Cell-permeable transgelin-2 as a potent therapeutic for dendritic cell-based cancer immunotherapy

Previously, we reported that transgelin-2 is highly expressed in immune-related tissues, such as the thymus, spleen, and LNs [16]. Transgelin-2 is also dominantly expressed in lymphocytes and plays an important role in stabilizing the IS, thereby enhancing T cell-mediated immune responses [16, 17, 23]. In addition, we found that transgelin-2 is physically associated with the integrin lymphocyte function-associated antigen-1 (LFA-1), which enhances the adhesion of cytotoxic T cells to intercellular adhesion molecule-1 (ICAM-1)-positive tumor target cells, such as E0771 cells, but not ICAM-1-negative B16F10 cells [17]. In accordance with the previous results, overexpression of transgelin-2 (TG2) in OTI T cells did not enhance T cell adhesion to ovalbumin (OVA)-pulsed B16F10 target cells in vitro and showed only a mild effect on B16F10 tumor regression in vivo (Fig. 1a, b). Interestingly, however, we observed that whole-body transgelin-2-KO (Tagln2^−/−) mice, as compared to wild-type (WT) mice, developed rapid B16F10-derived tumor growth, resulting in increased tumor weights and sizes (Fig. 1c). Kaplan–Meier survival studies 40 days after tumor injection showed that WT mice had a median survival time of 35 days. However, Tagln2^−/− mice had a median survival time of only 28 days (Fig. 1c). These results suggest that cell types other than T cells can participate in tumor regression independently of the LFA-1/ICAM-1 interaction between cytotoxic T cells and tumor target cells.

Since DCs play a major role in processing and presenting antigen peptides to antigen-specific T cells, we next asked whether DCs express transgelin-2 or other isotypes, such as transgelin-1 and -3. Interestingly, immature DCs isolated from bone marrow (BM) did not express transgelin family members (Fig. 1d). However, transgelin-2 was highly expressed during GM-CSF-induced differentiation (Fig. 1d). The expression of transgelin-2 was also induced in fully differentiated BMDCs in response to LPS (Fig. 1e), suggesting a specific role of transgelin-2 in mature BMDCs.

To understand the significance of transgelin-2 expression in mature DCs, we used mature BMDCs from WT (Tagln2^+/+) and Tagln2^−/− mice (Fig. 1f), which were previously generated in our laboratory [16]. To determine whether transgelin-2 regulates the functions of BMDCs, OVA257–264-pulsed WT or transgelin-2-knockout (KO) BMDCs were i.v. injected into mice, thereby generating cytotoxic T cell clones against the OVA257–264 peptide in vivo (Fig. 1g). At 7 days after DC injection, the mice were i.v. injected with OVA⁺B16F10 melanoma cells. The metastatic colonies were evaluated at day 14, and a significant increase in metastatic nodules was observed in mice with adoptively transferred OVA257–264-pulsed Tagln2^−/− BMDCs compared to OVA257–264-pulsed WT BMDCs (Fig. 1h). For the solid tumor model, OVA⁺B16F10 cells were implanted into the mammary fat pads of C57BL/6 female mice previously injected with WT BMDCs or Tagln2^−/− BMDCs. In accordance with the metastatic model, Tagln2^−/− BMDCs could not optimally control tumor growth in mice (Fig. 1i). Kaplan–Meier survival studies showed that mice injected with OVA257–264-pulsed WT BMDCs had a median survival time of 40 days. However, mice adoptively transferred with OVA257–264-pulsed Tagln2^−/− BMDCs had a median survival time of only 34 days similar to PBS control group (Fig. 1j). Further, we observed a significant reduction in CD8⁺ tumor-infiltrating lymphocytes (TILs) in the B16F10 tumors from Tagln2^−/− DC-injected mice (WT vs. KO, 9.8 ± 1.5 vs. 1.8 ± 1.1) (Fig. 1k). In addition, the tumor-infiltrating CD8⁺ T cells from Tagln2^−/− DC-injected mice expressed a lower level of IFNγ than that in WT DC-injected mice (Fig. 1l). Among various DC subsets, conventional type 1 DCs (cDC1s) are the main cellular source of IL-12, a fundamental cytokine for anti-cancer CD8⁺ CTL activation, and the super promising DC subset for induction of anti-tumor immunity due to their superior capacity to uptake dying or dead cell materials and to process tumor-associated antigens for cross-presentation [24‐26]. We therefore determined whether transgelin-2 is also expressed in Flt3L-induced cDC1s. To this end, BM-isolated immature DCs were treated with Flt3 ligand for 9 days, and analyzed the expression of surface markers and transgelin-2 (Additional file 1: Fig. S1A and B). We found that transgelin-2 was also dramatically induced during Flt3L-induced cDC1 differentiation, while transgelin-2-KO followed the normal differentiation patterns as judged by the expressions of CD103, CD24, and XCR1 (Additional file 1: Fig. S1A and B). We next evaluated the antitumor activity of Tagln2^−/− cDC1s against s.c. injected OVA⁺B16F10 cells. Similar to the GM-CSF-induced BMDCs, Tagln2^−/− cDC1s were not able to optimally control tumor growth in mice (Additional file 1: Fig. S1C), suggesting that transgelin-2 is also important for the function of cDC1s to facilitate optimal antigen presentation.

To investigate the mechanism by which transgelin-2 affects the function of DCs, we analyzed mature BMDCs obtained from Tagln2^−/− mice and compared them with WT BMDCs in terms of their morphology, actin dynamics, and signaling. Morphologically, although there were no gross differences between WT and KO mice, Tagln2^−/− BMDCs exhibited an absence of filopodia-like membrane protrusions, as determined by scanning electron microscopy (SEM) (Fig. 2a, yellow arrowheads). To understand this morphological phenotype, LPS-stimulated WT BMDCs were stained for F-actin and transgelin-2 and visualized by confocal microscopy. As shown in Fig. 2b, transgelin-2 was specifically localized at the leading edge of lamellipodia, where filopodia are formed from the pre-existing meshwork of Arp2/3 complex-branched filaments. In addition, transgelin-2 co-localized with F-actin in the actin-rich cores of podosomes, which were surrounded by vinculin in the stereotypical podosome organization (Fig. 2b, right). Podosomes are characteristics of cells in the myeloid lineage, including DCs, macrophages, and osteoclasts, which can degrade extracellular matrices and play a role in the migration of DCs through tissues [27]. Thus, the localization of transgelin-2 in filopodia tips, as well as in podosomes, demonstrated that transgelin-2 is involved in the membrane protrusions that support DC migration and interactions with T cells for proper antigen presentation. To this end, we next assessed podosome formation and spreading during DC activation in response to LPS and fibronectin (Fn), respectively.

DCs have been shown to exhibit transient podosome loss at approximately 20 min after their activation with LPS and then recover them again after 2 h [27, 28]. Likewise, in our study, BMDCs from both WT and KO mice showed a substantial reduction in the number of cells with podosomes after 30 min of stimulation with LPS. However, podosomes were significantly slower to reform in Tagln2^−/− BMDCs after 2 h (Fig. 2c). Next, since the interaction of BMDCs with extracellular matrix (ECM) proteins, such as Fn, induces a dynamic cytoskeletal rearrangement driven by actin polymerization, thereby mediating cell adhesion and migration through integrins [28, 29], we evaluated the effects of the transgelin-2 KO on DC spreading and actin polymerization. Compared to WT BMDCs, Tagln2^−/− BMDCs exhibited reduced cell spreading and lower levels of polymerized actin when seeded on Fn (Fig. 2d). Taken together, these results strongly suggest that transgelin-2 has a specific role in DC migration by modulating filopodia and podosome formation, and cell spreading on ECM.

CCR7 is necessary for directing DCs to secondary lymphoid nodes and eliciting an adaptive immune response [30, 31]. Furthermore, CCR7 induces actin rearrangements by activating Akt and Erk signaling, as well as the G_i-dependent activation of MAPK members Erk1/2, Jnk, and p38 [32]. To determine whether transgelin-2 is linked to dynamic actin signaling in mature BMDCs, we examined signaling downstream of CCR7 activation by CCL19. PI3K and its downstream effector Akt were both significantly attenuated in Tagln2^−/− BMDCs. Similarly, p38 and Erk signaling were also reduced in these cells (Fig. 2e).

Reduced CCR7-mediated signaling led us to examine whether transgelin-2 deficiency was connected with the expression of chemokine receptors or adhesion molecules on the surface of DCs. We found no differences in the expression levels of surface proteins between WT and KO BMDCs. However, Tagln2^−/− BMDCs exhibited dramatic defects in migration in response to various chemokines, such as CCL3, 5, 19, 21, and SDF-1α (Fig. 3a). We further performed live time-lapse imaging to further monitor how DCs responded to stimulation with the chemokine CCL19. Although WT BMDCs readily spread, formed membrane protrusions, and showed a typical dendritic morphology, Tagln2^−/− BMDCs were round and irregularly shaped, with a relative absence of a protrusive morphology (Fig. 3b and additional movie files show this in more detail. [See Additional files 2–5]), suggesting that the spatial and temporal regulation of the actin cytoskeleton is impaired in Tagln2^−/− BMDCs.

We therefore next asked whether DC migration was also impaired in vivo in Tagln2^−/− mice. To this end, we co-injected differentially labeled BMDCs from WT or Tagln2^−/− mice into the footpads of normal recipient mice. Draining popliteal LNs were collected at 24 h post-injection, and the presence of migrated DCs was assessed by flow cytometry (Fig. 3c). Fixed tissues were also examined under confocal microscopy (Fig. 3d). In contrast to WT BMDCs, which reached the draining LNs normally, there was a significant reduction in the number of Tagln2^−/− BMDCs observed in the draining LNs (Fig. 3c, d), thus demonstrating that transgelin-2 influences the migration rate of DCs in vivo.

Although enlarged draining popliteal LNs were observed in recipient OTII TCR mice adoptively transferred with OVA323–339-pulsed WT BMDCs, no significant change in LN size was seen with Tagln2^−/− BMDCs (Fig. 4a). This result suggests that antigen-specific OTII CD4⁺ T cells may be less proliferative in mice injected with Tagln2^−/− BMDCs. As expected, the population of Tagln2^−/− BMDCs (CD11c⁺ cells) that migrated into the draining LN was significantly smaller than that of WT BMDCs (Fig. 4b). Interestingly, however, the expression of co-stimulatory molecules, such as CD80 and CD40, was not altered in Tagln2^−/− BMDCs in vivo, suggesting that transgelin-2 has little effect on the ability of DCs to express these co-stimulatory factors. This result led us to ask whether the migratory defects seen in Tagln2^−/− BMDCs were the only driver of the reduced antitumor response (Fig. 1h, i) or whether another mechanism was also involved. To this end, we investigated the role of transgelin-2 in the direct DC-mediated T cell response in vitro. OVA323–339-pulsed BMDCs were co-cultured with OTII CD4⁺ T cells, and the surface expression of CD69 or CD25 on T cells was determined. Interestingly, the expression of these activation markers was significantly reduced in T cells co-cultured with Tagln2^−/− BMDCs, thereby indicating a second role for transgelin-2 in DCs.

To understand how Tagln2^−/− BMDCs attenuated T cell activation in vivo, we performed an in vitro activation assay. We first determined whether transgelin-2 KO affected the differentiation of immature DCs into mature DCs. During GM-CSF-induced differentiation, Tagln2^−/− BM cells followed the normal differentiation patterns based on the surface expression of CD11c (Fig. 5a). In addition, the expression levels of MHC class II and co-stimulatory molecules, such as CD80, CD86, and CD40, were not significantly different in either WT or Tagln2^−/− BMDCs by LPS stimulation (Fig. 5b). Further, the expression levels of DC cytokines such as IL-1β and IL-12 were similar in both WT and Tagln2^−/− BMDCs (Fig. 5c). By contrast, the activation of OTII CD4⁺ T cells was significantly reduced in OVA323–339-pulsed Tagln2^−/− BMDCs as determined by the expression of CD69 and CD25, and the secretion of IL-4, IL-2, and IFN-γ (Fig. 5d–f). Indeed, higher levels of antigen peptide were required to produce a similar amount of cytokines in the Tagln2^−/− BMDCs (Fig. 5e, f). Thus, we asked whether Tagln2^−/− BMDCs would be unable to optimally activate antigen-specific T cells to expand in vivo. To this end, C57BL/6 mice were injected with OVA257–264-pulsed WT or Tagln2^−/− BMDCs via the footpad. After 7 days, CD3⁺ T cells were purified and examined for proliferation and cytokine secretion in response to the same OVA257–264-pulsed WT BMDCs in vitro. As shown in Fig. 5g, h, Tagln2^−/− BMDCs induced lower levels of cell proliferation and cytokine secretion, suggesting that they could not optimally support T cell clonal expansion in vivo.

T cell activation and differentiation require sustained interaction with cognate DCs via integrin affinity and avidity regulation. In a previous study, we showed that transgelin-2 in T cells is associated with LFA-1 and can regulate LFA-1 avidity [17]. We therefore tested whether transgelin-2 is also involved in DC-mediated T cell adhesion. However, SEM analysis revealed no differences between WT and Tagln2^−/− BMDCs (Fig. 6a). To assess the adhesion strength between BMDCs and T cells in a quantitative manner, we performed a conjugation assay in the presence of an OVA peptide. We found that Tagln2^−/− BMDCs showed a significant reduction in conjugation (Fig. 6b), demonstrating that transgelin-2 supports adhesion between DCs and antigen-specific T cells.

To evaluate the role of transgelin-2 in regulating interactions between DCs and T cells in the live LN, we performed three-dimensional live-imaging of DCs using two-photon microscopy. OVA323–339-pulsed WT or Tagln2^−/− BMDCs labeled with Cell Tracker CMFDA were s.c. injected into the rear footpads of C57BL/6 mice. At 24 h post-injection, OTII CD4⁺ T cells labeled with CMRA were i.v. injected (Fig. 6c). As previously observed (Fig. 3d), a significantly smaller number of Tagln2^−/− BMDCs migrated into the LN as compared to WT BMDCs (Fig. 6c). Moreover, T cells formed more stable contacts with WT BMDCs than with Tagln2^−/− BMDCs (Fig. 6d). As a result, the T cell speed in the LN was reduced, and track displacement differences were not significant (Fig. 6d). Collectively, these results demonstrate that transgelin-2 is essential not only for the delivery of antigen materials through the dynamic movement of DCs but also for maintaining sustained interactions between DCs and T cells for the initiation of adaptive immune responses.

In a previous report, we demonstrated that cell-permeable recombinant transgelin-2 fused with PTD (TG2P) enhanced cytotoxic T cell-mediated anticancer activity through increased LFA-1 avidity [17], suggesting that TG2P can compensate for endogenous transgelin-2 without viral transduction. WT-TG2P was rapidly internalized into the DCs (Fig. 7a). Figure 7b shows a schematic of the secondary structure of transgelin-2 fused with PTD. However, because we expected that natural transgelin-2 could be degraded by the ubiquitin pathway, as predicted by the program UbiSite [33], we substituted the potential ubiquitination site lysine (K)78 with arginine (R) to generate de-ubiquitinated TG2P (dU-TG2P) (Fig. 7b). Interestingly, dU-TG2P (K78R) showed remarkable stability and persisted longer than 24 h in the cytosol of BMDCs (Fig. 7c). Therefore, we next asked whether dU-TG2P could mimic the actions of transgelin-2 in Tagln2^−/− BMDCs. To this end, Tagln2^−/− BMDCs were incubated with dU-TG2P (10 μM) for 2 h, and then, the cells were seeded on Fn-coated plates. After 90 min, the size of the BMDCs was determined by flow cytometry, revealing that dU-TG2P significantly increased the size of Tagln2^−/− BMDCs (Fig. 7d). To rule out a potential effect of LPS contamination and to confirm the specificity of the recombinant transgelin-2, dU-TG2P was heated inactivated (heat = H, H/dU-TG2P), and its effects on Tagln2^−/− BMDCs were examined. Heat inactivation completely abolished the ability of dU-TG2P to increase the size of Tagln2^−/− BMDCs (Fig. 7d), suggesting that the efficacy of dU-TG2P was solely mediated by transgelin-2. dU-TG2P (10 μM) also significantly increased the spreading of Tagln2^−/− BMDCs on Fn, whereas H/dU-TG2P had a little effect as determined by confocal microscopy and image analysis (Fig. 7e). Consistently, dU-TG2P remarkably increased T cell adhesion to Tagln2^−/− BMDCs in a dose-dependent fashion (Fig. 7f). Further, dU-TG2P-treated Tagln2^−/− BMDCs supported more rapid antigen-specific (OT-II) T cell proliferation and cytokine production than untreated or H/dU-TG2P-treated Tagln2^−/− BMDCs (Fig. 7g, h).

Although we have previously shown that TG2P could enhance cytotoxic T cell-mediated anticancer activity [17], its efficacy may be limited unless a sufficient number of tumor-specific T cell clones are obtained. From this point of view, the amplification of the DC functions may be more important because they can generate numerous T cell clones targeting tumor antigens. We therefore asked whether dU-TG2P could promote the efficacy of WT BMDCs, which would prove useful for DC-based therapeutic applications. WT BMDCs treated with dU-TG2P exhibited significantly increased spreading on Fn (Fig. 8a). Moreover, dU-TG2P treatment produced increased adhesion between WT BMDCs and activated T cells (Fig. 8b), which stimulated the release of increased levels of T cell cytokines (Fig. 8c). Likewise, a larger number of T cells proliferated with co-cultured dU-TG2P-treated BMDCs than with untreated BMDCs (Fig. 8d).

We further asked whether dU-TG2P also could promote the efficacy of WT cDC1s. Similar to the result of WT BMDCs, dU-TG2P increased conjugation between CD8⁺ OTI T cells and WT cDC1s (Additional file 1: Fig. S2A). In addition, increased conjugation correlated with the increased cytokine productions (Additional file 1: Fig. S2B). Accordingly, a larger number of T cells proliferated with co-cultured dU-TG2P-treated cDC1s than with untreated cDC1s (Additional file 1: Fig. S2C).

To evaluate the effects of dU-TG2P-treated BMDCs in the in vivo tumor model, OVA257–264-pulsed untreated BMDCs or dU-TG2P-treated BMDCs were i.v. injected into mice, followed by OVA⁺B16F10 melanoma cells, as described in Fig. 1g. A significant reduction in metastatic nodules was observed in mice with adoptively transferred OVA257–264-pulsed dU-TG2P-treated BMDCs compared to OVA257–264-pulsed untreated BMDCs (Fig. 8e). Consistent with the metastatic model, dU-TG2P-treated BMDCs significantly suppressed tumor growth in mice (Fig. 8f). Kaplan–Meier survival studies showed that mice injected with OVA257–264-pulsed BMDCs had a median survival time of 43 days and mice adoptively transferred with OVA257–264-pulsed dU-TG2P-treated BMDCs had an 80% probability of surviving longer than 47 days (Fig. 8g). Collectively, these results indicate that dU-TG2P is a promising therapeutic approach for DC-based cancer immunotherapy.

Rabbit polyclonal anti-transgelin-2 antibody was raised in rabbits using purified full-length transgelin-2 (AbFrontier, Seoul, Korea). In addition, the following antibodies were used: goat polyclonal anti-TAGLN1 (Santa Cruz Biotechnology, Dallas, TX, USA); rabbit polyclonal anti-β-actin; rabbit polyclonal antibodies against p-PI3K, t-PI3K, p-AKT, t-AKT, p-p38, t-p38, p-ERK, t-ERK, His, HRP-conjugated anti-mouse IgG, anti-goat IgG, and anti-rabbit IgG (Cell Signaling Technology, Danvers, MA, USA); mouse monoclonal anti-TAGLN3 and anti-vinculin (Abcam, Cambridge, MA, USA); and antibodies for FITC-conjugated CD40 (MA5-16506), MHCII (11-5322-82), CD18 (LFA-1β; 11-0181-82), ICAM-1 (11-0541-82), CD11c (17-0114-82), PE-conjugated CD86 (12-0862-82), CD80 (12-0801-82), CD25 (120251-82), CD69 (12-0691-82), CXCR4 (12-9991-82), CCR7 (12-1971-82), APC-conjugated CD11c (17-0114-82), and B220 (17-0452-82) (eBioscience, San Diego, CA, USA); FITC-conjugated CD24 (M1/69), CD103 (2E7), and SIRPα (P84) (BioLegend, San Diego, CA, USA). All antibodies for flow cytometry were used at a dilution of 1:100. Phalloidin-TRITC, lipopolysaccharide, and Fn were purchased from Sigma Aldrich (St. Louis, MO, USA). GM-CSF, CCL-3, CCL-5, CCL-19, CCL-21, and SDF-1α were purchased from Peprotech Inc. (Rocky Hill, NJ). Alexa647-phalloidin, CellTracker CMFDA-green, CMRA-Orange dyes, anti-mouse Alexa 647, and anti-rabbit Alexa488 were purchased from Invitrogen (Carlsbad, CA, USA). A CellTrace™ Violet (CTV) Cell Proliferation Kit was purchased from Thermo Fisher Scientific (Waltham, MA, USA). OVA peptide fragments (323–339 and 257–264) were purchased from GeneScript (San Francisco, CA, USA). Flt3L-Ig was purchased from Bio-X-Cell (West Lebanon, NH, USA). CD45R (B220) MicroBeads was purchased from Miltenyi Biotec. (Bergisch Gladbach, Germany).

B16F10 (CRL-6475) cell lines were purchased from ATCC. A stable B16F10 cell line expressing membrane-bound OVA (OVA⁺B16F10) was produced by transient transfection with pCL-neo-mOVA (Addgene, Cambridge, MA) using Lipofectamine 2000 reagent (Invitrogen) and selection with G418 (InvivoGen, San Diego, CA, USA). For BMDCs cultures, 5 × 10⁶ BM cells were cultured in 10 mL of RPMI supplemented with 20 ng/mL recombinant murine GM-CSF for 7 to 9 days. GM-CSF was added every 3 days. To generate cDC1s, 3 × 10⁶ BM cells were incubated in 3 mL of RPMI supplemented with 200 ng/ mL Flt3-L for 9 days. Flt3-L was added every 2 days and cDC1s (CD11c⁺B220⁻) were isolated by anti-B220 positive selection beads to exclude plasmacytoid DCs (CD11C⁺B220⁺) for further experiments. However, unless otherwise indicated (for Additional file 1: Figs. S1 and 2), we used GM-CSF-induced BMDCs for most of the experiments. Naive CD4⁺ T cells were purified from the mouse spleen and LNs by negative selection using an EasySep magnetic separation system (Stemcell Technologies, Vancouver, Canada). To generate mouse T cell blasts, OTII CD4⁺ T cells were incubated in 2 µg/mL anti-CD3/28-coated culture plates with 100 U/mL rIL-2 for 48 h and cultured further for 3 days with 100 U/mL rIL-2.

C57BL/6 wi mice and OTII TCR transgenic mice (C57BL/6 background) were purchased from Damul Science (Korea) and Jackson Laboratories (Bar Harbor, ME, USA), respectively. All mice were housed under specific pathogen-free conditions. Transgelin-2 (Tagln2^−/−) KO mice have been described previously [16]. All experimental methods and protocols were approved by the Institutional Animal Care and Use Committee of the School of Life Sciences, Gwangju Institute of Science and Technology, and carried out in accordance with their approved guidelines (IACUC GIST-2015–04).

WT or Tagln2^−/− BMDCs (1 × 10⁶) were activated with 200 ng/mL of lipopolysaccharide (LPS), harvested, and blocked with a rat anti-mouse CD16/CD32 antibody (mouse Fc Block, BD Pharmingen). Cells were then stained with activation markers, including CD11c, CD80, CD86, CD40, and MHC-II, for flow cytometry. To examine T cell-mediated DC activation in vitro, 1 μg/mL of pOVA (323–339)-pulsed WT or Tagln2^−/− BMDCs (1 × 10⁵) was co-cultured with OTII CD4⁺ T cells (5 × 10⁵) for 24 h, and the supernatants were subjected to ELISA assay to examine cytokine production.

BMDCs were plated on coverslips coated with or without 10 μg/mL of Fn for 90 min. The cells were fixed for 10 min with 4% paraformaldehyde and permeabilized with 0.1% Triton-X (Sigma-Aldrich) in PBS for 10 min at room temperature (RT). The coverslips were then incubated with TRITC-conjugated phalloidin at RT for 30 min, washed, mounted onto slide glass using Vectashield (VectorLabs, Burlingame, CA), and imaged using a FV-1000 confocal microscope (Olympus, Tokyo, Japan) To measure cell spreading area, the captured images were analyzed using ImageJ software (NIH) as follows: threshold values were set to define the cell edge, and a mask was then created for each cell to get the total cell area (with arbitrary units) within the mask. Cell size was determined by flow cytometry after detachment of the BMDCs with 10% EDTA.

OTII CD4⁺ or OTI CD8⁺ T cells were stained with Cell Tracker Green CMFDA, and WT or Tagln2^−/− BMDCs or Flt3L-induced cDC1s were stained with Cell Tracker Orange CMRA for 30 min. The cells were then washed and resuspended in RPMI 1640 media. For conjugation, DCs were incubated with T cells (1:5 ratio) for 2 h in the presence or absence of pOVA (323–339, 1 μg/mL) or pOVA (257–264, 1 μg/mL). The relative proportion of green, orange, and green and orange-positive events in each tube was determined by FACS Canto (BD Biosciences, San Jose, CA) and analyzed with FlowJo software (Treestar, San Carlos, CA). The number of gated events counted per sample was at least 10,000. The percentage of conjugated T cells was determined as the number of dual-labeled (green and orange-positive) events divided by the number of green-positive T cells.

To examine in vitro T-cell activation, the indicated concentration of pOVA (323–339, 10⁻³–10¹ μg/mL)-pulsed WT or Tagln2^−/− BMDCs (1 × 10⁵) was co-cultured with OTII CD4⁺ T cells (5 × 10⁵) for 24 h, and the supernatants were subjected to ELISA assay to determine cytokine secretion. In addition, the cells were stained with anti-CD69 or CD25 to determine DC activation. To determine the effects of dU-TG2P, BMDCs or Flt3L-induced cDC1s (1 × 10⁵) were treated with dU-TG2P along with pOVA (323–339, 1 μg/mL) or pOVA (257–265, 1 μg/mL) for 2 h and co-cultured with OTII CD4⁺ T cells (5 × 10⁵) or OTI CD8⁺ T cells (5 × 10⁵) for 24 h, and the supernatants were subjected to ELISA assay to determine cytokine secretion. For in vivo T-cell priming, WT or Tagln2^−/− BMDCs (3 × 10⁶) were pulsed with pOVA (323–339, 1 μg/mL) for 2 h and injected s.c. into the footpads of OTII mice, and the cells were isolated from the popliteal lymph node at 24 h post-injection. In general, to stain the cells with fluorescently conjugated antibodies, the cells were blocked with anti-FcγR antibody (CD16/32, clone 2.4G2) and then stained for surface activation markers. Data were acquired using a FACS Canto and analyzed with FlowJo software.

Transwell cell migration was assayed using a 96-well Boyden chamber (ChemoTx plate, Neuroprobe, Inc., Gaithersburg, MD) according to the manufacturer’s instructions. The Boyden chamber was assembled with polyvinylpyrrolidone-free polycarbonate filters (3–5-µm pore size). WT or Tagln2^−/− BMDCs (1 × 10⁶) were added to the Fn-coated upper compartment, and media containing 200 ng/mL of each chemokine were added to the lower compartment. The apparatus was incubated for 4 h at 37 °C in a humidified CO₂ incubator. Cells on the bottom wells were harvested and resuspended in 300 µL of PBS, and the number of cells was counted using a FACS Canto for a fixed period of time (300 s) under constant middle pressure.

To evaluate DC migration, a mixture of 2 × 10⁶ WT (CMFDA-green) and 2 × 10⁶ Tagln2^−/− BMDCs (CMRA-orange) was injected into the footpads of WT recipient mice. A popliteal lymph node was harvested at 24 h post-injection. The popliteal LNs were fixed in 4% paraformaldehyde in PBS at 4 °C overnight. On the next day, the samples were washed and incubated in PBS with 30% sucrose (w/v) (Sigma-Aldrich) overnight at 4 °C. The samples were then embedded in Tissue-Tek® O.C.T. Compound (Thermo Fisher Scientific) and frozen using 2-methylbutane, cooled with liquid nitrogen. 10 μm sections were cut using a Leica CM1800 cryostat. For immunostaining, tissue sections were blocked for 2 h at RT in 10% normal goat serum (Sigma-Aldrich). The sections were incubated with fluorescently conjugated anti-B220 antibody at RT for 30 min in 10% goat normal serum. The samples were washed three times to remove unbound antibody and mounted in Permount solution (Thermo Fisher Scientific). Images were acquired with a confocal microscope and analyzed with Fluoview software. To quantify the number of migratory DCs, single-cell suspensions from the draining popliteal LNs were obtained by digestion in collagenase D, and the % of migrating DCs was quantified using FACS Canto. In some experiments, the excised lymph node was photographed for size determination. Draining popliteal LNs were harvested from the left hind limb, which were injected with cells through the footpad, whereas nondraining LNs were excised from the right hind limb.

To analyze the podosome dynamics, BMDCs were harvested and seeded on poly-L-lysine-coated glass coverslips in a 12-well plate (2 × 10⁶ cells/well in supplemented medium) and incubated for the indicated time in the presence of 200 ng/mL of LPS at 37 °C and 5% CO₂. To analyze the localization of transgelin-2, BMDCs were incubated for 24 h in the same condition. Coverslips were washed once with warmed PBS, followed by fixation with 4% paraformaldehyde in PBS at RT for 10 min. After permeabilization using 0.1% Triton X-100, the cells were stained with 10 µg of anti-vinculin and anti-transgelin-2 antibodies at 4 °C overnight. The next day, the coverslips were washed with PBS two times, and the cells were stained with mouse anti-Alexa647 and rabbit anti-rabbit Alexa488 secondary antibodies. For actin staining, permeabilized cells were incubated with anti-Alexa 647- or TRITC-conjugated phalloidin (1:100) for 1 h at RT. The coverslips were mounted onto slideglass using Vectashield (Vector Labs) and imaged using a confocal microscope. For quantitation of cells with podosomes, the cells were assessed for the presence of at least one clearly identifiable podosome with an F-actin-rich core. At least 100 cells were scored per sample, with a minimum of three biological replicates.

For in vivo imaging, pOVA (323–339, 1 µg/mL)-pulsed WT or Tagln2^−/− BMDCs (3 × 10⁶) were stained with Cell Tracker CMFDA-green and injected s.c. into the footpads of WT recipient mice. After 24 h, purified OTII CD4⁺ T cells were stained with CMRA-orange and adoptively transferred to recipient mice intravenously. Mice were anesthetized with isoflurane, and the popliteal LNs were surgically exposed. Imaging was performed on a Zeiss LSM 880 microscope equipped with a MaiTai laser (Coherent) tuned to 750 nm in combination with an NDD2 BIG2 GaAsP detector and a 20 × water-dipping lens (NA 1.0, Zeiss) using ZEN v2.1 acquisition software. Images were collected with a typical voxel size of 0.593 × 0.593 × 1.0 μm and a volume dimension of 607.28 × 607.28 × 200 μm to create a three-dimensional data set. For four-dimensional data sets, images were collected with a typical voxel size of 0.45 × 0.45 × 1.5 μm and a volume dimension of 425.1 × 425.1 × 30 μm. This volume collection was repeated every 60 s for up to 2 h. To analyze the number of DC-contacted T cells and the speed, displacement, and duration of these interactions, tracks were generated for T cells and analyzed using Imaris software. Data were plotted using Prism (GraphPad).

For the dynamic analysis of DC spreading and protrusion, time-lapse imaging was conducted on an EVOS system (EVOS™ FL Digital Inverted Fluorescence Microscope, Fisher Scientific, Paisley, Scotland, UK). WT or Tagln2^−/− BMDCs (3 × 10⁵) were seeded on 10 µg/mL of Fn-coated 12-well non-tissue culture plates for 10 min at 37 °C, and the plates were immediately placed in the chamber of the EVOS unit (which was programmed to supply 5% CO₂ and maintained at 37 °C constant temperature). The cells were recorded in the presence or absence of 200 ng/mL CCL19 every 10 s for up to 1 h. Sequential images were analyzed using ImageJ software.

For scanning EM, cells were fixed with 2.5% glutaraldehyde solution for 2 h, rinsed with PBS for 5 min, and fixed in OsO₄ for 2 h. The samples were then dehydrated through incubation with a graded ethanol series over 30 min and dried in a critical point dryer. The samples were prepared by sputter coating with 1–2 nm gold–palladium and analyzed using FE-SEM (HITACHI, Tokyo, Japan).

To evaluate transgelin-2 functions in T cells for tumor suppression (for Fig. 1b), PBS (none), WT (OTI-T), KO (Tagln2^−/− OTI-T), or transgelin-2-overexpressing T (TG2OE OTI-T) cells were adoptively transferred into tumor-bearing C57BL/6 mice at days 7, 10, and 13 post-tumor implantation. All mice were sacrificed at day 25, and tumors were isolated and weighed.

To evaluate the effect of Tagln2^−/− in mice (for Fig. 1c), OVA⁺B16F10 cells (3 × 10⁵) were s.c. injected into the dorsal flank region of age- and sex-matched WT or Tagln2^−/− mice. To evaluate transgelin-2 functions in DCs for tumor suppression, BMDCs (1 × 10⁷) from WT or Tagln2^−/− were pulsed with 1 µg/mL of pOVA for 2 h and i.v. injected into 8-week-old WT mice. At day 7, OVA⁺B16F10 cells (3 × 10⁵) were s.c. or i.v. injected into the dorsal flank region to induce solid and metastatic tumor models, respectively. Mice were sacrificed at day 8 post-inoculation with tumor cells. At the end of the experiments, tumors were isolated, weighed, and photographed for gross morphology. To analyze TILs, tumor tissues were dissected and mechanically disaggregated before digestion with collagenase D (1 mg/mL, Roche) for 30 min at 37 °C. After digestion, all of the cells were passed through 70-µm filters, and leukocytes were isolated by centrifugation using 38% Percoll for 30 min. Pellets were resuspended with PBS, stained with anti-CD3 and CD8 antibodies, and analyzed by flow cytometry. To analyze cytokine production, isolated TILs were stimulated with PMA/Ionomycin (200 nM/ 1 µM) in the presence of Brefeldin A (1 µg/mL) for 4 h at 37 °C. Cells were subsequently collected and stained for CD8 followed by fixation with IC fixation Buffer (eBioscience) for 20 to 30 min at RT. Then, the cells were washed twice with 1 × Permeabilization Buffer (eBioscience) and stained with IFN-γ antibody. After washing, cells were analyzed by flow cytometry.

To measure proliferation and cytokine production from OVA-specific T cells, DCs from WT or Tagln2^−/− (1 × 10⁷) were pulsed with 1 µg/mL of pOVA (257–265) for 2 h and i.v. injected into 8-week-old C57BL/6 WT mice. At day 7, CD3⁺ T cells were isolated from the LNs and spleens, stained with CTV, and co-cultured with 1 µg/mL of pOVA (257–265)-pulsed WT BMDCs for 4 days for clonal expansion. The proliferative CTV-positive cells that were in a live cell gate were quantified by flow cytometry. To measure the IL-2 or IFN-γ cytokines from OVA-specific T cells, co-cultured supernatants were harvested at 24 h and subjected to ELISA.

Recombinant transgelin-2 protein fused with PTD (TG2P) was described previously [17]. To generate a de-ubiquitinated mutant of TG2P, a potential ubiquitinated amino acid residue was predicted using the program UbiSite [33], and the resulting residue, K78, was mutated to an R by site-directed mutagenesis. The resulting PCR-amplified cDNAs encoding dU-TG2P were ligated into the pET-28a vector (Novagen, Madison, WI). Expression of recombinant TG2P and dU-TG2P in Escherichia coli BL21 (DE3) cells has been described previously [17].

DCs were washed and incubated with 10 μM of TG2P or dU-TG2P for 2 h at 37 °C in serum-free media. The cells were then washed with PBS and resuspended in media for further assays. To rule out the potential for LPS contamination, dU-TG2P was heat-inactivated by boiling for 5 min.

OTII CD4⁺ or OTI CD8⁺ T cells were stained with CTV and co-cultured with 1 µg/mL of pOVA (323–339)-pulsed BMDCs or pOVA (257–264)-pulsed cDC1s (1 × 10⁵) for 2, 3 or 5 days. Total cells were fixed with Cell Fixation & Cell Permeabilization Kit (Thermo Fisher Scientific) for 1 h at 4 °C and stained with anti-Ki67 along with 1 μL of NucSpot Far-Red for 1 h at 4 °C in the dark. The samples were washed with PBS and analyzed by flow cytometry. The percent of proliferative populations was acquired from the gate in a CTV-positive population.