TAGLN2 regulates T cell activation by stabilizing the actin cytoskeleton at the immunological synapse

Activated T cells play a central role in adaptive immunity through cytokine secretion or the destruction of antigen-bearing cells. T cell activation and function require physical contact with antigen-presenting cells (APCs) at a specialized junction structure known as the immunological synapse (IS; Monks et al., 1998; Grakoui et al., 1999). Once formed, signal-dependent rearrangements in actin cytoskeleton dynamics are absolutely necessary to sustain correct temporal and spatial control of the activation process (Huang and Burkhardt, 2007; Yu et al., 2013). Alteration of actin dynamics at the IS results in immune dysfunction.

Our understanding of actin cytoskeleton reorganization during T cell activation has advanced rapidly in the last decade by adding new players and gaining mechanistic insight into T cell biology. The actin-related protein 2/3 (Arp2/3) complex is a major regulator of actin dynamics in T cells, as in other cell types (Higgs and Pollard, 2001). Of the Arp2/3-activating factors, WASp (Wiskott–Aldrich syndrome protein), WIP (WASP-interacting protein), WAVE (WASP-family verprolin-homologous protein), and HS1 (hematopoietic lineage cell-specific protein 1) are well studied in T cells, and depletion of these proteins individually results in poor actin polymerization and weakens the ability of T cells to form the IS with APCs (Huang and Burkhardt, 2007). Formins are a class of actin nucleators that are independent of Arp2/3 (Pruyne et al., 2002). Two formins, mDia1 and FMNL-1 (formin like-1), are necessary for microtubule-organizing center translocation toward the IS, whereas their depletion has no effect on the development of normal lamellipodium and IS (Gomez et al., 2007). Additionally, several negative regulators of F-actin have been described in T cells (Ichetovkin et al., 2002; Eibert et al., 2004). Cofilin is well characterized and functions by severing actin filaments and sequestering actin monomers. Depletion of cofilin results in the formation of increased and longer lamellipodial structures in T cells (Kim et al., 2014).

The structure of the actin network is also regulated through the actions of a variety of actin cross-linking or bundling proteins. An actin cross-linking protein, α-actinin, regulates IS formation (Gordón-Alonso et al., 2012). l-Plastin is involved in efficient spreading of T cells on surfaces with immobilized T cell receptor (TCR) ligands (Wang et al., 2010). Filamin A interacts with CD28 at the IS (Tavano et al., 2006). Overall, the studies on each of these proteins suggest that different actin regulators participate in different aspects of T cell activation and only a fully rearranged actin cytoskeleton is optimal for full T cell activation.

TAGLN (transgelin) family members have been identified as actin cross-linking/gelling proteins. The family comprises three isoforms, namely TAGLN1 (also known as transgelin1 or SM22α, highly expressed in smooth muscle cells; Camoretti-Mercado et al., 1998), TAGLN2 (also known as transgelin2 or SM22β, associated with various types of cancers; Zhang et al., 2002, 2010; Rho et al., 2009), and TAGLN3 (also known as transgelin3 or NP25, abundant in brain tissues; Mori et al., 2004). The term “transgelin” was originally derived from the transformation-sensitive and rapid actin-gelling properties of these proteins (Shapland et al., 1993). In this study, we found that TAGLN2 had a modest effect on actin cross-linking or bundling, whereas it substantially blocked depolymerization of and competed with cofilin to bind F-actin. These findings suggest that TAGLN2 has a unique role in actin reorganization, in addition to its previously reported function. Indeed, there is little information available regarding the role of actin-stabilizing proteins in IS formation and subsequent T cell activation.

Our study shows that TAGLN2 is the only transgelin isoform present in leukocytes, and it is segregated at the distal supramolecular activation cluster (SMAC; d-SMAC) within the IS after TCR stimulation. These results prompted us to explore the possibility of TAGLN2 as a potential regulator of IS formation and subsequent T cell activation. Through biochemical and genetic analyses, we uncovered evidence that TAGLN2 stabilizes the actin cytoskeleton, leading to an increase in F-actin content at the IS and subsequent activation of integrin function. TAGLN2^−/− T cells are unable to sustain a conventional F-actin ring, thus decreasing cell adhesion and spreading after TCR stimulation. As a result, cytokine expression and effector function are impaired.

The calponin homology (CH) actin-binding domain is found in cytoskeletal and signal transduction proteins, including TAGLNs. A phylogenetic tree of the family of single CH domain-containing proteins is shown in Fig. S1 A. The TAGLN family, including TAGLN2, is evolutionarily related to Vav1 and IQGAP1, which are key cytoskeletal-regulatory proteins involved in T cell immunity.

The expressions of TAGLN1 and TAGLN3 are restricted to the smooth muscle (Camoretti-Mercado et al., 1998) and brain (Mori et al., 2004), respectively. However, the expression pattern of TAGLN2 has not yet been fully characterized. Thus, we first tested for the presence of TAGLN2 in various human tissues by Northern blot analysis and observed high expression in the spleen (Fig. S1 B). Western blot and real-time quantitative PCR (qPCR) analyses of mouse tissues confirmed that TAGLN2 was most highly expressed in the thymus, spleen, and lymph nodes (Fig. S1 C). TAGLN2 expression was also detected in primary immune cells or immune cell lines, and we did not detect the presence of other TAGLNs in these cells (Fig. S1 D). The predominant expression of TAGLN2 in T cells prompted us to explore its role in cytoskeletal regulation, which is essential for T cell function. A dramatic reduction in IL-2 mRNA and IL-2 protein levels was observed in human peripheral blood lymphocytes (PBLs) transfected with TAGLN2 siRNA after stimulation with Staphylococcal aureus enterotoxin E (SEE)–pulsed Raji B cells (Fig. S1 E), suggesting a potential role for TAGLN2 in T cell immunity.

Just like TAGLN1 and TAGLN3 (Camoretti-Mercado et al., 1998; Mori et al., 2004), TAGLN2 colocalized with actin stress fibers and also localized to the lamellipodial leading edge, where actin polymerization actively occurs in CHO-K1 cells (Fig. 1 A). Coimmunoprecipitation of actin with GFP-tagged TAGLN2 (TG2_GFP) before and after activation revealed that TAGLN2 binding to actin is stimulus independent (Fig. 1 B). These findings led us to analyze endogenous TAGLN2 localization in primary T cells. TAGLN2 was localized to the F-actin ring of mouse CD3⁺ T cells loaded onto anti-CD3/28–coated coverslips (Fig. 1 C). Ectopic overexpression of TG2_GFP in Jurkat T cells (J-TG2_GFP) revealed TG2_GFP expression not only in the cytoplasm but also in the membrane compartment (Fig. 1 D). Upon J-TG2_GFP cell conjugation with SEE-loaded Raji B cells, TG2_GFP localized to the outer border of ICAM-1, which is a representative peripheral SMAC molecule present on B cells (Fig. 1 E and Video 1). To examine the localization of TAGLN2 in greater detail, J-TG2_GFP cells were loaded on anti-CD3/28–coated coverslips, and then, the cells were fixed and stained with F-actin, LFA-1, Arp3, and myosin IIA. TAGLN2 was most clearly seen at the outer actin ring (d-SMAC; Fig. 1 F, d), although it was also observed in the nucleus (Fig. 1 F, Fig. S2, and Video 2). TG2_GFP signals strongly overlapped with Arp3 at the outer ring, but not with LFA-1 and myosin IIA, which are representative of the middle ring (peripheral SMAC; Fig. 1 F, p; Fig. S2; and Videos 3 and 4; Yi et al., 2012). Collectively, we concluded that TAGLN2 associates with F-actin and is a d-SMAC protein in the IS.

We next investigated the function of TAGLN2 in actin reorganization in Jurkat T cells. A kinetic study revealed that, although initial cell spreading and F-actin ring formation were similar, TAGLN2 overexpression prolonged the duration of T cell spreading and F-actin ring formation on anti-CD3/28–coated coverslips (Fig. 2, A and B; and Video 5 and 6). This was correlated with the up-regulation of total F-actin content in J-TG2_GFP cells after stimulation (Fig. 2 C). Accumulation of F-actin at the IS was also evident during T cell–APC conjugation. TG2_GFP was clearly overlapped with LifeAct (mCherry; Fig. 2 D, left), and the F-actin retention time at the contact region in J-TG2_GFP cells was longer than that of J-GFP cells (Fig. 2 D, right).

Conversely, targeted knockdown of TAGLN2 by siRNA (Fig. 2 E) resulted in diminished cell spreading and F-actin ring formation (Fig. 2 F). Moreover, the F-actin retention time at the contact region was shorter than that of scrambled siRNA-transfected cells (Fig. 2 G). Collectively, these results demonstrate that TAGLN2 controls total F-actin content at the IS that subsequently influences F-actin ring formation and spreading.

Previous experiments have demonstrated that the actin cytoskeleton can regulate integrin-mediated T cell adhesion (Burn et al., 1988). Inhibitors such as cytochalasin B and D (Cyt D) have been used to inhibit β2- and β1-mediated adhesion to their ligands (Rothlein and Springer, 1986). Therefore, we explored whether prolonged accumulation of F-actin at the IS may reflect integrin activation. We performed a T cell–APC conjugation assay and found that J-TG2_GFP cells increased conjugate formation over time (Fig. 3 A). Strong inhibition of conjugate formation in J-TG2_GFP cells treated with Cyt D or TS1/18 (a blocking antibody for LFA-1) demonstrated that TAGLN2 controls LFA-1–mediated T cell adhesion to APC (Fig. 3 B). TAGLN2 overexpression did not change the expression levels of the cell surface molecules CD3, 28, LFA-1, or ICAM-1 (Fig. 3 C). An ICAM-1 adhesion assay further supported TAGLN2-mediated control of integrin function (Fig. 3 D). Conversely, TAGLN2 deficiency reduced conjugate formation (Fig. 3 E).

To understand the mechanism of TAGLN2 involvement in actin assembly at the IS, we first analyzed TAGLN2 and actin interactions in vitro. We confirmed these interactions using a high-speed cosedimentation assay. TAGLN2 was recovered with F-actin in the pellet (Fig. 4 A), confirming the in vivo observation of association with stress fibers or localization at the lamellipodial leading edge. The binding of TAGLN2 to actin monomers was saturated at a 1:1 ratio (B_max = 0.9175 ± 0.102 mol/mol), with a K_d of 7.395 µM (Fig. 4 B). TAGLN1 was reported to bundle actin filaments (Han et al., 2009); thus, we tested for a correlation between the bundling activity of TAGLN2 and the increase in the F-actin content of T cells. We found, by low speed centrifugation, that the well-known actin-bundling protein α-actinin (1 µM) precipitates most of the F-actin; however, TAGLN2 does not show any bundling activity up to a 1:2 ratio of actin to TAGLN2 (8 µM; Fig. 4 C). We saw bundling activity only at higher concentrations (16–24 µM), as assessed by low-speed centrifugation and EM (Fig. 4 D). In contrast to the strong, rope-like structure formed by α-actinin, TAGLN2-actin bundles were loosened, gel-like structures (Fig. 4 D). The optimal ratio of actin to common actin-bundling proteins such as fascin, fimbrin, filamin, and α-actinin is ∼1:0.05 to 1:0.5 (Glenney et al., 1981; Stokes and DeRosier, 1991; van der Flier and Sonnenberg, 2001; Gordón-Alonso et al., 2012). However, the requirement of exceptionally higher concentrations of TAGLN2 needed to cross-link actin implies that TAGLN2 has another function other than F-actin cross-linking.

The aforementioned results prompted us to consider the other functions of TAGLN2 in actin reorganization. We performed an actin polymerization assay with pyrene-labeled actin and compared the polymerizing activity of TAGLN2 with that of Arp2/3 and/or the VCA (verprolin-homology, cofilin homology, and acidic) domain of neuronal WASp (Blanchoin et al., 2000). Although Arp2/3 and VCA successfully enhanced spontaneous polymerization under actin polymerization buffer (APB), TAGLN2 slightly reduced the kinetics of spontaneous actin polymerization (Fig. 5 A). We also measured the concentration of globular actin (G-actin) required for polymerization (critical concentration) in the presence or absence of TAGLN2 and found that addition of TAGLN2 does not change the critical concentration of G-actin (Fig. 5 B).

The binding of fluorescently labeled TAGLN2 along actin filaments (Fig. 5 C) led us to examine whether TAGLN2 may be involved in F-actin stabilization as a side-binding protein. TAGLN2 blocked spontaneous actin depolymerization in concentrations ranging from 1 to 8 µM, which were lower than those needed for bundling (Fig. 5 D). This demonstrated that TAGLN2 stabilizes F-actin. Moreover, TAGLN2 blocked cofilin-mediated depolymerization, as determined by a depolymerization assay and confocal analysis (Fig. 5 E).

The crystal structure of TAGLN1 was reported to contain two cross-linked monomers of TAGLN1 (Li et al., 2008). Using the TAGLN1 homodimer as a template, we predicted the dimerization of TAGLN2 monomers (Fig. 5 F). Interestingly, we found a similar mode of homodimerization for TAGLN2. It was mediated by amino acids at the binding interface that correspond to those of the TAGLN1 homodimer (Li et al., 2008). The structure of the homodimeric TAGLN2 model enabled us to predict an underlying mechanism of TAGLN2 action in F-actin stabilization. As shown schematically in Fig. 5 F, dimerization of TAGLN2 monomers, through the interaction of one side of the first monomer (helix I and II) with the other side of second monomer (helix IV and V), can generate multimers of TAGLN2 that stabilize an incoming actin molecule’s position on the growing F-actin lining. The 1:1 stoichiometry of TAGLN2 to actin also suggests that there is only one actin-binding motif (ABM) on each TAGLN2 monomer (Fig. 4 B). These observations suggest a model in which TAGLN2 functions as a “molecular staple” on polymerized actin, binding to actin at one end (fulcrum) and undergoing multimeric interactions with two arms (Fig. 5 F, right).

Sequence alignment with TAGLN1 and TAGLN3 revealed that TAGLN2 contains a predicted ABM between residues 153 and 160. Deletion of amino acids 153–160 (ΔABM) reduced TAGLN2 association with F-actin (Fig. 6, A and B). This mutant also lost actin-stabilizing activity (Fig. 6 C), further corroborating that the actin stabilizing activity of TAGLN2 is fully dependent on its binding to F-actin. To investigate whether TAGLN2 involvement in T cell activation is related to its binding to actin, we also generated J-TG2ΔABM_GFP cells (Fig. 6 D). We observed diminished accumulation of TAGLN2 at the IS, and the expression of IL-2 mRNA was attenuated in these cells compared with J-TG2_GFP cells (Fig. 6, D and E). These cells also failed to increase conjugate formation (Fig. 6 F). We questioned whether T cell activation by TAGLN2 is caused by its primary role in actin signaling or a secondary effect stemming from increased conjugate formation. To test this, J-GFP, J-TG2_GFP, or J-TG2ΔABM_GFP cells were loaded on anti-CD3/28–coated plates with or without ICAM-1 (±ICAM-1), and IL-2 production was measured. J-TG2_GFP cells on the anti-CD3/28+ICAM-1–coated plates produced the most IL-2. This suggested that TAGLN2 function is enhanced by integrin LFA-1 binding to its ligand ICAM-1 (Fig. 6 G). On anti-CD3/28–coated plates, however, production of IL-2 in J-TG2_GFP cells was also increased when compared with J-GFP or J-TG2ΔABM_GFP cells (Fig. 6 G). This suggests that the primary role of TAGLN2 in regulation of actin is essential for TAGLN2-mediated T cell function.

To connect the in vitro and in vivo data, we generated mice deficient in TAGLN2 (Fig. S3, A and B). The sizes of primary and secondary lymphoid organs were not changed, and the weights of age-matched littermates also did not differ significantly among wild-type (WT; +/+), heterozygous (+/−), and TAGLN2-null (−/−) mice (Fig. S3, C and D). The absence of TAGLN2 expression was confirmed using Western blotting (Fig. S3 E). The TAGLN2^−/− mice showed normal T cell development, in terms of the number and percentage of CD4⁺ and CD8⁺ single-positive cells in the thymus, lymph nodes, and spleen (Fig. S3 F), as well as normal B cell and dendritic cell populations in the spleen (Fig. S3 G). CD3⁺ T cells showed normal surface expression of CD3-ε and CD28 (Fig. S3 H). Response to lipopolysaccharide was not significantly altered in either dendritic cells (DCs; Fig. S3 I) or B cells (Fig. S3 J) taken from TAGLN2^−/− mice.

We also generated transgenic mice by using an Lck (lymphocyte-specific protein tyrosine kinase) promoter-based vector containing GFP or TG2_GFP, and we verified the GFP signals within well-defined T cell regions in lymphoid organs such as the thymus, spleen, and lymph nodes (Fig. S4 A). The expression levels of GFP and TG2_GFP were confirmed (Fig. S4 B). The mice exhibited normal T cell development as determined by the ratio of CD4⁺/CD8⁺ cells (Fig. S4 C).

We compared cell spreading and F-actin ring formation in CD3⁺ T cells purified from TAGLN2^+/+ and TAGLN2^−/− mice before and after stimulation on anti-CD3/28–coated coverslips. A kinetic study revealed that the TAGLN2 deficiency inhibited cell spreading as well as diminished F-actin ring formation (Fig. 7 A). These cells showed a reduction in total F-actin content after T cell stimulation (Fig. 7 B). TAGLN2^−/− CD3⁺ T cells showed a significant reduction of conjugate formation with SEE-loaded CD19⁺ B cells purified from WT mice (Fig. 7 C). The ability to adhere to ICAM-1 was also diminished in these cells, suggesting weakened LFA-1 function (Fig. 7 D).

We next examined the relationship between TAGLN2 and cofilin in vivo. Because active cofilin functions to rapidly disassemble F-actin, we tested the ability of TAGLN2 to overcome cofilin-mediated actin depolymerization in CHO-K1 cells. We generated an active form of cofilin by substitution of Ser-3 to Ala-3 (cofilin/S3A; Moriyama et al., 1996). Overexpression of TAGLN2 markedly protected F-actin within stress fibers from cofilin-mediated destruction, whereas the actin-binding mutant (ΔABM) had little effect (Fig. 8 A). In Jurkat T cells, cofilin/S3A led to impaired spreading (area) on anti-CD3/28–coated plates, presumably as a result of the excessive depolymerization of F-actin (Fig. 8 B and Video 7). However, cotransfection with WT-TG2, but not TG2ΔABM, blocked the cofilin effect (Fig. 8 B), thereby corroborating that TAGLN2 antagonizes cofilin activity in vivo as it does in vitro. Cofilin almost similarly colocalized with TAGLN2 (Fig. 8 B and Video 8). Surprisingly, however, careful examination revealed that cofilin mainly localized at two distinguishable regions in the IS—the central SMAC (corresponding to the “F-actin hole”) and upper region of d-SMAC—where TAGLN2 is rather excluded (Fig. 8 C), suggesting specific locations of two proteins for opposite functions in the IS.

Although most studies have associated cofilin activation with F-actin depolymerization, this relationship does not always hold. Indeed, cofilin activity favors F-actin polymerization via its actin-severing activity in vivo and is therefore known to positively regulate T cell activation (Samstag et al., 2003). If this is the case, there must be signaling cross talk or a cooperative function, but not opposition, between TAGLN2 and cofilin in vivo. Interestingly, we found that the dephosphorylation (activation) of cofilin was not obvious in TAGLN2^−/− CD3⁺ T cells after anti-CD3/28 stimulation, whereas the total amount of cofilin was similar in both types of T cells (Fig. 8 D). The opposite of this effect was also seen in transgenic TG2_GFP CD3⁺ T cells (Fig. 8 D). In addition, we observed reduced phosphorylation of phosphoinositide 3-kinase (PI3K) and extracellular signal–regulated kinase (ERK), an upstream Ras-signaling cascade of cofilin dephosphorylation (Wabnitz et al., 2006), in TAGLN2^−/− CD3⁺ T cells (Fig. 8 E).

TCR triggering leads to proximal and distal intracellular signaling events that are related to actin dynamics. As shown in Fig. 8 F, none of the proximal molecules showed impaired phosphorylation after TCR stimulation in TAGLN2^−/− CD3⁺ T cells. This could be explained by the fact that none of these signaling molecules require actin for recruitment to the IS (Larghi et al., 2013). On the other hand, the phosphorylation of PLC-γ was attenuated in TAGLN2^−/− CD3⁺ T cells (Fig. 8 F). This impairment attenuated distal signaling by reducing PKC-θ and ERK activation (Fig. 8 F). PLC-γ phosphorylation depends not only on recruitment to the IS but also on correct spatial organization, driven by F-actin polymerization and retrograde flow (Babich et al., 2012). These results suggest that TAGLN2 plays a role in the spatial organization of IS, which is essential to maintain signaling activation.

Finally, we examined the physiological significance of TAGLN2 in T cell immunity. CD3⁺ T cells from TAGLN2^−/− mice showed decreased secretion of IL-2 after incubation with the S. aureus enterotoxin B (SEB)–loaded CD19⁺ B cells or stimulation with anti-CD3/28 antibodies (Fig. 9 A). To evaluate more antigen-specific effector functions, we generated OTII TCR⁺ and OTI TCR⁺ mice with or without TAGLN2 gene. Purified CD4⁺ or CD8⁺ T cells from OTII TCR⁺/TAGLN2^−/− and OTI TCR⁺/TAGLN2^−/− mice, respectively, showed a significant reduction of IFN-γ and IL-2 protein levels (Fig. 9 B) along with the decreased proliferation in response to the specific ovalbumin (OVA) peptides (Fig. 9 C). Conversely, CD3⁺, CD4⁺, and CD8⁺ T cells from transgenic mice with TG2_GFP exhibited approximately a twofold increase in IL-2 or IFN-γ expression in response to the indicated stimuli (Fig. S4, D–F).

We next assessed the ability of CD8⁺ T cells to kill target cells. TAGLN2^−/− CD8⁺ T cells exhibited a profound decrease in target cell lysis (Fig. 9 D). Flow cytometric analysis of TAGLN2^−/− CD8⁺ T cells showed reduced expression of lytic granules (granzyme B and perforin) after stimulation (Fig. 9 E), along with the consistent reduction in the secretion of mature granzyme B (Fig. 9 F). On the other hand, CD8⁺ T cells also showed a reduced conjugation to target cells (Fig. 9 G). These results suggest that decreased cytotoxicity of TAGLN2^−/− CD8⁺ T cells is resulted from both reduced expression of lytic granules and weakened adhesion to target cells.

To further confirm that defects in the actin-binding ability of TAGLN2 also affect T cell spreading and activation in vivo, we performed retroviral transduction/reconstitution experiments in TAGLN2^−/− T cells. In contrast to WT cells, TG2ΔABM cells did not properly form F-actin rings when reconstituted cells were loaded on anti-CD3/28–coated coverslips (Fig. 9 H). Moreover, the mutant failed to mediate normal T cell activation when the reconstituted cells were incubated with SEB-loaded CD19⁺ B cells (Fig. 9 I). To further determine whether the signaling defects in TAGLN2 knockouts are a direct result of TAGLN2 function on F-actin polymerization, or a secondary defect in integrin-mediated cell adhesion, the reconstituted cells were loaded on anti-CD3/28–coated plates with or without ICAM-1 (±ICAM-1), and intracellular signaling was examined. In contrast to the TAGLN2^+/+ or WT-transduced TAGLN2^−/− T cells, ΔABM-transduced T cells were not optimally phosphorylated in either condition (±ICAM-1; Fig. 9 J). These results strongly suggested that direct modulation of actin reorganization is most critical for TAGLN2 function and that integrin-mediated adhesion is important.

Once the IS is established, the surface of the distal region is enriched in molecules with large extracellular domains, such as CD43 and CD45 (Delon et al., 2001; Freiberg et al., 2002), whereas a stable centripetal retrograde actin flow drives and regulates T cell activation in the cytoplasmic region (Babich et al., 2012). To meet the needs of spatial and temporal regulation in the IS, actin polymerization and depolymerization are tightly regulated by many accessory proteins like Arp2/3, WASp, WAVE, cofilin, and HS12 (Huang and Burkhardt, 2007). We have found that another actin binding protein, TAGLN2, colocalizes with F-actin in d-SMAC and controls F-actin levels at the IS by stabilizing F-actin and in part antagonizing cofilin-mediated disassembly. This function is also connected to the up-regulation of LFA-1 activity in T cells. Furthermore, transgenic or knockout T cells showed up-regulated or impaired activation, demonstrating that TAGLN2 is an important actin regulator in T cells.

The similarities between TAGLN2 and its other two family members, TAGLN1 and TAGLN3, indicate actin-bundling or -gelling functions (Shapland et al., 1993; Han et al., 2009); however, we detected actin bundling effects only at exceptionally high concentrations of TAGLN2 (four to eightfold greater than actin). Because most actin-bundling proteins can optimally cross-link actin into bundles at concentrations 2–20-fold lower than actin (Glenney et al., 1981; Stokes and DeRosier, 1991; van der Flier and Sonnenberg, 2001; Gordón-Alonso et al., 2012), it is unlikely that TAGLN2-actin bundling is related to its biological function under physiological conditions. Instead, we found that TAGLN2 could stabilize and protect F-actin from cofilin-mediated depolymerization, which was also confirmed by confocal microscopy. The inability of the actin-binding mutant to protect F-actin and to enhance T cell activation suggests that TAGLN2 function is dependent on binding to actin. The mechanism of TAGLN2 stabilization and, further, protection of F-actin from cofilin attack is currently undetermined. We can rule out a role in capping of barbed ends, as the proteins in this category significantly affect spontaneous polymerization. Instead, we hypothesize that TAGLN2 binds along the side of actin filaments and may stabilize filaments by altering their structure, as previously reported (Stokes and DeRosier, 1987). Thus, we suggest a “molecular staple” model for TAGLN2-mediated F-actin stabilization, as described in Fig. 5 F. This feature may be especially beneficial for blocking cofilin activity, as cofilin severs actin filaments by increasing their bending and twisting elasticity (McCullough et al., 2011).

Cofilin is a small, ubiquitously expressed protein that is essential for cell survival and for T cell activation. Although cofilin is an actin-severing protein, its activity can favor F-actin polymerization, because cofilin activity generates actin monomers and new barbed ends where active Arp2/3 complexes can associate and promote F-actin elongation (Huang and Burkhardt, 2007). In this regard, it is very interesting to find that cofilin mainly localizes at upper region of d-SMAC—TAGLN2 is rather excluded at this region and enriched at the bottom region—during IS formation. As these two molecules function oppositely in vivo as well as in vitro, these results suggest that TAGLN2 and cofilin may not compete with each other in vivo but rather cooperate to regulate a stable centripetal retrograde actin flow during IS formation. From this point of view, it is interesting that TAGLN2 deficiency results in attenuated cofilin activity in T cells. There are at least two explanations for the down-regulation of cofilin activity in TAGLN2^−/− T cells. First, TAGLN2^−/− T cells may attenuate cofilin activity to restore the F-actin content, which was reduced by loss of TAGLN2. Therefore, active cofilin may further exacerbate the deficiency of TAGLN2 in T cells. Alternatively, TAGLN2^−/− may attenuate actin-linked signaling pathways including PLC, PI3K, and ERK, presumably caused by the immature actin structure at the IS. Indeed, both PI3K and ERK overlap with the Ras-signaling cascade that regulates dephosphorylation of cofilin in human untransformed T cells (Wabnitz et al., 2006). Thus, down-regulation of cofilin activity might be a natural result of attenuated actin-linked signaling in TAGLN2^−/− T cells.

Another question is why cells do not simply make less cofilin, instead of expressing TAGLN2 to keep actin stable. In this regard, if cellular concentrations of TALN2 are substoichiometric to active cofilin, its role as a cofilin antagonist is unclear. However, we found that the cellular concentration of TAGLN2 is much higher than cofilin in T cells, as determined by quantitative Western blotting (Fig. S5). Moreover, our preliminary results demonstrated that substitution of putative phosphorylation sites in TAGLN2 influences its actin-stabilizing function, while having little effect on F-actin binding (unpublished data). Although more careful experiments are needed to understand how TAGLN2 phosphorylation controls F-actin stabilization, these results strongly suggest that T cells require TAGLN2 to stabilize F-actin levels, thereby controlling T cell function under diverse physiological conditions.

Although intracellular signals such as Zap70 and Erk1/2 are necessary to activate LFA-1, the actin cytoskeleton also plays a role in LFA-1–mediated adhesion (Lub et al., 1997; Sampath et al., 1998). In the present study, increase in T cell conjugation with APC or adhesion to ICAM-1, with WT transfection but not with ΔABM, suggests that actin regulation by TAGLN2 is linked to LFA-1 affinity or avidity changes. However, this fact raises an important question about whether the biological effectiveness of TAGLN2 is primarily caused by actin regulation, a secondary effect stemming from LFA-1 adhesion, or both. The current results show that primary actin signaling is critical, but a secondary function to enhance LFA-1–mediated T cell adhesiveness may also be important, particularly in cases of cytotoxic T lymphocyte–mediated cytotoxicity against target cells. However, further studies are necessary to elucidate how TAGLN2/actin signaling is connected to LFA-1 activation.

In summary, our results indicate that TAGLN2 is a d-SMAC molecule in the IS and plays a key role in mature synapse formation by binding and stabilizing actin. We also provide initial evidence linking TAGLN2 with T cell immunity through its role in cytokine production and cytotoxic effector function. TAGLN2 is the only transgelin family member expressed in immune cells. Moreover, the ability of TAGLN2 to influence activation of T cell immunity may be involved in a variety of immune responses and diseases including cancer, infectious diseases, septic shock, autoimmune diseases, and immediate and delayed type hypersensitivity reactions.

Rabbit polyclonal anti-TAGLN2 and anti-GFP antibodies were raised in rabbits using purified full-length TAGLN2 and GFP proteins (AbFrontier). Rabbit polyclonal anti-TAGLN antibody, mouse polyclonal anti-GFP, and anti–PLC-γ antibodies, and goat polyclonal anti–β-actin antibody were purchased from Santa Cruz Biotechnology, Inc. Rabbit polyclonal antibodies against p-PI3K, p-Zap70, Zap70, Lck, ERK, p-ERK, p-PLC-γ, p-PKC-θ, PKC-θ, and p-cofilin; horseradish peroxidase–conjugated anti–mouse IgG; and anti–rabbit IgG antibodies were obtained from Cell Signaling Technology. Mouse monoclonal antibodies against TAGLN3, myosin IIA, Arp3, and cofilin were purchased from Abcam. Anti–human and –mouse CD28 antibodies were acquired from R&D Systems. FITC-conjugated anti–mouse CD4, and peridinin chlorophyll protein complex (PerCP) Cy5.5-conjugated CD8-α antibodies were purchased from eBioscience. Anti–goat IgG, anti–hamster IgG, TRITC-conjugated anti–mouse IgG, TRITC-phalloidin, poly-l-lysine (PLL), and OVA were purchased from Sigma-Aldrich. OKT3 (human anti-CD3; CRL-8001) and 145-2C11 (mouse anti-CD3; CRL-1975) hybridoma cell lines were purchased from the ATCC. TS1/18 (anti–human LFA-1; HB-203) and R6.5 (anti–human ICAM1) hybridoma cell lines were a gift from T.A. Springer (Harvard Medical School, Boston, MA). Lipofectamine 2000 reagent, Alexa Fluor 647 phalloidin, and carboxyfluorescein succinimidyl ester (CFSE) were obtained from Life Technologies. siRNA-targeting TAGLN2 (M-011468-02; 5′-GCAAGAACGUGAUCGGGUU-3′; 5′-UAUGUGAGCUCAUUAAUGC-3′; 5′-CAACUGGUUCCCUAAGAAA-3′; 5′-UGCAAGCAGCUGAGCGCUA-3′) and a scrambled siRNA (D-001810-01; 5′-UAAGGCUAUGAAGAGAUAC-3′) were obtained as a pool of four siRNA duplexes from Thermo Fisher Scientific. SEE and SEB were obtained from Toxin Technology. OVA 257–264 and OVA 323–339 peptides were purchased from InvivoGen.

Jurkat T cells (TIB-152; ATCC), A7r5 (CRL-1444; ATCC), HEK293T (CRL-1573; ATCC), CHO-K1 (10061; Korean Cell Line Bank) cells, EL4 (TIB-39; ATCC), and Raji B cells were maintained in RPMI 1640 medium or DMEM medium (Gibco/Invitrogen) supplemented with 10% (vol/vol) FBS (Gibco/Invitrogen). After written informed consent, human primary PBLs were isolated from healthy donors using dextran sedimentation and centrifugation through a discontinuous Ficoll gradient (GE Healthcare). Mouse CD3⁺ T cells were purified from the mouse spleen and lymph nodes on a T cell enrichment column (R&D Systems). Mouse splenocytes were dispersed and purified to CD4⁺, CD8⁺, and CD19⁺ populations by MACS cell separation (Miltenyi Biotec). The purity of each population was confirmed to be >95% by flow cytometry.

For generation of TAGLN2 knockout mice, murine genomic DNA for TAGLN2 was obtained from 129/SvJ mouse J1 embryonic stem (ES) cells by PCR. A targeting vector was constructed to delete nucleotides 14,691–15,479 containing exon 2 of TAGLN2 using a long arm fragment and two short arm fragments ligated into the pOSDupDel.Neo vector. The targeting vector was then electroporated into 129/SvJ ES cells after linearization using NotI. Positive recombinant ES cells were identified using PCR and then microinjected into C57BL/6 blastocysts. The resulting chimeric mice were then mated with C57BL/6 mice. Germline transmission was confirmed using Southern blot analysis (Macrogen). Interbreeding of the heterozygous mice was performed to generate homozygous TAGLN2-deficient mice.

Transgenic mice expressing GFP or TG2_GFP on T cells were established. The cDNA fragments corresponding to GFP or human TAGLN2_GFP were subcloned into a transgenic expression vector containing the proximal murine p56Lck promoter and the introns and exons of the human growth hormone gene (Chaffin et al., 1990). The resulting plasmids were used to generate transgenic mouse lines according to standard protocols (Macrogen).

Ova-specific TCR transgenic line I (OTI) mice contain transgenic inserts for mouse Tcra-V2 and Tcrb-V5 genes for recognition of OVA 257–264 in the context of H2-K^b, and ova-specific TCR transgenic line II (OTII) mice contain transgenic inserts for mouse α-chain and β-chain TCR subunits expressed by a CD4⁺ T cell hybridoma designated clone 1.1, which is specific for OVA 323–339 in the context of H2-Ab1 (I-A^b; The Jackson Laboratory). TAGLN2^−/−, GFP, and TG2_GFP mice were then crossed with OTI mice to generate OTI TCR⁺/TAGLN2^−/−, OTI TCR⁺/GFP, and OTI TCR⁺/TG2_GFP mice for five generations. OTII TCR⁺/TAGLN2^−/−, OTII TCR⁺/GFP, and OTII TCR⁺/TG2_GFP mice were also generated by backcrossing in the same way as OTI mice. All mouse lines were confirmed by PCR using genomic DNA. All mice were housed in specific pathogen-free conditions, and all experiments were approved by the Animal Care and Use Committee of the School of Life Sciences, Gwangju Institute of Science and Technology.

To generate TAGLN2 constructs, a human TAGLN2 clone coding for the full-length open reading frame was purchased from ImaGene. The human WT TAGLN2 coding sequence region was ligated to pCDNA3 vector (cytomegalovirus [CMV] promoter; Invitrogen), pEGFP-N1 vector (CMV promoter; Takara Bio Inc.), and pHJ-1 as a lentiviral vector (CMV promoter). TAGLN2ΔABM (Δ154–161) was generated by overlapping PCR. The 408– and 81–base pair products from first-round PCR were used together as an overlapping template to generate TAGLN2ΔABM.

To produce His-tagged TAGLN2 and TAGLN2ΔABM recombinant proteins, the pET-28a vector was used as an expression vector, which has a T7 promoter and provides six His residues at the N terminus of the expressed protein. The coding sequences of TAGLN2 and mutant TAGLN2 were amplified by PCR, and the products were incorporated into pET-28a. For retroviral transduction to mouse primary T cells, TAGLN2 and TAGLN2ΔABM genes were subcloned into the pRV-GFP vector, which is a retroviral plasmid that directs the expression of GFP.

Cofilin cDNA was amplified from cofilin mRNA, and then, the constitutively active mutant cofilin (cofilin/S3A) was constructed by substituting Ser-3 with alanine. The PCR product was inserted into pERFP_N1 (CMV promoter; Takara Bio Inc.).

Transient transfection to human PBLs and Jurkat T cells was mostly performed by Amaxa technology using the human T cell nucleofector and nucleofector kitV (Lonza). Transfection to CHO-K1 cells was performed using Lipofectamine 2000 (Life Technologies). To establish stable cell lines, cDNAs in pHJ-1 vector were cotransfected with lentiviral packaging vectors into 293T cells. After 48 h, the supernatants were collected and spin infected into Jurkat T cells by centrifugation at 800 g for 30 min in the presence of 8 µg/ml polybrene.

Mouse CD3⁺ T cells, PBLs, and Jurkat T cells were stimulated with 10 µg/ml anti-CD3 (OKT3 for human; 145-2C11 for mouse)/2 µg/ml CD28 and secondary goat or hamster IgG antibody (5 µg/ml) for the times indicated in the text. In some cases, T cells were stimulated by 25 µg/ml anti-CD3/2 µg/ml CD28 with or without ICAM-1–coated coverslips. For superantigen stimulation, T cells were incubated with 1 µg/ml SEE-pulsed Raji B cells or 1 µg/ml SEB-pulsed mouse CD19⁺ B cells. For OVA stimulation, CD4⁺ and CD8⁺ T cells isolated from spleens of OTII TCR⁺/TAGLN2^+/+,−/− and OTI TCR⁺/TAGLN2^+/+,−/− mice, respectively, were cultured with T cell–depleted splenocytes pulsed with 10 µg/ml of OVA 323–339 (OTII) and OVA 257–264 (OTI) for 24 h.

Cells or homogenized tissues from a C57BL/6 mouse were lysed in ice-cold lysis buffer (50 mM Tris-HCl, pH 7.4, containing 150 mM NaCl, 1% Triton X-100, one tablet of Complete protease inhibitors, and phosphatase inhibitors) for 1 h on ice. Cell lysates were centrifuged at 16,000 g for 25 min at 4°C, and the supernatant was eluted with SDS sample buffer (100 mM Tris-HCl, pH 6.8, 4% SDS, and 20% glycerol with bromophenol blue) and heated for 5 min. The proteins were separated through 10–12% SDS-PAGE gels and were transferred to a nitrocellulose membrane by means of Trans-Blot SD semidry transfer cell (Bio-Rad Laboratories). The membrane was blocked in 5% skim milk (1 h), rinsed, and incubated with intended antibodies in TBS containing 0.1% Tween 20 (TBS-T) and 3% skim milk overnight. Excess primary antibody was then removed by washing the membrane four times in TBS-T. The membrane was then incubated with 0.1 µg/ml peroxidase-labeled secondary antibody (anti–rabbit or –mouse) for 1.5 h. After three washes with TBS-T, bands were visualized by Western Blotting Detection reagents (iNtRON Biotechnology) and were then exposed to x-ray film.

For Northern blot analysis, human Northern RNA blot (12 major tissues; OriGene) was hybridized with TAGLN2 (5′-GCTGAGCGCTATGGCATTAACACCACTGAC-3′) and β-actin (5′-GGGAAATCGTGCGTGACATTAAGGAGAAGC-3′) cDNAs as probes labeled with α-[³²P]dCTP (PerkinElmer) by random priming using the Prime-It kit (Agilent Technologies), according to the manufacturer’s protocol. Blots were prehybridized for 30–60 min at 68°C in Rapid-hyb buffer (GE Healthcare) and hybridized for 90 min at 68°C in the presence of cDNA probes, following the protocol provided. Blots were washed three times in 2× SSC/0.05% SDS at room temperature for 10 min and twice in 0.1× SSC/0.1% SDS at 68°C for 5 min then exposed to x-ray film with intensifying screens at −70°C.

Total RNA was isolated from cells or homogenized tissues of C57BL/6 mice with TRI Reagent (Molecular Research Center, Inc.) and reverse transcribed using RT Premix (iNtRON Biotechnology). PCR was performed with the following primers (the respective forward and reverse pairs are indicated): human IL2, 5′-CACGTCTTGCACTTGTCAC-3′ and 5′-CCTTCTTGGGCATGTAAAACT-3′; mouse IL2, 5′-TGAGCAGGATGGAGAATTACAGG-3′ and 5′-GTCCAAGTTCATCTTCTAGGCAC-3′; human GAPDH, 5′-CGGAGTCAACGGATTTGGTCGTAT-3′ and 5′-AGCCTTCTCCATGGTGGTGAAGAC-3′; mouse GAPDH, 5′-GCACAGTCAAGGCCGAGAAT-3′ and 5′-GCCTTCTCCATGGTGGTGAA-3′; and mouse TAGLN2, 5′-GGTGCAGCAGAAGATTGAGAA-3′ and 5′-GATCTTCTTTACTGGGGCCTG-3′. The amplification protocol consisted of denaturation at 94°C for 30 s, annealing at 60–62°C for 20 s, and extension at 72°C for 40 s. The 30 cycles were preceded by denaturation at 72°C for 7 min. The expression levels of the human and mouse IL2 genes were evaluated by real-time RT-PCR, unless otherwise indicated. PCR amplification was performed in StepOne Real-Time PCR Systems (Applied Biosystems) for continuous fluorescence detection in a total volume of 10 µl of cDNA/control and gene-specific primers by using SYBR Premix Ex Taq (Takara Bio Inc.). The mRNA levels of the target genes, relative to GAPDH, were normalized using the following formula: relative mRNA expression = 2^{−(ΔCt of target gene − ΔCt of GAPDH)}, where Ct is the threshold cycle value. In each sample, the expression of the analyzed gene was normalized to that of GAPDH and described as the mRNA level relative to GAPDH.

For dynamic T cell imaging, J-GFP and J-TG2_GFP cells were transiently transfected with LifeAct (mCherry) by using Amaxa (Lonza) and plated on coverslips (18-nm diameter; Marienfeld) coated with human anti-CD3/28. After preincubation, time-lapse series of randomly selected, LifeAct-positive cells in imaging medium (137 mM NaCl, 5 mM KCl, 1 mM sodium phosphate, 6 mM d-glucose, 1 mM CaCl₂, 0.5 mM MgCl₂, 1% BSA, and 10 mM Hepes, pH 7.3) at 37°C were imaged using a 100×, NA 1.40 oil immersion objective of a laser-scanning confocal microscope (FV1000; Olympus). Images from the bottom area of each coverslip were recorded every 20 s for a total of 1,000 s. For F-actin accumulation at T cell–B cell contact sites, SEE-loaded Raji B cells (stained for ICAM-1 with cy5-conjugated CBRIC1/11) were mixed with T cells and preincubated in solution for 10 min. After incubation, conjugated cells were placed on PLL-coated coverslips and initiated imaging. Time-lapse and z-stack images were collected every 10 min, for a total of 120 min at 37°C. The data were obtained, processed, and analyzed by FLUOVIEW software (Olympus).

For immunofluorescence staining, T cells were incubated for 5 min or indicated times on PLL or human or mouse anti-CD3/28–coated coverslips. CHO-K1 cells were incubated for 2 h on 10 µg/ml fibronectin-coated coverslips. The cells were fixed, washed with PBS, and permeabilized with 0.1% Triton X-100 in PBS for 10 min. The cells were then blocked with 1% BSA in PBS for 1 h, rinsed with PBS, and incubated with primary antibodies in blocking buffer overnight at 4°C. Secondary antibodies were added after washing, incubated for 1 h at room temperature in the dark, rinsed with PBS, and incubated for 30 min at room temperature with TRITC-phalloidin or Alexa Fluor 647–phalloidin in PBS to stain actin. The cells were then washed and mounted with Fluorescent Mounting Medium (Dako). Images were obtained using a 100×, NA 1.40 oil immersion objective of a laser-scanning confocal microscope (FV1000). For the 3D localization assay, z stacks were generated from optical sections taken at 0.2-µm intervals, and FLUOVIEW program and Volocity software (PerkinElmer) were used for 3D image reconstruction.

The fluorescence intensity profiles were measured by the FLUOVIEW program. To determine the percentage of F-actin ring formation on anti-CD3/28–coated coverslips, actin rings were assessed as ring (R), partial (P), and not cleared (N) and quantified using the following equation: percentage of actin ring formation = (R + P)/(R + P + N) × 100. Actin accumulation at the contact sites in the T cells during IS formation was calculated as the ratio between the mean fluorescence intensity of LifeAct in the half portion of contact side (a) and distal side (b) in the T cells. At least 50–100 cells or conjugates were counted in each of three independent experiments. All data were plotted using KaleidaGraph (Synergy Software).

96-well flat-bottom plates (Thermo Fisher Scientific) were coated with human or mouse rICAM-1-Fc (R&D Systems) or 1% BSA (Sigma-Aldrich) as a negative control. Cells were labeled with 0.5 µM BCECF (2′,7′-bis-(2-carboxyethyl)-5-(and-6)-carboxyfluorescein; Invitrogen), washed, resuspended in culture media, and then pretreated with or without 10 µg/ml TS1/18 or 2 µM Cyt D. For TCR stimulation, cells were incubated for 30 min at 37°C in media supplemented with anti-CD3/CD28 antibodies. T cells were added in triplicate to coated plates (10⁵ cells/well) containing secondary anti-IgG as a cross-linker and incubated for 30 min at 37°C. Input was determined by using a fluorescence plate reader (Molecular Devices). Unbound cells were removed by washing three times with warm PBS, and cell adhesion was calculated by dividing the fluorescence emission of bound cells by the total emission of all cells seeded per well.

Jurkat T cells and Raji B cells were stained with CellTracker Orange CMTMR (5-(and-6)-((4-chloromethyl)benzoyl) amino) and Green CMFDA (5-chloromethylfluorescein diacetate), respectively, according to the manufacturer’s protocol (Invitrogen). Raji B cells (2 × 10⁵) were incubated with 5 µg/ml SEE or vehicle control for 30 min, washed, and resuspended in RPMI 1640 medium. For a blocking experiment, Jurkat T cells were incubated with TS1/18, TS2/4 (10 µg/ml each), and Cyt D (2 µM) for 30 min. For conjugation, equal volumes of B and T cells were mixed and incubated at 37°C for indicated time points. The relative proportion of orange, green, and orange–green events in each tube was determined by two-color flow cytometry using a FACSCanto instrument (BD) and analyzed with FlowJo software (Tree Star). The number of gated events counted per sample was at least 20,000. The percentage of conjugated T cells was determined as the number of dual-labeled (CMFDA and CMTMR positive) events divided by the number of CMFDA-positive T cells. Mouse CD3⁺ T cells and SEB-loaded CD19⁺ B cells conjugates were also stained, incubated, and determined as described as described in Jurkat T cell–Raji B cell conjugation assay.

In cytotoxic T cell–target cell conjugation assay, CD8⁺ T cells from OTI TCR⁺/TAGLN2^+/+ or OTI TCR⁺/TAGLN2^−/− mice were stimulated in vitro by OVA 257–264 peptide-loaded DCs in the presence of 10 U/ml rIL-2 to generate OVA-specific cytotoxic T cells. 2 d after stimulation, cells were washed and seeded into new media daily for 3–6 d. OVA-specific cytotoxic T cells were purified by MACS beads, and mixed with EL4 cells, which were used as target cells, loaded with 10 µg/ml OVA 257–264 peptide for 1 h. Conjugates were determined as described in the previous paragraph.

Expression of recombinant proteins in Escherichia coli BL21 (DE3) cells followed the protocol used for transformation of the aforementioned recombinant plasmids. Expression of recombinant proteins was induced by addition of 0.3 mM IPTG to culture media for 4 h at 37°C, and then, cells were collected. The cell pellets were resuspended in PBS, sonicated, and centrifuged. After centrifugation, recombinant proteins in the supernatant were purified by affinity chromatography on a His-selected Nickel Affinity Gel (Sigma-Aldrich). The gel was equilibrated with 10 volumes of buffer (50 mM sodium phosphate, pH 8.0, and 0.3 M NaCl) and incubated with the supernatant. The gel was washed with 5 volumes of wash buffer (50 mM sodium phosphate, pH 8.0, 0.3 M NaCl, and 10 mM imidazole). The protein was eluted with increasing concentrations of imidazole up to 250 mM, and the collected fractions were dialyzed to remove imidazole. The proteins were assessed by SDS-PAGE and Coomassie blue staining.

Actin cosedimentation assays were performed as described previously (Kwon et al., 2013). Nonmuscle actin derived from human platelets was purchased (Cytoskeleton, Inc.). Unless otherwise indicated, actin was used at a final concentration of 4 µM. Actin was mixed in G buffer (0.2 mM CaCl₂, 0.2 mM ATP, and 5 mM Tris-HCl, pH 8.0) to produce an actin stock. Actin was polymerized in APB (50 mM KCl, 2 mM MgCl₂, 0.2 mM ATP, and 5 mM Tris-HCl, pH 8.0) at room temperature for 1 h and then incubated with His-tagged TAGLN2 (8–24 µM) for 30 min at room temperature. 1 µM α-actinin and 1 µM BSA were used as positive and negative controls. Actin filaments with bound proteins were pelleted by centrifugation at 100,000 g for 2 h (for the F-actin binding assay) or 10,000 g for 10 min (for the F-actin bundling assay) at room temperature. Equal amounts of pellet and supernatant were resolved using SDS-PAGE, and proteins were visualized by Coomassie blue staining. The percentage of actin in supernatant and pellet was quantified densitometrically using ImageJ (National Institutes of Health).

Quantitation of actin binding affinity was performed as previously described (Kwon et al., 2013). In brief, the intensity ratio of recombinant protein to F-actin in each pellet was converted to a molar ratio (TAGLN2/actin) using standard curves run on each gel that contained known amounts of TAGLN2 (0.4–5.6 µM) and actin (4 µM) in molar/molar ratios. Cosedimentation binding data were plotted as a function of the free TAGLN2 concentration added and fitted according to the equation: [C_bound]/[actin] = B_max[C_free]/K_d + [C_free], where [C_bound] and [C_free] are the bound and free concentrations of TAGLN2, [actin] is the total actin concentration, B_max is the maximal molar binding ratio, and K_d is the dissociation constant (micromolars). Data were fit using the Excel Solver package (Microsoft) by varying the values of B_max and K_d and minimizing the sum of squares between the actual and predicted binding ratios.

Actin polymerization assays were performed using the Actin Polymerization Biochem kit (Cytoskeleton, Inc.) in which the rate of conversion of pyrene-labeled G-actin into F-actin was monitored. In brief, pyrene-labeled G-actin was thawed at 4°C and centrifuged at 14,000 g for 30 min at 4°C to remove residual F-actin. Unless otherwise indicated, 4 µM G-actin, 10 nM bovine brain Arp2/3 complex (Cytoskeleton, Inc.), recombinant human GST-tagged VCA domain of WASp (400 nM; Cytoskeleton, Inc.), and His-tagged TAGLN2 (4 µM) were used for polymerization assay. Actin polymerization was initiated by adding G-actin stock to the wells with APB and target proteins. The plate was read with a fluorometer (GEMINI XS; Molecular Probe), with excitation at 355 nm and emission at 405 nm. Fluorescence data were zeroed by subtracting baseline fluorescence from all fluorescence values and normalized by dividing all fluorescence values by the mean fluorescence of at the plateau.

For monitoring actin filament disassembly, 20 µM pyrene G-actin was polymerized in APB for 1 h in the presence or absence of various concentrations of TAGLN2 or phalloidin (20 µM; Cytoskeleton, Inc.). Reactants were then diluted to 4 µM with G-actin buffer to induce spontaneous depolymerization. In the case of cofilin-mediated depolymerization, 1 µM cofilin (Cytoskeleton, Inc.) in G-buffer was added to F-actin reactants during the dilution step. Decreasing pyrene fluorescence was measured as in the actin polymerization assay. Normalization for fluorescence intensity was performed by dividing fluorescence value of reactants with fluorescence value of the reactant in the presence of phalloidin at each time point. Changes in fluorescence intensity (ΔF) during depolymerization were calculated by subtracting the fluorescence measured at last time point from the fluorescence measured at initial time point. Changes of each sample (ΔF of TAGLN2 or TAGLN2 + cofilin) were subsequently divided by ΔF of cofilin.

4 µM G-actin was incubated with APB in the presence or absence of 8 µM TAGLN2 conjugated with Dylight488 (Thermo Fisher Scientific) for 30 min at room temperature. One-tenth of the reaction volume was subsequently stained with TRITC-phalloidin and placed on a PLL-coated coverslip. In some cases, G-actin was exchanged with 4 µM 488-actin (Hypermol), and mixed with TAGLN2 (0–8 µM) and cofilin (0–1 µM) on coverslips coated by N-ethylmaleimide–myosin II. The fluorescent images were collected using confocal microscopy.

Negative staining was used to visualize F-actin under EM. After actin polymerization, the samples were loaded onto cooper grids coated with Formvar and carbon for 30 s and excess liquid was removed with filter paper. Actin bound on the grids was stained with 1% uranyl acetate (Sigma-Aldrich) for 10 min and then washed with distilled water. Excess liquid was removed, and grids were air dried. EM observation was performed using a transmission electron microscope (TECNAI G2; FEI) operated at 120 kV.

Jurkat T cells or mouse T cells were maintained in serum-free medium for 12 h and incubated with anti-CD3/28 for the indicated times at 37°C. The reactions were terminated by adding 4% paraformaldehyde. Fixed cells were washed once with PBS and resuspended in PBS containing 1% BSA and 0.25% Triton X-100 for 5 min. After permeabilization, cells were washed, stained for 30 min with TRITC-phalloidin, and then analyzed by flow cytometry.

The crystal structure of human TAGLN1 was used as a template to superimpose a TAGLN2 monomer on each of two TAGLN1 molecules using the online server SuperPose (Maiti et al., 2004). In addition, the prediction of a TAGLN2 homodimer was performed computationally by the ClusPro online server, and the same interface was observed as for the TAGLN1 template. Interactions of the TAGLN2 homodimer were analyzed by University of California, San Francisco Chimera and Discovery Studio Visualizer software (Accelrys Software, Inc.) and optimized by virtual inspection to overcome the steric hindrances.

Jurkat T cells, mouse splenocytes, T cells, B cells, and DCs were suspended in PBS containing 2% FBS. Cells were stained with fluorescence-conjugated antibodies against human CD3-ε, CD28, LFA-1, ICAM-1, and mouse CD4, CD8, CD3-ε, CD28, B220, CD11c, MHCII, CD86, and CD80 for 30 min at 4°C. The corresponding isotype controls were used in each antibodies. For intracellular granzyme B and perforin staining, OVA-specific cytotoxic T cells were restimulated with 200 nM PMA and 1 µM ionomycin for 4 h at 37°C. Brefeldin A was added during the last 2 h of restimulation. Cells were stained with anti-CD8, and then fixed, permeabilized, and stained with antibodies specific for granzyme B and perforin (eBioscience). Cells were acquired on a FACSCanto, and data were analyzed with FlowJo software.

Splenic CD3⁺, CD4⁺, or CD8⁺ T cells (5 × 10⁴ cells/sample) were stimulated as described in T cell activation section. After 24–48 h, the amounts of IL-2 and IFN-γ in the supernatants were determined by ELISA with Duo Set Mouse ELISA kits (R&D Systems). To measure granzyme B secretion, OVA-specific cytotoxic T cells generated previously described in Conjugation assay section were restimulated with EL4 cells loaded with 10 µg/ml OVA 257–264 at a 20:1 (effector cells/target cells) ratio for 4 h. The amounts of granzyme B in the supernatants were determined by ELISA.

CD4⁺ and CD8⁺ T cells were labeled for 30 min at 37°C with 10 µM CFSE (Molecular Probe) and washed twice in PBS. Labeled T cells were stimulated with anti-CD3/28 antibodies. After 72 h, T cells were collected and analyzed by flow cytometry for dilution of CFSE intensity.

For the in vitro cytotoxic T cell activity assay, OVA-specific T cells were generated and incubated with target cells described in Measurement of secretion of cytokines and granzyme B section at various ratios of effector cells to target cells. After 4 h, cell-mediated cytotoxicity was determined using CytoTox 96 NonRadioactive cytotoxicity assay kit (Promega), and the percentage of cytotoxicity was calculated according to manufacturer’s instructions.

Before retroviral transduction, mouse CD3⁺ T cells from TAGLN2^+/+ and TAGLN2^−/− mice were incubated in 10 µg/ml anti-CD3–coated coverslips with 2 µg/ml anti-CD28 and 10 µg/ml rIL-2 for 48 h. A total of 6 × 10⁶ retroviral packaging cells (platinum-E cells; Cell Biolabs) were plated overnight in 10-cm² dishes. Retroviral particles were generated by transfection with retroviral vectors (empty vector, TG2, and TG2ΔABM) and pCL-Eco packaging vector using Lipofectamine 2000 (Invitrogen). After 48 h, virus supernatants (1 ml) were harvested, mixed with 10⁶ mouse T cells and polybrene (8 µg/ml) in a 12-well plate and centrifuged for 90 min at 2,000 g at 32°C with 20 U/ml rIL-2. The transduced T cells were maintained with fresh media with rIL-2 and expanded for 5–8 d.

Mean values were calculated using data taken from at least three independent experiments conducted on different days. Where significance testing was performed, unpaired Student’s t tests and one-way analysis of variance tests were used. Differences between groups were considered significant at P < 0.05.

Fig. S1 shows that TAGLN2 is expressed in immunogenic tissues and cells and knockdown of TAGLN2 reduces T cell activation. Fig. S2, which corresponds to Fig. 1 F, shows individual z-stack images acquired as optical sections from bottom to top. Fig. S3 shows the strategy to generate TAGLN2 knockout mice and characterization of their population, development, and expression of indicated surface molecules. Fig. S4 shows the strategy to generate GFP-T or TG2_GFP-T transgenic mice and characterization of their T cell development and cytokine secretion in response to TCR activation. Fig. S5 shows the cellular concentrations of TAGLN2, cofilin, and actin, as determined by quantitative Western blotting. Video 1, corresponding to Fig. 1 E, shows a 3D view of TAGLN2_GFP and ICAM-1 localization at the contact site of Jurkat T cells and Raji B cells. Videos 2–4, corresponding to Fig. 1 F, show localization of TAGLN2, Arp3, myosin IIA, and F-actin in Jurkat T cells placed on anti-CD3/28–coated coverslips. Videos 5 and 6, corresponding to Fig. 2 A, show real-time F-actin ring formation and dynamic localization of TAGLN2 and actin during J-GFP or J-TG2_GFP cell spreading on anti-CD3/28–coated coverslips. Videos 7 and 8, corresponding to Fig. 8 B, show colocalization of cofilin, TAGLN2, and F-actin in Jurkat T cells placed on anti-CD3/28–coated coverslips.

Image acquisition was performed using the imaging facility of the Bio-Imaging Research Center at the Gwangju Institute of Science and Technology (GIST).

This work was supported by the Bio & Medical Technology Development Program (2011-0030157), Basic Science Research Program (2012R1A2A1A03002115), Basic Science Research Program through the National Research Foundation of Korea funded by the Ministry of Education (2013R1A6A3A04064259), and Cell Dynamics Research Center Program (2007-0056244) through the National Research Foundation grants funded by the Ministry of Science, ICT (Imagination, Creativity, and Science Technology) and Future Planning, Korea. This work was also partially supported by the Bio-Imaging Research Center at GIST.

The authors declare no competing financial interests.