Modulation of T cell signaling by the actin cytoskeleton

T cells have a central role in adaptive immunity, and their activation involves many spatially and temporally coordinated signaling processes on multiple time and length scales. There are three distinct stages during the activation process: T cell receptor (TCR) triggering, signal persistence and signal termination. A crawling T cell constantly scans the surfaces of antigen-presenting cells (APCs) in search of antigen peptides that are bound to major histocompatibility complex proteins (pMHCs). Upon recognition of such peptides, ligation of integrin molecules facilitates the initial close cell–cell contact, and engagement between TCRs and agonist pMHCs subsequently triggers the initial TCR signaling across the plasma membrane. TCRs and their co-receptors, other accessory proteins, the microtubule organization center (MTOC) and secretory organelles all reorient to a position near the cell–cell contact site. At the same time, the T cell membrane protrudes over the APC surface, leading to the formation of a specialized membrane junction, namely the immunological synapse (Fig. 1A). Here, we focus our discussion on the canonical pattern of concentric protein domains, but it is important to note that a variety of protein patterns exist for different subsets of T cells as well as in other immune cells (Davis and Dustin, 2004). A hallmark of TCR triggering is the rapid increase in the concentration of intracellular calcium ions (Lewis, 2001). Following TCR triggering, signaling is amplified and sustained in microclusters (Seminario and Bunnell, 2008; Choudhuri and Dustin, 2010; Dustin and Groves, 2012). TCRs in the cell periphery rapidly assemble into microclusters consisting of tens to hundreds of molecules. Concurrently, other signaling factors (Box 1) are recruited to the macromolecular signaling complex to propagate downstream signaling events, including kinases such as ζ-chain-associated tyrosine-protein kinase 70 (ZAP70) (Wang et al., 2010; Sherman et al., 2011) and the lymphocyte-specific tyrosine-protein kinase Lck (Nika et al., 2010), adaptor proteins such as the linker for activation of T cells family member 1 (Lat) (Lillemeier et al., 2010) and the SH2-domain-containing leukocyte protein of 76 kDa [SLP-76 (also known as lymphocyte cytosolic protein 2, LCP2)] (Barda-Saad et al., 2010), as well as actin-polymerizing factors such as Wiskott–Aldrich syndrome protein (WASp) (Gomez and Billadeau, 2008; Reicher and Barda-Saad, 2010). These signaling microclusters undergo continuous translocation and become spatially organized in the synapse to form a ‘bull's eye’ pattern of several distinct concentric domains, which are known as supramolecular activation clusters (SMACs; Fig. 1B) (Monks et al., 1998; Grakoui et al., 1999). TCR microclusters are transported such that they accumulate in the central SMAC (cSMAC). Distinct from the newly formed microclusters in the cell periphery, these centralized microclusters are mostly associated with receptor internalization and signaling degradation (Mossman et al., 2005; Varma et al., 2006), thus making the term ‘supramolecular activation cluster’ a misnomer. Integrins, such as lymphocyte function-associated antigen 1 (LFA-1), together with many cytoskeletal linker proteins, reorganize to form a ring structure surrounding the cSMAC, forming the peripheral SMAC (pSMAC). The distal SMAC (dSMAC) is enriched with proteins with large extracellular domains, including the protein-tyrosine phosphatase CD45 (also known as receptor-type tyrosine-protein phosphatase C). All signaling events must be coordinated in time and space to achieve accurate T cell activation, and each of these activities is dependent on the actin cytoskeleton.

The actin cytoskeleton is a filamentous network known to provide mechanical forces for cell polarization and motility, to scaffold proteins for binding and assembly and also to transport molecules. Its importance in T cell activation has long been established, ever since the first studies of the actin cytoskeleton in T cell activation (Geiger et al., 1982; Ryser et al., 1982). However, the nature of the mechanisms linking actin with T cell activation has only recently begun to be unveiled (Fig. 2). Actin drives the process of cell polarization and maintains the cell–cell contact – the very first steps for T cell activation; it also likely provides a scaffold for clustering, translocation and spatial segregation of proteins, key steps to amplify and sustain T cell signaling (Barda-Saad et al., 2005; Billadeau et al., 2007; Dustin, 2007; Gomez and Billadeau, 2008; Beemiller and Krummel, 2010). It has been proposed that mechanical forces from actin might be required in TCR triggering, although the exact mechanism underlying this is still debated (van der Merwe and Dushek, 2011). In addition, studies also suggest that T cell activation requires actin retrograde flow, instead of static filaments (Kaizuka et al., 2007; Yu et al., 2010; Babich et al., 2012). An emerging picture is that the actin cytoskeleton provides a dynamic framework for the spatial and temporal regulation of both mechanical and biochemical signaling pathways in T cell activation. Here, we discuss, with an emphasis on the mechanistic aspects, how the actin cytoskeleton regulates T cell activation during TCR triggering, assembly and translocation of signaling proteins and the formation of the immunological synapse.

TCR triggering is the process by which TCR–pMHC ligation leads to intracellular signaling. It is not only the initial step of T cell activation but is also extremely sensitive and specific – signaling can be triggered by extremely low densities of activating agonist peptides in the presence of high levels of self-pMHCs (stochastic noise); T cells must also differentiate agonists of various potency. The TCR complex consists of a TCR heterodimer (TCRαβ) that is responsible for ligand recognition and multiple CD3 molecules (CD3γ, CD3δ, CD3ε and CD3ζ) that mediate all proximal signaling events. TCR engagement with the antigen peptide–MHC protein complexes triggers tyrosine phosphorylation of the immunoreceptor tyrosine-based activation motifs (ITAMs), which are present in the cytoplasmic domains of all CD3 subunits in the TCR complex (van der Merwe and Dushek, 2011). Three main models have been proposed so far to understand how TCR engagement leads to biochemical signaling across the plasma membrane: conformational change, kinetic segregation and receptor aggregation. We discuss below the possible roles of actin in each mechanism.

The conformational change model explains T cell triggering at the level of the single TCR. The central hypothesis of this model is that TCR–pMHC binding alone, without contribution from other forces, exerts a pulling force on the TCR, leading to conformational changes in the CD3 subunits and thus downstream biochemical signaling (Sun et al., 2001; van der Merwe, 2001; Beddoe et al., 2009). As a result, the proline-rich sequence in the CD3ε cytoplasmic tail becomes accessible to interact with a cytoskeletal adaptor protein, the non-catalytic region of the tyrosine kinase cytoplasmic protein Nck1 (Gil et al., 2002). These observations serve as indirect evidence to support the involvement of actin in the conformational change model. A more significant role for actin was proposed in a modified version, called the receptor deformation model (Ma et al., 2008; Ma and Finkel, 2010). In this model, TCR–pMHC binding itself does not initiate TCR signaling. Instead, as the T cell moves in an actin-dependent manner with respect to its attached APC, it generates tension across its membrane. The effect of this tension is a pulling force that is transmitted to the TCR, thereby inducing a conformational change in the TCR–CD3 complex and making the cytoplasmic ITAMs more accessible to the tyrosine-protein kinase Lck.

As the structure of an intact TCR–CD3 complex is still unsolved, there have been limited structural data to prove the conformational changes directly. But considerable evidence pinpoints mechanical effects from the actin cytoskeleton as an important factor in TCR triggering. First of all, actin might be essential to ensure the specificity and sensitivity of T cell triggering, by enabling the rapid association–dissociation of TCR–pMHC binding. The pulling force from actin proposed in the receptor deformation model is expected to cause an increase in the dissociation of TCR–pMHC, and this has been demonstrated experimentally in recent studies. Huppa and colleagues used single-molecule fluorescence resonance energy transfer (FRET; Box 2) to measure the TCR–pMHC binding kinetics between TCRs in the T cell membrane and pMHCs on glass-supported lipid bilayers (Box 2) (Huppa et al., 2010). The dissociation rate (k_off) at the membrane junction (two-dimensional) was faster than that obtained from measurements in solution (three-dimensional) – a key feature that is required for T cells to sample rapidly a small number of cognate pMHCs among a high level self-pMHCs on an APC. Interestingly, k_off decreased significantly when actin filaments were depolymerized. A similar effect of the actin cytoskeleton on TCR–pMHC binding was also independently observed by Huang and colleagues, who used a micropipette and a biomembrane force probe to measure the binding kinetics between a T cell and an antigen-presenting red blood cell (Huang et al., 2010). Both studies demonstrate that actin is crucial to ensure rapid TCR–pMHC binding and possibly for the high specificity and sensitivity of T cell activation. Second, the model postulates that the mechanical force generated by actin must be transmitted to the TCR–CD3 complex by means of TCRs. This is structurally plausible as the part of the TCRβ chain is in close proximity to the relatively rigid CD3ε ectodomain (Ghendler et al., 1998; Kim et al., 2010). More importantly, mechanical forces applied on the TCR heterodimer can trigger T cell signaling (Kim et al., 2009; Li et al., 2010; Husson et al., 2011). Additional evidence to support the necessity of actin-generated force for TCR triggering is the observation that monomeric agonist pMHC can trigger T cell signaling when anchored on the cell surface, but the same agonist in solution cannot.

The conformational change model is particularly appealing to explain the possible role of actin in the high sensitivity and specificity of T cell triggering. By regulating the TCR–pMHC binding strength, actin enables T cells to detect signals rapidly at extremely low densities and differentiate agonists of different potency. This mechanism is supported by compelling evidence, but thus far no structural data have been obtained to demonstrate structural changes directly in the intact TCR–CD3 complex upon the application of physical forces or upon binding of agonist pMHC. The conformational change concept is also challenged by some recent studies (Fernandes et al., 2012; James and Vale, 2012) and, furthermore, it is unclear how the pulling force is transmitted from the TCR to the CD3 cytoplasmic domains and whether actin has any role in that step.

In addition to its role in directly influencing single TCR–pMHC complexes, actin might also have a role in TCR triggering at the level of TCR clusters. The kinetic segregation model and receptor aggregation model (discussed below) are better suited to describe the possible underlying mechanisms. Proposed in 1996, the kinetic segregation model postulates that TCR triggering is caused by size-dependent segregation of kinases from phosphatases present at the membrane junction between T cell and APC (Davis and van der Merwe, 1996; Davis and van der Merwe, 2006). Membrane receptors with small extracellular domains, including the TCR–pMHC pair and kinases, accumulate in many small ‘close-contact zones’ (∼15 nm apart), from which large glycoproteins such as the inhibitory tyrosine phosphatase CD45 are excluded owing to steric effects. In resting T cells, the ITAMs have a low phosphorylation level because they are constantly phosphorylated by tyrosine kinase Lck and dephosphorylated by phosphatase CD45. However, the size-dependent exclusion of CD45 shifts the kinase–phosphatase balance, leading to prolonged phosphorylation of the ITAMs and thus TCR triggering (Choudhuri et al., 2009). This mechanism is supported by mounting evidence. For example, exclusion of phosphatases from areas of TCR triggering has been confirmed experimentally (Leupin et al., 2000; Lin and Weiss, 2003; Varma et al., 2006). Moreover, T cell signaling is reduced when protein segregation is disrupted by truncation of the extracellular domains of CD45 and CD148 (Irles et al., 2003; Lin and Weiss, 2003) or elongation of the ectodomain of agonist pMHCs (Choudhuri et al., 2005; Choudhuri et al., 2009).

Actin probably has important roles in many aspects of the kinetic segregation model. First of all, actin is required for forming and maintaining the ‘close-contact zones’ (Valitutti et al., 1995a; Bunnell et al., 2001; Cannon and Burkhardt, 2002</emph>), the very first step of TCR signaling. Second, actin and actin-based molecular motors are coupled to the plasma membrane and regulate membrane tension (Nambiar et al., 2009; Skruzny et al., 2012), a potential facilitator of protein segregation in the ‘close-contact zones’ during TCR triggering. A recent study by James and Vale suggests that protein segregation is not solely driven by the extracellular sizes of cell-surface molecules but also by a certain membrane force (James and Vale, 2012). Intriguingly, by depolymerizing actin with pharmacological agents, they observed a negligible role for actin in protein segregation in the small population of cells that did form cell–cell contacts out of the large population of cells that did not. The observation is surprising and might be attributable to the heterogeneity of cell response to pharmacological treatment. Finally, as the kinetic segregation model is also applicable to the triggering of other membrane receptors such as the co-stimulatory molecules (Davis and van der Merwe, 2006), actin might further modulate TCR triggering by transducing co-stimulatory signals to lower the threshold number of agonist pMHCs required for activation (Bachmann et al., 1997; Suzuki et al., 2007).

Actin could also modulate TCR triggering at the cluster level by promoting formation of lipid rafts. Lipid rafts are defined as nanometer-sized membrane domains that are enriched in cholesterol, sphingolipids and signaling proteins (Lingwood and Simons, 2010). In this model, it is hypothesized that the lipid rafts, formed upon the TCR–pMHC interactions, trigger TCR signaling by facilitating the association of the TCR–CD3 complexes with tyrosine kinases while excluding inhibitory phosphatases such as CD45 (Harder et al., 2007; Filipp et al., 2012). The actin cytoskeleton might be crucial in transporting TCRs into the kinase-rich lipid rafts and maintaining the lipid raft environment. Evidence to support the role of actin in this mechanism includes studies in which disruption of the actin cytoskeleton leads to disappearance of lipid rafts and CD45 in the immunological synapse (Valensin et al., 2002; Chichili et al., 2010; Chichili et al., 2012). However, a recent study suggests that formation of lipid rafts is not sufficient for Lck clustering (Rossy et al., 2013). It should also be noted that the existence of lipid rafts is controversial, mainly because many of the experimental approaches used were found to be questionable and some groups were unable to observe these small and dynamic membrane domains directly (Shaw, 2006; Leslie, 2011).

The receptor aggregation model simply states that binding between TCRs and pMHCs leads to aggregation of TCR–CD3 complexes and thus allows for enhanced local TCR signaling. The model stemmed from the longstanding observations that forced aggregation of TCRs, using antibodies or soluble multimeric forms of agonist pMHCs, can trigger T cell activation. In addition to the role of actin in maintaining cell–cell contact, actin might also contribute to TCR–CD3 aggregation by acting as a scaffold to stabilize and support protein aggregation – which will be discussed further in the next section. However, the receptor aggregation model has many limitations. One major problem is that the model cannot explain why T cell signaling can be triggered by extremely low densities of pMHCs, at which TCR microclusters are small, sparse or unobservable (Sykulev et al., 1996; Irvine et al., 2002; Manz and Groves, 2010). As different modifications have been made to the original model, the TCR aggregation model looks more plausible when other mechanisms such as conformational change (Reich et al., 1997; Alam et al., 1999) and TCR serial triggering (Valitutti et al., 1995b) are incorporated.

Taken together, the actin cytoskeleton is likely to be indispensable for the generation of signals from TCRs regardless of the models proposed. All three models – conformational change, kinetic segregation and receptor aggregation – are not mutually exclusive. They describe the TCR triggering process on different length scales: the conformation change model focuses on the level of single TCRs, whereas the kinetic segregation and receptor aggregation models concern the molecular cluster level. In fact, many experimental observations support more than one model. Therefore, it is possible that actin has multiple roles in a combination of these mechanisms at various stages in TCR triggering.

After TCR triggering, TCRs and other receptors as well as intracellular proteins cluster into distinct domains, creating large-scale spatial organization in the immunological synapse. The entire process of TCR clustering, translocation and large-scale segregation is correlated with the T cell biochemical signaling and thus provides a direct readout of T cell activation in time and space.

TCRs, together with other signaling molecules, form microclusters following their ligation with the agonist pMHCs. Although TCR microclusters that accumulate in the cSMAC were originally believed to be the site for T cell signaling, it has become clear that, under normal triggering conditions, the newly formed microclusters in the periphery of the cell contact area are the active signaling units (Varma et al., 2006; Cemerski et al., 2008), also referred to as ‘signalosomes’ (Werlen and Palmer, 2002). Actin is required for signaling at the microcluster level because either disruption or over-stabilization of actin abrogates the formation of TCR microclusters, Ca²⁺ signaling (Campi et al., 2005; Varma et al., 2006) and tyrosine phosphorylation (Shen et al., 2005). One important question that many recent studies aimed to answer is the exact nature of the contribution of actin to T cell signaling within the microclusters.

Actin might regulate T cell signaling by maintaining the association between TCRs and other signaling proteins. That explains the observation that, upon actin depolymerization by pharmacological drugs, the concentration of intracellular Ca²⁺ is reduced within one minute and no new TCR clusters form, but existing TCR microclusters remain intact for at least a few further minutes (Varma et al., 2006). Interestingly, if T cells are treated with jasplakinolide, a pharmacological drug that interrupts actin depolymerization, Ca²⁺ signaling also stops, but phosphorylation of ZAP70 in the TCR microclusters remains (Babich et al., 2012). This observation suggests that the integrity of signaling clusters only requires the scaffolding function of actin, but keeping the dynamic association of molecules inside microclusters for sustained signaling requires actin retrograde flow. Indeed, the scaffolding effects appear to be reciprocal as one of our recent studies has suggested that microclusters possess the capacity to organize the actin cytoskeleton locally around themselves (Smoligovets et al., 2012). The scaffolding properties of actin might be mediated by its regulatory proteins (Fig. 3). The guanine-nucleotide-exchange factor (GEF) Vav1 and Wiskott-Aldrich syndrome protein WASp, both key regulators for F-actin nucleation and remodeling, have been observed to be recruited to TCR–pMHC binding sites (Sasahara et al., 2002; Zeng et al., 2003; Barda-Saad et al., 2005; Tybulewicz, 2005; Yokosuka et al., 2005; Miletic et al., 2006; Miletic et al., 2009). It has been suggested that Vav1 associates with the Lat–SLP-76 adaptor complex (Sylvain et al., 2011; Pauker et al., 2012) and that WASp binds to the proline-rich sequences in CD3 chains (Gil et al., 2002). Moreover, Vav1 was shown to stabilize signaling microclusters during T cell activation independently of its intrinsic GEF activity (Miletic et al., 2009; Sylvain et al., 2011), suggesting that it functions as a linker protein to load microclusters onto the actin cytoskeleton.

Actin might also modulate T cell signaling by stabilizing preformed protein clusters in quiescent T cells. With the help of high-resolution imaging techniques (Box 2), Lillemeier and colleagues revealed that TCR and the adaptor protein Lat exist in separate membrane domains with a diameter of 70–140 nm (Lillemeier et al., 2010), whereas Sherman and co-workers reported pre-existing Lat nanoclusters as small as dimers in resting T cells (Sherman et al., 2011). The difference could in part be due to the differences in sample preparation, but both studies confirmed that actin is required for nanocluster stability. As protein preclustering has been proposed as a regulatory ‘safety-check’ mechanism in T cell activation (Lillemeier et al., 2010; Chung et al., 2012), the prescaffolding effect exerted by actin probably has an important role. However, the idea that pre-existing protein clusters contribute to T cell activation was recently challenged by the finding that Lat molecules participate in early T cell signaling by means of subsynaptic vesicles instead of preclusters (Williamson et al., 2011).

In addition, as discussed in the context of the kinetic segregation and receptor aggregation models, actin might also modulate TCR microcluster formation and signaling by controlling the mechanical properties of the cell membrane or promoting membrane compartmentalization. Besides the aforementioned lipid raft hypothesis, an alternative mechanism for the role of actin in the assembly of microclusters is the so-called ‘picket fence model of diffusion restriction’ (Kusumi et al., 2005). This model posits that the high local density of actin and actin-associated membrane proteins restricts the diffusion of other membrane-associated or membrane-proximal cytoplasmic proteins and thus promotes assembly of microclusters. Although this model is supported by some data (Andrews et al., 2008; Treanor et al., 2010; Treanor et al., 2011), it does not explain why proteins can rapidly coalesce or dissociate from the clusters or how TCR microclusters are transported from the cell periphery to the cSMAC.

Within a few minutes of cluster formation, TCR microclusters are centripetally translocated to coalesce into cSMACs, and ligated integrins and associated proteins accumulate around the periphery of the synapse. It has long been recognized that receptor transport and spatial organization in the immunological synapse require retrograde flow of actin (Dustin and Cooper, 2000; Burkhardt et al., 2008; Gomez and Billadeau, 2008; Beemiller and Krummel, 2010). To reveal the physical interactions between TCR clusters and actin, Kaizuka and colleagues showed that TCR microclusters and integrins move centripetally together with the actin cytoskeleton by using imaging of fluorescently labeled protein clusters in Jurkat cells that are triggered by a glass-supported lipid bilayer containing stimulatory ligands (Kaizuka et al., 2007). Based on the observation that TCR microclusters move slower than the underlying retrograde actin flow, they proposed that a ‘stick–slip’ movement of TCR microclusters occurs on the actin cytoskeleton (Kaizuka et al., 2007). However, considering that each TCR microcluster contains tens to hundreds of molecules, simultaneous breakage of all the molecular linkages to actin in the ‘slip’ phase seems unlikely. The Groves group later proposed a frictional coupling mechanism (Fig. 4), based on a study that employed physical barriers on glass-supported lipid bilayers (DeMond et al., 2008). When the centripetal transport of TCR microclusters was hindered by a physical barrier, the microclusters did not stop but continued their motion with a different direction (DeMond et al., 2008). This suggests that the coupling between microclusters and actin is rather transient – multiple weak links form and break at different times to generate a frictional force between microclusters and actin. Two subsequent studies provided further support for the idea of the frictional coupling model and showed that actin itself slows down as it passes over mechanically trapped TCRs (Yu et al., 2010; Smoligovets et al., 2012).

The transient coupling mechanism predicts that the strength of attachment to the actin cytoskeleton correlates positively with the number of molecular linkages. This could explain why TCR and LFA-1 microclusters, although both coupled with actin, are transported differentially to destinations that are micrometers apart. Indeed, it has been found that, when the LFA-1 clusters are crosslinked by using bivalent and tetravalent antibodies, larger clusters that presumably have more linkages to actin are transported closer to the center of the immunological synapse and that tetravalently crosslinked LFA-1 clusters are able to reach the cSMAC (Fig. 4B) (Hartman et al., 2009). This observation cannot be explained by the size-exclusion mechanism in the kinetic segregation model. Instead, it suggests that spatial sorting of proteins in the immunological synapse can be driven by their specific coupling interactions to the actin cytoskeleton.

Formation of the immunological synapse is tightly associated with signaling activities: signaling is sustained during microcluster formation (Bunnell et al., 2002; Lee et al., 2003; Yokosuka et al., 2005) and attenuated (Varma et al., 2006) or amplified under conditions of weak agonist (Cemerski et al., 2008) in cSMACs. Therefore, actin-mediated protein cluster formation, translocation and spatial sorting might provide a dynamic framework for T cells to regulate signaling activities in time and space.

Compelling evidence indicates that the retrograde flow of actin is an essential part of the formation of immunological synapses and of T cell activation, and multiple processes might generate and control this flow. Actin retrograde motion is likely to be driven by the polymerization of its filaments (Henson et al., 1999). Additionally, the actin-associated molecular motor myosin II has been found to contribute to actin flow by exerting contractile forces on actin filaments (Vicente-Manzanares et al., 2009; Betapudi, 2010; Arii et al., 2010; Bhuwania et al., 2012). Non-muscle myosin IIA is the dominantly expressed isoform of myosin II in T cells (Jacobelli et al., 2004). The importance of non-muscle myosin IIA in T cell polarity and migration is well established (Smith et al., 2003; Sanchez-Madrid and Serrador, 2009; Jacobelli et al., 2009; Jacobelli et al., 2010), but its role in formation of the immunological synapse and T cell activation is still controversial. Studies from some groups have shown that myosin IIA is necessary for TCR microcluster movement and immunological synapse organization (Wülfing and Davis, 1998; Ilani et al., 2009; Yu et al., 2012; Yi et al., 2012; Kumari et al., 2012), whereas other groups have found that neither TCR cluster transport nor Ca²⁺ signaling is dependent on myosin IIA function (Jacobelli et al., 2004; Babich et al., 2012; Beemiller et al., 2012). It has been found that myosin drives a rapid inward translocation of TCR microclusters only during the early stage of signaling, whereas actin polymerization provides a slower basal rate of motion that persists throughout the entire life span of an immunological synapse (Yu et al., 2012). Consequently, actin retrograde flow alone is able to organize the microclusters spatially in response to myosin inhibition, albeit at a slower rate than in the presence of uninhibited myosin IIA (Yu et al., 2012). In agreement with these observations, Yi and colleagues reported that contraction of myosin II and actin retrograde flow together transport microclusters, but that myosin II is not required for formation of immunological synapses. In their study, the specific locations of myosin II and the actin filaments were also nicely examined. By contrast, others have recently reported that actin retrograde flow is driven solely by actin polymerization that pushes against the edge of the cell membrane (Babich et al., 2012; Beemiller et al., 2012). Interestingly, Babich and colleagues also observed that, although myosin IIA is not required, actin retrograde flow is not the only driving force for microcluster centralization (Babich et al., 2012).

The discrepancies arising from all these studies could be due to a combination of factors. First, different T cell types were used: primary murine cells (Jacobelli et al., 2004; Yu et al., 2012; Beemiller et al., 2012), human Jurkat cells (Ilani et al., 2009; Babich et al., 2012; Yi et al., 2012) or primary human cells (Ilani et al., 2009; Babich et al., 2012). Although no detailed comparison across different cell types is available, there might be some heterogeneity with regards to the expression and requirement of motor proteins, or cell responses to drug treatment. Second, the discrepancy might be due to the use of different stimulatory surfaces, such as surfaces with immobilized anti-CD3 versus supported lipid bilayers with agonist pMHCs, as T cell activation is known to be sensitive to the type, concentration and mobility of agonists (van der Merwe and Dushek, 2011; Hsu et al., 2012). Third, different actin labeling methods might restrict quantification of actin in certain locations (i.e. pSMAC versus dSMAC) or even interfere with actin polymerization (Burkel et al., 2007; Riedl et al., 2008; Milroy et al., 2012).

Since the first study of the role of actin in T cell activation more than three decades ago, tremendous progress has been made towards identifying individual factors that link the complex signaling pathways and cytoskeletal elements of T cells through the use of traditional biochemical studies (Burkhardt et al., 2008; Reicher and Barda-Saad, 2010). Many recent breakthroughs, such as those that relate actin dynamics to TCR activation in real-time, were made possible by employing interdisciplinary approaches such as high-resolution imaging techniques. It has now been established that the importance of the actin cytoskeleton in T cell activation transcends the maintenance of cell morphology and transport of proteins. In fact, actin regulates T cell activation by maintaining cell–cell contact, facilitating TCR triggering and scaffolding protein assemblies and organization. In spite of our progress towards understanding of the complex role of actin in T cell signaling, many important questions remain. For example, an emerging idea is that the physical forces from actin and actin-associated proteins also modulate T cell activation, but we do not know the exact mechanisms or proteins involved that transduce mechanical forces to biochemical signaling pathways. The distinct filamentous structures of actin and the spatial distribution of molecular motors in T cells need to be distinguished with high resolution, possibly by using transmission electron microscopy (TEM) or super-resolution imaging techniques. Another key focus of research is to understand the role of actin in temporally coordinating different signaling pathways. Additionally, the question of actin integration in T cell co-stimulation remains to be answered. New regulatory proteins will doubtless be identified and novel strategies must be developed to dissect the spatial and temporal connectivity between signaling pathways in order to answer all these questions – only then will a more comprehensive understanding of the role of actin in T cell functions be achieved.