding NPKF motif are impaired in their ability to centralize myosin and SLP-76, but are not as impaired as the β2-KD line. Finally, cells reconstituted with the dual NPxF mutant are more defective than cells expressing the NPKF mutant, implying that the proximal NPLF motif weakly impacts centripetal movements in cells lacking the distal NPKF motif. Thus, kindlin-3, rather than talin-1, plays a decisive role in the LFA-1-mediated central accumulation of myosin filaments and SLP-76 microclusters.

The contractile filaments observed here resemble stress fibers, implying that other stress-fiber constituents contribute to the centralization of SLP-76 microclusters (Tojkander et al., 2012). We observed that the shRNA-mediated suppression of zyxin, filamin-A (FLNA), or vinculin (VCL) either abolishes or impairs the centralization of SLP-76 microclusters (Fig. 7A; Fig. S1B,C). A later screen revealed that α-actinin-1 (ACTN1) and α-actinin-4 (ACTN4) also participate in SLP-76 microcluster centralization, although α-actinin-1 appears to play a more decisive role (Fig. S7). Consistent with these findings, filamin-A and α-actinin-4 also appear in our analysis of the shear-resistant signaling complexes retained on stimulatory beads after sonication (Fig. S3F). To characterize the distribution of zyxin in the immune synapse, we stained for zyxin in our cytoskeletal reporter line. Zyxin puncta appear throughout the immune synapse, but do not consistently colocalize with SLP-76 microclusters and do not accumulate in the center of the contact (Fig. 7B). Nevertheless, the suppression of zyxin reduces the central accumulation of MLC-9 and the centralization of SLP-76 microclusters (Fig. 7C,D).

To determine how proteins that impact microcluster and myosin transport influence T cell activation, we examined the upregulation of the activation marker CD69 and the secretion of interleukin (IL)-2 by J14.SY cells stimulated on substrates identical to those used in our imaging assays. Integrin β2, kindlin-3 and zyxin are all required for the maximal upregulation of CD69 by J14.SY cells; however, talin-1 is not required (Fig. 8A,B). In contrast, all four proteins are required for optimal IL-2 secretion (Fig. 8C). All four knockdown cell lines adhere to anti-CD3ε-coated plates over a broad range of shear forces, indicating that the signaling defects observed in these lines are not secondary to defects in contact formation (Fig. 8D) (Nguyen et al., 2008; Ophir et al., 2013).

Actomyosin arcs clearly contribute to the centripetal movement of TCR microclusters within the immune synapse (Yi et al., 2012; Yu et al., 2012; Hammer and Burkhardt, 2013; Murugesan et al., 2016; Hong et al., 2017; Hammer et al., 2019). However, the participation of these arcs in the centripetal transport of SLP-76 microclusters, which play distinct and critical roles in T cell activation, has received less study (Babich et al., 2012). Here, we visualized the interactions of SLP-76 microclusters with actively contracting myosin filaments using antibodies against phosphorylated myosin light chain (pMLC) and a fluorescent chimera of human MLC-9. This allowed us to demonstrate that, in Jurkat T cells, TCR ligand density and SLP-76 expression favor the co-accumulation of myosin with SLP-76 microclusters in compact central structures. The centripetal movement of SLP-76 microclusters is blocked by the same factors that prevent the centralization of TCR microclusters: Rho-family kinase inhibitors, myosin II ATPase inhibitors, and the formin inhibitor SMIFH2. As previously reported, the partial suppression of myosin IIA does not suppresses microcluster transport (Babich et al., 2012). However, we show that myosin IIB is expressed in Jurkat T cells, and that the simultaneous suppression of myosin IIA and myosin IIB inhibits the sustained centripetal movement of SLP-76 microclusters in these cells. Jointly, these findings indicate that TCR and SLP-76 microclusters are identically dependent on the contraction of formin-nucleated actomyosin filaments, even though the overall topology of myosin in T cells expressing exogenous SLP-76 deviates from the largely circumferential pattern observed in unmanipulated Jurkat T cells and primary T cells.

Our data confirm that SLP-76 microclusters are enclosed by, but remain distinct from, actomyosin-containing rings (Hashimoto-Tane et al., 2016). The separation of SLP-76 microclusters from their surrounding actin-containing structures is consistent with previous nanoscale analyses of LAT and SLP-76 microclusters (Sherman et al., 2011, 2016). Remarkably, these rings are unaffected by the expression of exogenous SLP-76. VLA-4 and LFA-1 are both excluded from SLP-76 microclusters but are enriched in the surrounding region, at a distance compatible with their participation in actomyosin rings. The recruitment of integrins into these rings is less pronounced than in the transient structures observed by Hashimoto-Tane et al. This may reflect the later timepoints used in our study, or other differences between our model systems. Importantly, these weak integrin rings form in the absence of exogenous integrin ligands. Thus, ADAP could stabilize SLP-76 microclusters in cells stimulated via the TCR alone by tethering nascent microclusters to surrounding actomyosin rings, even when the salient exogenous integrin ligands are absent (Lewis et al., 2018).

Our observations also provide insights into processes occurring in primary T cells and cells stimulated via physiological ligands. SLP-76 microclusters accumulate in the center of the stimulatory contacts formed by human peripheral blood T cell (PBT) blasts (Nguyen et al., 2008). In addition, SLP-76 microclusters arise in the distal regions of the T cell–B cell conjugates induced by antigenic peptides or superantigens, and these microclusters accumulate in the centers of the associated synapses, even when exogenous SLP-76 is absent (Yokosuka et al., 2005; Bunnell et al., 2006; Purbhoo et al., 2010; Roybal et al., 2015). The levels of SLP-76 that support sustained centripetal movements can be as low as 2.3-fold above those in the parental Jurkat E6.1 cell line. Comparable and greater increases in SLP-76 expression accompany the reactivation of human memory T cells and the activation of primary murine T cells by antigen (Clements et al., 1998; Hussain et al., 2002). Thus, physiological increases in SLP-76 expression could contribute to the centralization of microclusters and contractile actomyosin filaments in primary T cells. TCR-transgenic CD8⁺ T cell blasts activated via high-quality ligands generate actomyosin arcs that are displaced towards the center of the immune synapse, partially separate from the distal actin cytoskeleton, and ultimately define the central domain in which TCR microclusters accumulate (Hong et al., 2017). Human T cell blasts stimulated on high-density TCR ligands develop ‘purse string’ like myosin structures that are also capable of separating from distal myosin filaments (Fig. S2A). These events closely parallel the formation of a central pool of SLP-76 microclusters bounded by a distinct inner ring of contractile myosin (Fig. 1B; Fig. S1D). The critical role of SLP-76 in these processes may be mediated via ADAP, as mutations that disrupt the interaction of ADAP with SLP-76 cause the immobilization and premature termination of SLP-76 microclusters and prevent the segregation of immune synapses into well-defined central and peripheral domains (Wang et al., 2004; Lewis et al., 2018). Thus, the formation of a well-defined cSMAC by agonist-stimulated effector T cells may require the compaction of contractile actomyosin structures by SLP-76- and ADAP-dependent processes (Grakoui et al., 1999; Schubert et al., 2012).

The spoke-like actomyosin filaments observed in SLP-76-expressing cells are oriented in a manner that could support the centripetal transport of SLP-76 microclusters via contraction along the long axis of these filaments. However, this would not explain the dependence of the central accumulation of TCR and SLP-76 microclusters on dynein motors and the microtubule cytoskeleton (Bunnell et al., 2002; Hashimoto-Tane et al., 2011). Based on our findings, we propose that SLP-76 centralization involves three distinct phases (Fig. S8A). In the first phase, SLP-76 microclusters are transported through the dSMAC via interactions with lamellipodial actin flows (Ditlev et al., 2019). In the second phase, SLP-76 microclusters engage dynein motors, but their movements through the pSMAC are constrained by interactions with slowly contracting actomyosin arcs (Murugesan et al., 2016). Finally, SLP-76 microclusters reach the inner boundary of the pSMAC and transition to a dynein-dependent mode that involves either their release from actomyosin arcs or the deformation of these arcs towards the center of the synapse. In the latter case, actomyosin spokes evolve from arcs experiencing centripetal forces applied by dynein motors. In this model, SLP-76 enhances the recruitment of dynein to the peripheral actomyosin system via the dynein-binding adapter ADAP, favoring arc deformation (Combs et al., 2006). The SLP-76-dependent coupling of dynein motors to lamellar actomyosin networks may also provide a platform for MTOC polarization and the maintenance of synaptic symmetry (Liu et al., 2013; Nath et al., 2016; Maskalenko et al., 2020). Thus, the movement of SLP-76 microclusters may provide insights into synapse stability, kinapse formation via symmetry breaking, and microtubule-dependent processes that support T cell signaling and cytolytic function. These systems might also be exploited to stabilize synapses and augment the effector functions of CAR T cells (Mayya et al., 2018, 2019).

VLA-4 and LFA-1 are specialized for distinct functions, such as rolling adhesion and synapse formation (Chigaev and Sklar, 2012). Here, we report that the suppression of β2 integrin mimics the suppression of myosin IIA and IIB and impairs the centripetal movement of microclusters and actomyosin filaments, even though the elimination of β1 integrin has no such effect. This unique role for LFA-1 may reflect its ability to stimulate the formation of lamellum-like ‘actin clouds’ via the adaptor ADAP (Suzuki et al., 2007). In addition, the loss of kindlin-3 impairs the adhesive function of LFA-1 to a greater extent than loss of VLA-4 (Manevich-Mendelson et al., 2009; Feigelson et al., 2011). Thus, the unique functions of LFA-1 might also reflect the preferential usage of specific effectors.

Although VLA-4 and LFA-1 both restrict the inward flow of actin when they encounter immobile ligands, VCAM-1 favors the immobilization of myosin filaments and SLP-76 microclusters in the extreme periphery of the contact, whereas ICAM-1 favors the retention of these structures in the pSMAC (Nguyen et al., 2008; Comrie et al., 2015b; Jankowska et al., 2018). The latter effect requires high doses of ICAM-1, most likely due to the limited expression of LFA-1 on Jurkat T cells (Jankowska et al., 2018). However, the distinct impacts of VLA-4 and LFA-1 on SLP-76 microclusters may reflect differences in their distributions, as high-affinity LFA-1 associates with actomyosin arcs, whereas the open form of VLA-4 is enriched in lamellipodia (Fig. S8B) (Mittelbrunn et al., 2004; Hyun et al., 2009; Yi et al., 2012; Murugesan et al., 2016).

In the absence of exogenously added integrin ligands, β2 integrin and kindlin-3 are required for the centripetal movement of SLP-76 microclusters and contractile myosin filaments, while β1 integrin and talin-1 are dispensable. This is consistent with the role of kindlin-3 in the centripetal movement of TCR microclusters and with the observation that the ‘tandem dimer’ (TD) variant of SKAP55 reconstitutes the centralization of contractile myosin filaments and SLP-76 microclusters, even though it can neither recruit talin-1 nor initiate integrin-dependent adhesion (Ophir et al., 2013; Kondo et al., 2017). These findings are incompatible with conventional models of integrin activation, where integrins transition from a closed configuration (E⁻H⁻), to an extended, talin-dependent state (E⁺H⁻), and then to a kindlin-dependent extended, high-affinity configuration (E⁺H⁺). However, LFA-1 can progress to the E⁺H⁺ state via a kindlin-dependent high-affinity configuration that does not extend fully (E⁻H⁺) (Lefort et al., 2012; Fan et al., 2016, 2019; Wen et al., 2021). Since kindlin-3 might be directly recruited into SLP-76 microclusters via ADAP, this E⁻H⁺ state may be achievable even when SKAP55 is replaced with the talin-nonbinding TD chimera, described above (Kasirer-Friede et al., 2014). Subsequently, LFA-1 in the E⁻H⁺ state must couple SLP-76 microclusters to force-generating cytoskeletal systems, such as actomyosin arcs. Since dynein plays roles in both microcluster and actomyosin transport, and binds to ADAP, the E⁻H⁺ state might also influence synaptic architecture by facilitating the peripheral capture of microtubules or the loading of microcluster-bound dynein motors onto microtubules (Combs et al., 2006; Yi et al., 2013; Byron et al., 2015). Another important feature of this model is its ability to explain the preferential centralization of high-affinity (mAb24⁺) LFA-1 relative to extended (Kim127⁺) LFA-1 in T cell–B cell conjugates, as LFA-1 in the E⁻H⁺ state engages ICAM-1 in cis and would be expected to experience less resistance to lateral movement (Comrie et al., 2015a,b; Fan et al., 2016, 2019).

Although signaling via centralized signaling complexes is ultimately terminated, the mechanical forces generated during microcluster transport contribute to T cell activation (Udagawa and McIntyre, 1996; Lee et al., 2003; Mossman et al., 2005; Varma et al., 2006; Ilani et al., 2009; Vardhana et al., 2010; Judokusumo et al., 2012; Kumari et al., 2012; Murugesan et al., 2016; Hong et al., 2017; Jankowska et al., 2018). Here, we demonstrate that factors required for the deformation of the cortical actomyosin system and the centripetal transport of SLP-76 microclusters, namely, LFA-1, kindlin-3, and zyxin, are required to upregulate CD69. In contrast, talin-1, which plays no role in these transport processes, has no impact on CD69 expression. Our work provides useful insights into animal models, where a kindlin-3-binding site in β2 integrin is required for optimal CD69 expression, even though talin-1 is dispensable for this event (Wernimont et al., 2011; Morrison et al., 2015). Our findings also highlight the unique roles of LFA-1 and kindlin-3 in the dynein-dependent centripetal movements of actomyosin arcs and SLP-76 microclusters, and link these transport processes to subsequent TCR-proximal signaling events. The tensile forces exerted during these processes may drive changes in gene expression via mechano-sensitive pathways, including the phosphorylation of CasL and the nuclear translocation of MRTF-A (Yu et al., 2012; Hong et al., 2017; Guenther et al., 2019). The resolution of these issues will clarify the role of mechano-transduction in immune cell activation and will shed light on how myosin II inhibitors, which are being explored as therapies for multiple human diseases, are likely to impact immune cell function (Bond et al., 2013; Orgaz et al., 2020; Naydenov et al., 2021).

All antibodies were from commercial vendors and bound the expected targets when used at the indicated dilutions as verified using knockdowns, mutant cell lines and chimeric proteins. Antibodies used for immunofluorescence (IF), flow cytometry (FACS) and western blotting (WB) were against: myosin IIA (Poly19098, Biolegend #909801, San Diego, CA; IF 1:500, FACS 1:2500, WB 1:2000) and myosin IIB (Sigma #M7939, St Louis, MO; IF 1:200, FACS 1:2500, WB 1:1000). Goat-anti-rabbit-IgG conjugated to Alexa Fluor 647 (Invitrogen, #A21245, Carlsbad, CA) was used at 1:1000 for immunofluorescence and flow cytometry. Antibodies used for immunofluorescence were against: integrin α4 [Chemicon #CBL485Z (44H6), Burlington, MA; 1:100], phospho-myosin light chain 2 (Ser19) [Cell Signaling Technology (CST) #3675, Danvers, MA; 1:50], talin-1 [Sigma #T3287 (8D4); 1:500], zyxin (Thermo Fisher Scientific #PA5-78236, Waltham, MA; 1:500), and goat anti-rabbit-IgG conjugated to Alexa Fluor 568 (Invitrogen #A11036, Carlsbad, CA; 1:1000), goat anti-mouse-IgG conjugated to Alexa Fluor 647 (Invitrogen #A21240; 1:1000), and goat-anti-mouse-IgG1 conjugated to Alexa Fluor 647 (Invitrogen #A21240; 1:1000). Antibodies used for western blotting were against: integrin β2 [CST #73663S (D4N5Z); 1:1000], GADPH [CST #2118S (14C10); 1:1000], Aequorea victoria GFP [Clontech #632381 (JL-8); 1:2000], kindlin-3 [CST #13843S; 1:1000], talin-1 [CST #4021S (C45F1); 1:1000], zyxin (CST #3553S; 1:1000), SLP-76 (CST #4958S;1:1000), and goat anti-rabbit-IgG conjugated to HRP (Thermo Fisher Scientific PI31462; 1:10,000), goat anti-mouse-IgG conjugated to HRP (Thermo Fisher Scientific PI31432; 1:10,000), and goat anti-rabbit-IgG conjugated to DyLight800 (CST #5151P; 1:30,000). Antibodies used for flow cytometry include: CD69-A647 [Biolegend #310918 (FN50); 1:20], CD3-PE [BD Pharmingen, San Diego, CA, #555333 (UCHT1); 1:8), CD29-APC (BD Pharmingen #561794 (MAR4); 1:20] and CD18-APC [BD Pharmingen 551060 (6.7); 1:20]. T cells were stimulated using anti-CD3ε (OKT3) and anti-CD28 (9.3) from BioXCell (Lebanon, NH). Anti-CD43 (BD Pharmingen #555474 (1G10)] was used to promote adhesion without triggering T cell activation. Recombinant human VCAM-1 (862-VC-100) and ICAM-1 (720-IC-050) were from R&D Systems (Minneapolis, MN). Para-amino-blebbistatin (DR-AM-89) was from OptoPharma (Budapest, Hungary). Blebbistatin (BML-EI315), H1152 (ALX-270-423), and Y27632 (BML-EI299) were from Enzo Life Sciences (Farmingdale, NY). SMIFH2 (S4826), CK666 (SML0006), and EHNA (E114) were from Sigma. Colchicine (64-86-8) was from Cayman Chemical (Ann Arbor, MI). Linear polyethylenimine 25,000 (23966-2) was from Polysciences Inc. (Warrington, PA).

Lentiviral shRNA expression vectors targeting α-actinin-1 (TRCN0000055827), α-actinin-4 (TRCN0000055784), β2 integrin (TRCN0000236135), filamin-A (TRCN0000062528), kindlin-3 (TRCN0000431404), myosin IIB (TRCN0000123074), SKAP55 (TRCN0000006369), talin-1 (TRCN0000123104), vinculin (TRCN0000116755), and zyxin (TRCN0000286232) were purchased from Sigma. The lentiviral packaging vector psPAX2 (Addgene #12260) the VSV.G pseudotyping vector pMD2.G (Addgene #12259) were deposited by Didier Trono. The RFP-marked shRNA expression vector was created by subcloning mRFP1 (Roger Tsien, UCSD, USA) into the Clontech EGFP-n1 backbone, generating CMV-mRFP1-n1. The H1 promoter-driven shRNA expression cassette targeting human myosin IIA was assembled by annealing oligonucleotides NRS801 through NRS804 and subcloning the resulting insert into the BglII and HindIII sites of pFRT.H1p (Dan Billadeau, Mayo Clinic, USA). The entire H1-shRNA cassette was amplified by PCR using primers NRS753 and NRS754. The resulting PCR product was cut with AseI and NdeI and was installed into the AseI site upstream of the CMV promoter in mRFP1-n1, yielding the vector H1p-shMYH9/CMV-mRFP-n1. The lentiviral vector used to expresses SLP-76.YFP has been described previously; the lentiviral vector used to expresses SLP-76.TRT was developed from the former vector by replacing YFP with TagRFP-Turbo from TRT-n1 (Lewis et al., 2018). The lentiviral vector used to express F-Tractin.mRuby3 and mTurquoise.MLC-9 was developed by Tobias Meyer (Addgene #85146) (Hayer et al., 2016). The pEYFP-Tubulin vector encoding a tagged form human α-tubulin was obtained from Clontech (#6118-1). The vectors expressing EGFP-tagged forms of human myosin IIA and IIB were from Addgene (Addgene #11347 and #11348) deposited by by Robert Adelstein (Wei and Adelstein, 2000). Vectors expressing β2.mYFP and αL.mCFP were provided by Timothy Springer (Harvard Medical School, Boston, MA, USA) (Kim et al., 2003). An intermediate tailless form of β2.mYFP was prepared by excising the region between the EcoR I and Age I sites and installing a PCR product generated using primers KPE017 and KPE018. Constructs expressing β2.mYFP tail mutants were prepared by installing synthetic gBlocks from IDT (KPEG026-028) into the tailless vector using a silent Age I site and Gibson Assembly Master Mix from New England BioLabs (E2611S). Oligonucleotide sequences are provided in Table S1.

Wild-type (E6.1) Jurkat leukemic T cells were obtained from the ATCC (TIB-152). SLP-76-deficient Jurkat J14 cells was a gift of Arthur Weiss (University of California San Francisco, CA, USA). β1-deficient Jurkat A1 cells were a gift of Yoji Shimizu (University of Minnesota Medical School, USA). The J14.SY, E6.1 ZAP-70.YFP, and E6.1 EGFP.β-actin cell lines have been described previously (Bunnell et al., 2001, 2002, 2006). E6.1 YFP.Tubulin was generated by transfecting E6.1 cells with linearized pEYFP-Tubulin and selecting in neomycin. Stable cell lines were generated via the lentiviral transduction of E6.1 cells with SLP-76.YFP, J14 cells with SLP-76.TRT (J14.ST), A1 cells with SLP-76.TRT (A1.ST) or SLP-76.YFP (A1.SY), and J14.SY cells with F-Tractin.mRuby3-P2A-mTurquoise.MLC-9 (J14.SY-CRL). Stable knockdown lines were generated via the transduction of J14.SY, J14.ST, or J14.SY-CRL cells with lentiviral particles targeting β2, SKAP55, talin-1, kindlin-3, zyxin, filamin-A, vinculin, α-actinin-1, or α-actinin-4. J14.SY cells lacking SKAP55 are referred to as JSKAP.SY cells. Knockdown lines were routinely re-derived to minimize adaptations to the loss of these proteins. Jurkat T cells and their derivatives were maintained in RPMI-1640 medium supplemented with 10% fetal bovine serum, 2 mM L-glutamine and 10 μg ml⁻¹ ciprofloxacin (‘complete RPMI’). Transient transfections were performed using an ECM 830 electroporator (BTX), as described previously (Bunnell et al., 2006). HEK 293T cells were obtained from Lawrence Samelson (NIH) and maintained in DMEM supplemented with 10% fetal bovine serum (FBS), 2 mM L-glutamine, 100 units ml⁻¹ of penicillin and 100 μg ml⁻¹ of streptomycin, as described previously (Lewis et al., 2018). All cell lines retain, lack or express the expected proteins and chimeras, as assessed by western blotting, flow cytometry and imaging.

Primary human T cells were isolated from human blood obtained commercially via New York Biologics. All samples were de-identified and acquired from consenting donors. Briefly, human primary CD4⁺ and CD8⁺ T cells were isolated from blood through an initial 30 min incubation with CD4⁺ and CD8⁺ human T cell Enrichment cocktails (Stem Cell Technologies, Cambridge, MA cat 15062 and 15063). Following this initial incubation, cells were diluted 1:1 in PBS and 2% FBS followed by Ficoll separation via centrifugation (1200 g for 20 min, no brake). Buffy coat layers were removed, diluted in PBS with 2% FBS, centrifuged (300 g for 10 min, low brake), and cultured in complete RPMI (see above) at a density of 10⁶ cells ml⁻¹ on tissue culture-treated Petri dishes coated with anti-CD3ε OKT3 (10 μg ml⁻¹) and anti-CD28 (2 μg ml⁻¹) for 3 days. Beginning on day 2, human recombinant IL-2 (Peprotech, 200-02; Rocky Hill, NJ) was added at a concentration of 200 units ml⁻¹. After day 3 of plate stimulation, CD4⁺ and CD8⁺ human primary T cells were removed from stimulatory coated petri dishes and cultured in T75 flasks in complete RPMI plus human recombinant IL-2 (200 units ml⁻¹).

Packaging reactions used HEK 293T cells at ∼70% confluency. For each 10 cm plate, a mix of 18 μl 1 mg ml⁻¹ linear polyethylenimine and 502 μl serum-free DMEM was incubated at room temperature for 5 min. In parallel, lentiviral transfer vectors (3 μg) were combined with psPAX2 (1.5 μg) and pMD2.G (0.5 μg) in 40 μl of serum-free DMEM. These cocktails were combined and incubated at room temperature for 30 min before adding dropwise to 293T cells. After 12–15 h, the transfection mixture was replaced with fresh DMEM. Lentiviral supernatants were harvested 36–48 h post transfection and any residual cells were removed using 0.45 µm filters (Thermo Fisher Scientific, SLHV033). To transduce Jurkat cells, equal parts of fresh RPMI, Jurkat cells, and lentiviral supernatants were combined. After 48 h, cells in the viral media were rinsed and immediately placed into selection medium containing either puromycin (1 µg ml⁻¹) or zeocin (200 µg ml⁻¹), as appropriate. After an additional 48 h, surviving cells were rinsed into fresh RPMI.

Jurkat cells were lysed, as described previously, in ice-cold buffer containing 20 mM Tris-HCl (pH 8.0), 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 10 mM NaF, 1 mM Na₃VO₄ and Complete protease inhibitor cocktail (Roche) (Lewis et al., 2018). Proteins were resolved using Nu-PAGE 4–12% Bis-Tris gels from Invitrogen (NP0336), transferred to PVDF membranes (Immobilon-P or -F; Sigma, IPVH00010 or IPFL00010), and blotted using antibodies at the concentrations indicated above. All incubations and washes were in Tris-bufferred saline (TBS, 150 mM NaCl, 50 mM Tris-HCl, pH 7.4) with 0.1% Tween-20 (TBST) with the following exceptions: blocking was performed in TBS with 5% BSA, primary antibodies were diluted in TBST with 1% BSA and 0.02% sodium azide, and fluorescently stained gels were subjected to a final rinse in TBS. Detection was performed using SuperSignal West Pico (Invitrogen, 34580) and CL-Xposure film (Thermo, PI34091) or using a Odyssey CLx imager (LI-COR Biosciences, Lincoln, NE).

Imaging and functional assays were performed in glass-bottomed 96-well plates (#4953) from Matrix Technologies (Indianapolis, IN). First, plates were incubated for 15 min in 1 M HCl in 70% ethanol and dried at 42°C for 30 min. Plates were then coated with 0.01% poly-L-lysine (Sigma P8920) for 20 min and dried at 42°C for 30 min. Subsequently, plates were sequentially coated with anti-CD3ε (OKT3), typically at 10 µg ml⁻¹ in PBS, and then, if required, with recombinant human (rh)VCAM-1 (1 µg ml⁻¹), rhICAM-1 (20 µg ml⁻¹) or anti-CD43 (10 µg ml⁻¹). Each binding was performed for 1 h at 37°C. Finally, plates were blocked with 1% BSA for 1 h at 37°C. Wells were rinsed with and stored in PBS. Plates were sealed and stored at 4°C for up to 48 h before use. Cells were stimulated on these glass substrates in complete RPMI (see above) supplemented with 25 mM Hepes (Cambrex Bio Science, East Rutherford, NJ).

For live-cell assays, imaging plates were maintained at 37°C, as described previously (Bunnell et al., 2003); Jurkat T cells were injected into the base of the well, allowed to initiate contacts, and then imaged at regular intervals for the indicated lengths of time. For fixed cell studies, Jurkat T cells were plated onto stimulatory substrates, allowed 7 min at 37°C to attach and spread, fixed in 1% PFA for 30 min at 37°C and rinsed with PBS. For immunofluorescence studies, all subsequent steps were performed in PBS supplemented with 0.1% saponin. Cells were then blocked in buffer containing 10% calf serum and 2% goat serum for 1 h at room temperature, followed by three 5 min washes. Primary antibodies were diluted into blocking buffer and allowed to bind for 1 h at room temperature, followed by three 5-min washes. Goat-derived secondary antibodies were diluted, allowed to bind and ultimately removed in the same manner. Confocal images were acquired with a Zeiss 200 M inverted microscope, 40× NA 1.3 Plan-Neofluar or 100× NA 1.45 Plan-fluar oil immersion objectives (Carl Zeiss AG, Jena, Germany), a CSU-10 confocal spinning-disk head (Yokogawa Electric, Tokyo, Japan). Image capture was performed with an Orca ER CCD camera, an ORCA Flash 4.0 v2 camera (Hamamatsu Photonics, Hamamatsu City, Japan) or an XR MEGA-10 intensified CCD camera (Stanford Photonics, Palo Alto, CA). Fluorescent proteins were resolved as previously described (Bunnell et al., 2006). Image acquisition was managed using Perkin-Elmer Ultraview or μManager software (Edelstein et al., 2010).

Image data was processed using custom AppleScripts (Apple Computer, Cupertino, CA) to manage tasks performed by the open source platforms ImageJ and Fiji (Schneider et al., 2012) or by iVision scripts (iVision, Atlanta, GA). All subsequent image processing and analysis was performed using iVision. All normalizations are linear unless otherwise stated. Dim images were routinely despeckled to reduce the impact of ‘hot’ camera pixels. For Fig. 1D and Fig. S5C, cells were center aligned and averaged; relative intensities are provided as a function of radial distance from the cell center. Because of variations in cell diameter, image regions in Fig. S5A were edge aligned before averaging; relative intensities are presented as a function of distance from the cell edge. All channels in Fig. 1F,G and Fig. S2B were background corrected using a rolling ball filter with a radius of 346.2 nm; in Fig. S2C–E the same filter was applied to the myosin IIA and pMLC channels, but the SLP-76 channel was not modified. The filter employed in Fig. 3G is the iVision Sharpen Hat 5×5 filter. For Fig. S2C–E, myosin IIA images were converted into binary masks by scaling images to 50 nm pixel⁻¹, applying a rolling ball background correction filter, then a Sobel edge filter, and subtracting the edge-filtered image from the background-corrected image. This process was iterated with rolling ball sizes from 100 nm to 350 nm, in 50 nm steps, and the resulting subtracted images were averaged. Using this averaged image, a binary mask was generated by setting thresholds as low as possible without detecting objects outside of cells. The permissive region for void detection was defined as the peripheral region of the contact minus the myosin IIA-masked region. Voids within the permissive mask were detected by sequentially searching for all possible circles with diameters of 600, 500, and finally 400 nm; after each search, the detected objects were removed from the permissive mask. All voids with eccentricities <0.9 were accepted. The outer diameter of the myosin rings surrounding these voids was determined using the point at which the myosin IIA signal falls at 90% of the distance from the peak to the second minimum. For Figs S2D,E and S5B, SLP-76 microclusters were identified using local maxima. Detailed scripts are available from the corresponding author upon request.

TCR-mediated adhesion assays were performed as previously described with the exception that the initial plate-binding time was reduced to 7 min (Nguyen et al., 2008; Ophir et al., 2013). Briefly, Jurkat T cells were loaded with 4 µM BCECF-AM dye from (Invitrogen, B-1170) in Hanks’ balanced salt solution (HBSS) for 30 min at 37°C and then washed three times in HBSS. 5×10⁵ cells were added per well, in triplicate, to control or anti-CD3-coated glass wells (10 µg ml⁻¹) substrates that were blocked with 5% BSA. Cells were allowed to bind for 7 min at 37°C and input BCECF fluorescence was read on a SpectraMax (Molecular Devices, San Jose, CA) plate reader (488 nm excitation, 530 nm emission, 515 nm cutoff). Unbound cells were removed by gentle pipetting and replaced with fresh medium following cycles of no or increasing shear, applied for 30 s using a Vortex-Genie2 plate shaker (Fisher Scientific) at the indicated speeds. Fluorescence readings was taken after each cycle of replacement.

Jurkat T cells were stimulated in 24-well tissue culture plates maintained in a humidified tissue culture incubator with 5% CO₂. Each well received 5×10⁵ cells in a final volume of 500 µl. Prior to the addition of cells, wells were left untreated or coated with anti-CD3 (10 µg ml⁻¹) in PBS for 1 h at 37°C. After 24 h, supernatants were harvested and the upregulation of CD69 was measured by flow cytometry. IL-2 secretion into the supernatant was measured using the Duo Set IL-2 ELISA kit (R&D Systems, DY202-05).

Surface-activated magnetic beads 4.5 µm in diameter (M-450 tosylactivated dynabeads, #14013) were obtained from Invitrogen and functionalized with anti-CD3 (10 µg ml⁻¹), anti-CD43 (10 µg ml⁻¹), and/or recombinant human VCAM-1 (1 µg ml⁻¹). Beads were then blocked in 1% BSA and residual active sites were capped via the addition of 20 mM Tris-HCl (pH 7.4). Stimulations were initiated by mixing beads and cells at 2:1 ratio, with 5×10⁷ cells ml⁻¹. After a 5 min co-incubation at 37°C, cells were fixed with 0.5 mM dithiobismaleimidoethane (DTME, Thermo Fisher Scientific #22335) for 15 min. Fixed cells were disrupted by sonication in detergent-free lysis buffer, beads were captured magnetically, and washed extensively. Magnetic isolates corresponding to 2×10⁷ cell equivalents of J14.SY cells were analyzed by Coomassie staining and individual bands were excised for in gel reduction, alkylation and trypsinization. Liquid chromatography tandem mass spectrometry (LC/MS/MS) analyses were performed using a Thermo LTQ ion trap mass spectrometer. MS/MS spectra were searched against the NCBI non-redundant protein sequence database using the SEQUEST computer algorithm.

Flow cytometry data was analyzed using FlowJo software. Statistical analyses were performed using Microsoft Excel and Prism. Based on previous studies, perturbations of SLP-76 microclusters typically yield normalized effect sizes ≥2. Under these conditions n=3 is sufficient to incorrectly reject the null hypothesis <5% of the time, with a power of 90%. Comparable effect sizes were observed here. All stable lines of lentiviral origin were derived at least twice. Cell scoring was conducted using iVision scripts that present de-identified images for analysis. For the analysis of clustering in each cell line, each category was analyzed by one-way ANOVA; post-hoc Bonferroni corrected comparisons versus the parental line are only shown if categories pass this initial test. CD69 and IL-2 assays were analyzed by two-way ANOVA; stimulation condition and cell-type yielded significant differences; post-hoc Bonferroni corrected comparisons within the stimulation group and versus the parental line are shown. Adhesion assays were analyzed by two-way ANOVA for shear and cell type using the Geisser–Greenhouse correction for matched samples; no differences were observed as a function of cell type. Two-tailed Student's t-tests for unpaired samples were used to compare the radial distributions of SLP-76 in the presence or absence of myosin IIA and/or IIB. Unless noted otherwise, *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001.

We thank M. A. Fray for visualizing EGFP-tagged forms of myosin relative to SLP-76 and R. R. Brewka for her contributions to establishing the role of myosin II in microcluster movement. We recognize Jon DeGnorre and the Tufts University Core Facility for mass spectrometry and the W. M. Keck Foundation and the Eshe Fund for generous support of core facilities at Tufts.