Introduction

T cell receptor (TCR) stimulation by antigen bound to major histocompatibility complex (MHC) molecules on the surface of an antigen-presenting cell (APC) induces the formation of the immune synapse (IS). The IS is a specialized cell-cell interface contact area that provides a signalling platform for integration of signals leading to intercellular information exchange, in order to ensure efficient TCR signal transduction, T cell activation and the proper execution of diverse T lymphocyte effector functions (1). Comprised among these functions, IS formation triggers the convergence of T lymphocyte secretory vesicles, including multivesicular bodies (MVB) (2) (3), towards the microtubule-organizing center (MTOC) and the polarization of MTOC together with secretory vesicles to the IS (4, 5). The canonical bullseye structure of a mature IS comprises three concentric areas referred to as supramolecular activation clusters (SMACs) that can be discerned based on the specific segregation of TCR, integrins and costimulatory molecules (6). The central supramolecular activation cluster (cSMAC) is enriched in ligand-bound TCR and associated signalling molecules, the peripheral SMAC (pSMAC) in adhesion molecules such as LFA-1, and the outer distal SMAC (dSMAC) in filamentous actin (F-actin) and CD45 (7) (8). IS assembly and subsequent secretory responses are coordinated by both actin and microtubule cytoskeletons, whose interplay regulates the polarized secretory response towards the IS (9) (7). IS formation is associated with an initial increase in cortical actin at the IS (10), followed by a decrease in cortical actin density at the central region of the immune synapse (cIS) including the cSMAC, that contains the secretory domain (11) (12). The partial depletion of F-actin at the cIS has been proposed, apart from allowing the focusing and docking of secretory vesicles for secretion (10), to initiate key subsequent events leading to MTOC polarization and delivery of secretory vesicles, such as lytic granules in cytotoxic T lymphocytes (CTL) (12) and cytokine-containing secretory vesicles in T-helper (Th) lymphocytes (13), to the secretory domain of the IS. The CTL degranulation of diverse secretory lytic granules with MVB structure results in the secretion to the synaptic cleft of Fas ligand (FasL)-containing exosomes (14) (15) (16) (17) (2) (18). These exosomes, along with perforin and granzymes, which are secreted in both soluble and nanoparticulate form in so-called supramolecular attack particles (SMAPs) (19) (20), lead to the induction of target cell apoptosis. In addition, FasL-containing exosomes have been involved in autocrine FasL/Fas dependent, activation-induced cell death (AICD) produced upon TCR triggering (16) (21) (17) (22), an important immunoregulatory event involved in the downregulation of T cell immune responses (23) (24).

We have previously described that cortical actin reorganization at the IS plays an important role in MVB polarized traffic leading to exosome secretion in Th lymphocytes (25) (26) (3). With respect to the molecular cues controlling this process, we have shown that TCR-stimulated protein kinase C δ (PKCδ) regulates cortical actin reorganization at the IS, thereby controlling MTOC/MVB polarization and ultimately leading to exosome secretion at the IS and AICD in Th lymphocytes (25). Moreover, PKCδ is necessary for the polarization of lytic granules and the induction of cytotoxicity by mouse CTL (27, 28), which supports a general role of PKCδ in secretory traffic leading to apoptosis in T lymphocytes. Besides PKCδ, several actin cytoskeleton regulators, including the formin FMNL1 and Diaphanous-1 (Dia1), also regulate MTOC polarization (29) (30) (31). In this context, we have shown both PKCδ-dependent phosphorylation of FMNL1 at the IS and PKCδ-dependent F-actin clearing at the cIS, which participate in MTOC/MVB polarization leading to exosome secretion in CD4⁺ Jurkat T lymphocytes forming IS (25) (26) (3). However, a formal connection between FMNL1 phosphorylation, F-actin regulation and MVB polarization/exosome secretion at the IS has not been established yet.

Regarding potential formin regulatory pathways, we have shown that FMNL1, but not Dia1, is strongly phosphorylated in T lymphocytes by PKCδ activators such as phorbol myristate acetate (PMA), but also upon TCR stimulation via an anti-TCR agonist as well as IS formation (26). FMNL1 phosphorylation was inhibited in PKCδ-interfered T lymphocytes (26), which is associated with a deficient MTOC polarization towards the IS (25), supporting a PKCδ role in both FMNL1 phosphorylation and MTOC polarization (25) (26). Interestingly FMNL2, a related formin which exhibits a high homology with FMNL1, is phosphorylated by PKCα and, to a lower extent, by PKCδ, at S1072 (32). This phosphorylation reverses FMNL2 autoinhibition mediated by interaction of N-terminal Diaphanous inhibitory domain (DID) with the C-terminal Diaphanous autoinhibitory domain (DAD), resulting in increased F-actin assembly, β1-integrin endocytosis, invasive motility (32) and filopodia elongation (33). In this regard, recent evidence suggests that the mutation of a particular residue in FMNL2 impacting the DID-DAD interaction may result in functional implications for the characteristic actin-regulating activity of FMNL2, specifically in the context of podosome formation in macrophages (34). Out of the three FMNL1 isoforms (α, β and γ) present in T lymphocytes and Jurkat cells (35), S1086 in FMNL1β is surrounded by a sequence displaying high homology to the one surrounding S1072 in FMNL2 (32) (26) (Fig.1). Moreover, we have shown that FMNL1β is the only FMNL1 isoform phosphorylated upon IS formation and capable of recovering MTOC polarization when it was reexpressed in cells simultaneously interfered for the three FMNL1 isoforms (26). Thus, we hypothesize that IS-induced, PKCô-dependent phosphorylation at S1086, a residue located within FMNL1P DAD, may release the autoinhibition mediated by DID/DAD-interaction, thereby leading to the activation of FMNL1β. Consequently, this activation would mediate the F-actin reorganization at the IS and the MTOC/MVB polarization, thereby enabling the subsequent exosome secretion. Here, we report that FMNL1β phosphorylation at S1086 regulates F-actin reorganization leading to MTOC/MVB polarization and exosome secretion.

Figure 1

Results

FMNL1 interference and YFP-FMNLip variants expression

According to our previous data (26), FMNL1β plays a crucial role in MTOC polarization towards the IS. Moreover, we have also shown that FMNL1β is strongly phosphorylated in T lymphocytes by PKCδ activators such as PMA, as well as by TCR stimulation and upon IS formation, which could potentially be related to FMNL1β activation (26). As previously stated, FMNL2 becomes active upon PKCα phosphorylation at S1072, enhancing F-actin assembly (32) (33). In FMNL1β, Si086 in arginine-rich DAD (labelled in Fig. 1) is surrounded by a sequence displaying high similarity to that around S1072 in FMNL2 (36) (32) (26) (Fig. 1A). Thus, we hypothesized that phosphorylation of FMNL1β at S1086 may release DID-DAD auto-inhibition, thereby activating FMNL1β and regulating MTOC polarization. To address this point, we introduced S1086A and S1086D mutations in the shFMNL1-HA-YFP-FMNL1β construct (35) (26), which act as non-phosphorylatable and phosphomimetic residues, respectively (Fig. 1B). These resulting bi-cistronic vectors produce both FMNL1 interference and the corresponding YFP-FMNL1β variant expression (35), acting as useful tools to analyse the role of FMNL1β in IS formation during transient expression experiments (26). To confirm the previously described efficacy of these vectors (33) (26), we first analysed endogenous FMNL1 interference and YFP-FMNL1β variants expression in C3 Jurkat clone transfected with the different vectors at the single cell level by immunofluorescence, using an anti-FMNL1 antibody that recognizes all its isoforms. This analysis was performed in transfected Jurkat cells forming synapses, to rule out any potential effect of IS formation and subsequent T lymphocyte activation on both endogenous FMNL1 interference and YFP-FMNL1β variants expression. As seen in Fig. 2A, C3 cells transfected with the FMNL1 interfering plasmid shFMNL1-YFP (YFP⁺ cells) (second row), showed very low anti-FMNL1 signal when compared with non-transfected cells in the same preparation or control YFP^– cells (first row). In contrast, cells transfected with YFP-FMNL1βWT, YFP-FMNL1βS1086A or YFP-FMNL1βS1086D (rows three to five, respectively), showed higher anti-FMNL1 fluorescence signals when compared to nontransfected cells in the same microscopy field (Fig. 2A). The Suppl. Fig. S1B shows, by single cell image analysis, that the mean fluorescence intensity (MFI) of the anti-FMNL1 in shFMNL1-YFP-expressing cells is extremely low when compared to the control YFP^– group, whereas it is much higher in YFP-FMNL1βWT, YFP-FMNL1βS1086A or YFP-FMNL1βS1086D-expressing cells. As shown in Suppl. Fig. S1A, anti-FMNL1 MFI correlated with YFP-FMNL1β MFI in YFP-FMNL1βWT, YFP-FMNL1βS1086A or YFP-FMNL1βS1086D-expressing cells. To further verify this data, lysates from bulk populations of transfected cells were analysed by western blot (WB) using the same anti-FMNL1 antibody. The apparent molecular weights (MW) of the bands observed in the WB analysis were compatible with the predicted MW of endogenous FMNL1 (150 kDa) and the chimeric YFP-FMNL1β variants (150 + 30 kDa) (Fig. 2B). The endogenous FMNL1 (150 kDa band) levels in lysates from cells transfected with YFP-FMNL1βWT, YFP-FMNL1βS1086A or YFP-FMNL1βS1086D were lower than in control untransfected cells (Fig. 2B). The reduction of endogenous FMNL1 expression, as assessed by WB (between 40–80% reduction relative to control YFP^– cells), was apparently lower than that observed by single cell imaging analysis (compare panels A and B in Fig. 2 with Suppl. Fig. S1B). This is not surprising, since the transfection efficiency was relatively low (20–50%) and thus cell populations analysed by WB contained transfected as well as untransfected cells.

Figure 2

YFP-FMNL1PWT is phosphorylated upon PKC activation but neither YFP-FMNL1PS1086A nor YFP-S1086D are.

Once the different YFP-FMNL1β variants were expressed, we analysed the ability of a PKCδ activator to induce their phosphorylation. To this end, we stimulated C3 clone transfected with the different constructs with the PKC8 activator PMA, which has been shown to induce strong FMNL1 phosphorylation (26). Subsequently, we immunoprecipitated FMNL1 with an anti-FMNL1 that recognises all FMNL1 isoforms (26) and analysed these immunoprecipitates (IPs) by WB with anti-phospho-Ser PKC substrate (32) (Fig. 3). In all the IPs from PMA-stimulated cells, anti-phospho-Ser PKC substrate detected a band corresponding to endogenous FMNL1 (150 kDa), which was 8- to 30-fold more intense than in IPs from unstimulated cells (Fig. 3, lower graph). In IPs from PMA-stimulated cells expressing YFP-FMNL1βWT, anti-phospho-Ser PKC substrate detected an additional band of the predicted MW (150 + 30 kDa), which was also recognized by anti-FMNL1 (Fig. 3). However, in IPs from PMA-stimulated cells expressing YFP-FMNL1βS1086A or S1086D no apparent phosphorylation of these FMNL1β chimeras could be detected (Fig. 3). Thus, substitution of S1086 by a non-phosphorylatable or a phosphomimetic residue fully abolished PKC8 activator-induced phosphorylation of FMNL1β, as detected by anti-phospho-Ser PKC substrate, confirming our hypothesis that S1086 is indeed the residue that is phosphorylated in FMNL1β upon PKC activation.

Figure 3

S1086 phosphorylation of FMNL1β is crucial for MTOC/MVB polarization towards the IS

Next, we wanted to assess whether S1086 phosphorylation was related to FMNL1β regulation of MTOC/MVB polarization towards the IS. To this end, untransfected (control YFP^–) C3 cells, or transfected with the plasmids interfering all FMNL1 isoforms and expressing YFP (shFMNL1-YFP), YFP-FMNL1βWT, YFP-FMNL1βS1086A, or YFP-FMNL1βS1086D (35) were challenged with SEE-pulsed Raji cells (blue). Next, we quantified MTOC/MVB polarization index (PI) in fixed synapses formed by the mentioned cell groups by epifluorescence microscopy and deconvolution, as previously described (37) (25) (26) (Suppl. Fig. S2A, Fig. 4). In the first series of experiments, average MTOC PI of untransfected (Control YFP^–) C3 cells forming synapses with unpulsed Raji cells was significantly lower than with SEE-pulsed Raji cells, and similar to MTOC PI of FMNL1-interfered cells (shFMNL1-YFP) forming synapses with SEE-pulsed Raji cells (Suppl. Fig. S2B). This confirms that antigenic stimulation at the IS indeed triggers MTOC polarization and validates our synapse model measurements. In addition, confirming our previous data (26), MTOC/MVB polarization was disrupted upon FMNL1 interference to the levels obtained with unpulsed Raji cells (Fig. 4A, compare the first two rows, Fig. 4B, Suppl. Fig. S2B), and was restored by YFP-FMNL1βWT expression (Fig 4A, third row and 4B). Interestingly, non-phosphorylatable YFP-FMNL1βS1086A expression was unable to restore MVB/MTOC polarization to control levels (Fig. 4A, fourth row and 4B), whereas phosphomimetic YFP-FMNL1βS1086D expression restored MVB/MTOC polarization to the levels achieved by YFP-FMNL1βWT expression (Fig. 4A, fifth row and 4B). Comparable results were obtained by confocal microscopy (Suppl. Fig. S3). From confocal images, en face views of the IS interface were generated (Suppl. Fig. S4, S5 and Suppl. Videos 1 and 2). These images show the polarized MTOC or the polarized accumulation of MVB (measured as high MTOC and MVB PI values, in white types), respectively, in the cIS region of untransfected (control YFP^–) C3 cells or cells expressing YFP-FMNL1WT or YFP-FMNL1S1086D, but not in cells transfected with shFMNL1-YFP or expressing YFP-FMNL1βS1086A (low MTOC and MVB PI values). Thus, phosphorylation of S1086 in FMNL1β appears to be essential for secretory polarized traffic towards the IS in T lymphocytes. Moreover, throughout the experiment, we observed that MTOC and MVB were polarizing together towards the IS. This fact, together with the previous work regarding segregation between MTOC movement and secretory granules traffic (13) (28, 38, 39), prompted us to simultaneously analyse both MTOC and MVB positioning with respect to the IS at the single cell level. Remarkably, we observed a robust linear correlation between MVB and MTOC PIs both in untransfected (control YFP^–) C3 cells and cells transfected with shFMNL1-YFP (Pearson’s linear correlation coefficients 0.96 and 0.95, respectively, Suppl. Fig. S6). Furthermore, it is noteworthy that MVB and MTOC centers of mass (MVB^C and MTOC^C, respectively) are very closely located in all the studied cell groups, regardless of polarization state (Fig. 4, Suppl. Fig. S3). Although the MTOC and MVB did not efficiently polarize in cells transfected with shFMNL1-YFP, their MVB^C were still located next to MTOC^C (Fig. 4A, second row), and equivalent results were obtained in all the analysed cell groups (Fig. 4). Thus, phosphorylation of S1086 in FMNL1β appears to be essential for both MTOC and MVB polarization towards the IS in T lymphocytes.

Figure 4

YFP-FMNL1PS1086D phosphomimetic mutant expression does not rescue MTOC polarization in PKCô-interfered T lymphocytes.

The former results supported the contribution of FMNL1β and its phosphorylation at S1086 to MVB/MTOC polarization, but did not address the sufficiency in this process of PKCδ-controlled, S1086 FMNL1β phosphorylation. In this context, we have previously shown that FMNL1β lies downstream of PKCδ in the same pathway controlling MTOC/MVB polarization (26) (3). Thus, in order to analyse this important issue we transfected the PKCδ-interfered P5 and P6 clones with the different plasmids and compared the MTOC PI among the different cell groups, as previously done in C3 clone. We have previously shown that neither PKC8 nor FMNL1 interference affects IS conjugate formation, since P5 and P6 clones and FMNL1-interfered C3 clone formed IS with SEE-pulsed Raji cells (25) (26). In addition, IS formed by P5 clone with SEE-pulsed Raji cells (fifth group in Suppl. Fig. S2B) had a MTOC PI comparable to those of IS formed by C3 clone with unpulsed Raji cells or P5 clone with unpulsed Raji cells (first and fourth group, respectively, in Suppl. Fig S2B), supporting that both FMNL1 or PKC8 interference fully inhibited MTOC polarization.

In contrast to what was found in the control C3 clone (Fig. 4 and Suppl. Fig. S3) or C9 clone (not shown), YFP-FMNL1βWT expression in PKCδ and FMNLl-interfered P5 clone (Fig. 5) or P6 clone (not show) did not restore MTOC PI to the values observed in control YFP^– C3 cells, which is compatible with the idea that FMNL1β lies downstream of PKCδ and its phosphorylation is indispensable for MTOC polarization. Consistently, expression in P5 clone of YFP-FMNL1βS1086A, a variant non-phosphorylatable by PKC, was also unable to enhance MTOC PI to the values observed in control YFP^– C3 cells (Fig. 5B). Moreover, phosphomimetic YFP-FMNL1βS1086D expression, which restored MVB/MTOC PI to the values achieved by YFP-FMNL1βWT expression in the C3 clone (Fig. 4 and Suppl. Fig. S3), did not restore MTOC PI in PKCδ and FMNL1-interfered P5 cells to the levels observed in control YFP^– C3 cells (Fig. 5B). Similarly, FMNL1βS1086D expression was unable to revert the deficient MTOC polarization occurring in P6 (not show), a different PKCδ-interfered clone previously described (25). We have shown that the deficient MTOC/MVB polarization in the PKCδ-interfered P5 and P6 clones is due exclusively to the reduction in PKCδ expression, since transient reexpression of PKCδ in all these clones recovered MTOC/MVB polarization to control levels (25). Since both the FMNL1 interference and FMNL1 variant re-expression for IS experiments were performed in transient assays (2–4 days after transfection), there was no chance for any clonal variation in these short-time experiments. Taken together, these results show that although FMNL1β phosphorylation at S1086 is necessary, it does not seem to be sufficient for MTOC polarization, at least in cells lacking PKCδ.

Figure 5

FMNLip translocation to the immune synapse is independent of S1086 phosphorylation and PKCδ.

Previous results have shown that FMNL1, apart from being mainly located at the cytosol and centrosomal areas, was also found located at the IS in a small percentage of synaptic conjugates, in end-point IS experiments (29) (26). If confirmed, the possibility of an IS-induced FMNL1 translocation to the IS may provide the molecular basis underlying a potential regulatory effect of FMNL1 on cortical actin cytoskeleton reorganization at the IS as well as on MTOC/MVB and secretion granule polarization, occurring both in CTL and Th cells, as suggested by several authors (9, 12) (13) (40) (3). However, although it could be inferred (29) (41) that the active translocation of any FMNL1 isoform from cytosol towards the IS is induced upon IS formation, it has not been formally demonstrated yet (29) (41). This is probably due to the fact that most early studies have used an end-point approach that does not allow to analyse the incipient IS (29) (42) (43) (44). Since FMNL1β is responsible for MTOC/MVB polarization to the IS (44), we analysed FMNL1β subcellular location in developing synapses by time-lapse, epifluorescence microscopy (43) (44). For this analysis, FMNLi-interfered, YFP-FMNL1βWT-expressing C3 clone was challenged with SEE-pulsed Raji cells. As evidenced in the example shown in Suppl. Video 3 (upper panel) and Fig. 6A (upper row), a transient relocation of initially cytosolic YFP-FMNL1βWT to the IS (white arrow) was observed in these cells. The average (±SD) duration of YFP-FMNL1βWT accumulation at the IS in C3 clone was around 6 min (6 min 18 s ± 1 min 34 s, n=6 synapses). Considering the major role of PKCα in FMNL2 relocation from the plasma membrane to the endosomal membrane (32), the high similarity between FMNL2 and FMNL1β C-terminal regions (Fig. 1) and the reported PKCδ-mediated FMNL1β phosphorylation (26), we next aimed to analyse the role of PKCδ in YFP-FMNL1βWT translocation to the IS. To this end we analysed YFP-FMNL1βWT translocation upon IS formation in the PKCδ-interfered P5 clone (25) (Suppl. Video 3 and Fig. 6A). The results in P5 clone showed a clear translocation of YFP-FMNL1WT to the IS (white arrow), with an accumulation of around 8 minutes at the IS (8 min 55 s ± 3 min 2 s, n=6 synapses, not significant differences with the C3 clone), so YFP-FMNL1βWT translocation to the IS in P5 and C3 clones showed comparable kinetics.

Figure 6

Moreover, to directly analyse the specific involvement of S1086 phosphorylation in this process, we challenged FMNLi-interfered, YFP-FMNL1βS1086A and YFP-FMNL1βS1086D-expressing C3 cells with SEE-pulsed Raji cells. As shown in Suppl. Video 4 and Suppl. Fig. S8, both YFP-FMNL1βS1086A and YFP-FMNL1βS1086D transiently relocated to the IS. The average time of YFP-FMNL1βS1086A accumulation at the IS was around 7 min (7 min 30 s ± i min 58 s, n=6 synapses), whereas YFP-FMNL1βS1086D showed an average accumulation time of around 6 min (6 min 52 s ± 2 min 9 s, n=7 synapses; not significant differences between YFP-FMNL1βS1086A and YFP-FMNL1βS1086D, nor between any of the mutants and YFP-FMNL1βWT). Therefore, these results confirm that the translocation and accumulation of FMNL1β in the IS region of Jurkat cells is a process that occurs independently of both PKCδ and S1086 FMNL1β phosphorylation.

A more detailed study of subsynaptic location of endogenous FMNL1 in C3 control clone using fixed synapses demonstrated that FMNL1 accumulated at the cIS but also at the edges of the synapse, around the dSMAC, both in C3 control clone (Fig. 6B, first and second rows, respectively) and in P5 PKCδ-interfered clone (Fig. 6B, third and fourth rows). At these subsynaptic locations, FMNL1 colocalized with F-actin (white pixels), which is compatible with the idea that FMNL1 regulates F-actin at the IS. To extend these results obtained in Jurkat cells we have performed experiments using primary T cells developing synapses mediated by TCR recognizing SEE plus SEB superantigens and a dual chimeric antigen receptor (CAR) recognizing CD19/CD22 on Raji cells (Suppl. Fig. S7A and B, respectively). As shown in this figure, endogenous FMNL1 colocalizing with F-actin at the IS was also found in both types of synapses developed by primary T lymphocytes.

Subsequently, we studied the subsynaptic locations of YFP-FMNL1βWT, YFP-FMNL1βS1086A, and YFP-FMNL1βS1086D in C3 control clone, to identify whether S1086 phosphorylation affected these locations. The results from fixed synapses, obtained through both epifluorescence (Suppl. Fig. S8 and S9) and confocal microscopy (Suppl. Fig. S10), revealed the presence of accumulations of the three YFP-FMNL1β variants in the IS. Additionally, in Suppl. Fig. S10, it can be observed that the three YFP-FMNL1β variants colocalized with F-actin at the IS.

Taken together, these observations show that, upon IS formation, FMNL1β translocates to the IS, with both PKCδ and S1086 phosphorylation being dispensable for the translocation to the IS of both endogenous FMNL1 and YFP-FMNL1β variants.

S1086 phosphorylation of FMNL1β is necessary for cortical F-actin reorganization during MTOC/MVB polarization

The observations that YFP-FMNL1β translocated to the IS, that cortical F-actin remodelling was considered to be sufficient for centrosome polarization in CTL (12), and that cortical F-actin remodelling was associated with MTOC and secretory granule polarization in CD4⁺ cells (13) (45), prompted us to study a potential role of FMNL1β and its phosphorylation at S1086 in synaptic F-actin architecture. To address this point, fixed synapses formed by the C3 clone transfected with the previously described plasmids were stained with phalloidin AF647 to label F-actin and imaged by confocal fluorescence microscopy. The relative area of the F-actin-low region at the cIS (Fact-low cIS area / IS area) in the IS interface, which mirrors cortical F-actin reorganization at the IS, was measured as indicated in Materials and Methods and represented for all cell groups. As seen in a representative example in Fig. 7A, the dot plot in Fig. 7B and Suppl. Video 5, the F-actin-low region at the cIS was smaller in FMNL1-interfered cells than in untransfected control YFP^– cells (Fig. 7A, second and first columns, respectively), as it was also in PKCδ-interfered cells (25) (26). Of note, YFP-FMNL1βWT expression restored F-actin depletion at the cIS to control levels (Fig. 7B). In contrast, YFP-FMNL1βS1086A expression did not rescue F-actin depletion to control levels, whereas YFP-FMNL1βS1086D did (Fig. 7B). Thus, FMNL1β phosphorylation at S1086 appears to be necessary for F-actin reorganization at the IS, as it was for MTOC/MVB polarization (Fig. 4).

Figure 7

Three-dimensional FMNL1β distribution at the synapse

We observed a positive effect of FMNL1β phosphorylation at S1086 on cortical F-actin rearrangement at the IS (former paragraph and Fig. 7) and YFP-FMNL1β translocated to the cIS, but also to the dSMAC (Suppl. Video 3 and 4, Fig. 6 and Suppl. Fig. S8, S9 and S10). To investigate this further, we analysed in more detail the subsynaptic distribution of the endogenous FMNL1, YFP-FMNL1βWT, and the YFP-FMNL1βS1086A and S1086D mutants with respect to the actin cytoskeleton. To this end, fixed synapses formed by the C3 clone transfected with the different plasmids were stained with phalloidin, anti-FMNL1 and anti-phospho-Ser PKC substrate antibody and imaged by confocal fluorescence microscopy (Suppl. Videos 6 and 7, Fig. 8 and Suppl. Fig. S11). Figs. 8A and 8B show representative examples of synaptic conjugates yx views (first row) and cropped synapse areas (white rectangles in first row) used to generate zx IS interfaces (second row), as shown in Suppl. Video 6 and 7. Subsequently, the colocalization pixels (white) along all the Z-stack at the IS interface (interface colocalization) were determined as indicated in Materials and Methods. This involved merging the indicated fluorescence channels (Fig. 8A, second row, F-actin in blue and anti-FMNL1 in red; Fig. 8B, second row, F-actin in blue and YFP fluorescence in green) on the IS interfaces of the synaptic areas generated as shown in Suppl. Video 6. MFI profile of each channel (F-actin in blue, anti-FMNL1 in red, YFP in green and colocalization in grey) along the indicated white arrows are shown below each IS interfaces. As shown in Fig. 8, and consistently with results from Fig. 7, both control YFP^– (first column) and YFP-FMNL1βWT-expressing cells (third and fourth columns) exhibited a wide F-actin depletion area at the cIS and an F-actin accumulation at the dSMAC. Interestingly, in some synapses from control YFP^– (first column) and YFP-FMNL1βWT-expressing cells (fourth column), F-actin and FMNL1 colocalized (white pixels) at certain small areas at the wide F-actin-low region at the cIS, although the MFI profiles show relatively less F-actin and FMNL1 at the cIS than at the dSMAC. In other synapses from both control YFP^– (not shown) and YFP-FMNL1βWT-expressing cells (third column), F-actin and FMNL1 colocalization occurred at the F-actin rich-area corresponding to the dSMAC (Fig. 8A), which was consistent with the data from Fig. 6B. In contrast, and consistently with Fig. 7, FMNL1-interfered cells (Fig. 8, second column) displayed small F-actin low areas at the cIS. Moreover, synapses from YFP-FMNL1βS1086A-expressing cells (fifth column) did not exhibit the F-actin depletion at the cIS, whereas synapses from the YFP-FMNL1βS1086D-expressing cells (sixth column) exhibited wide F-actin depletion at the cIS, comparable to those observed in the control YFP^– and in the YFP-FMNL1βWT-expressing cells (Fig. 8). In cells expressing the YFP-FMNL1β variants, colocalization of all the YFP-FMNL1β variants with F-actin was mainly in the F-actin low area at the cIS or at the dSMAC. When colocalization in all Z optical sections was assessed using directly the YFP fluorescence construction instead of anti-FMNL1 signal (Fig. 8B, second row, F-actin in blue and YFP fluorescence in green), comparable results were obtained. In addition, we analysed the subsynaptic location of FMNL1 with respect to Ser-phosphorylated PKC substrates by using an anti-FMNL1 and an anti-phospho-Ser PKC substrate. Both endogenous FMNL1 in control YFP^– cells (Suppl. Video 7 and Suppl. Fig. S11A, first column) and YFP-FMNL1βWT in YFP-FMNL1βWT⁺ cells, (Suppl. Fig. S11A, third column) strongly and specifically colocalized with anti-phosphoSer PKC substrate at the cIS. This is evidenced by the accumulation of white, colocalization pixels at low FMNL1 density areas located at the cIS, which is also an Factin-low area (Fig. 8, see MFI profiles for control YFP^– and YFP-FMNL1βWT⁺ cells). When colocalization at the synapse interface was assessed using directly the YFP fluorescence construction instead of anti-FMNL1 signal (Suppl. Fig. S11B second row, anti-phospho-Ser PKC substrate in blue and YFP construction in green and Suppl. Video 7) comparable results were obtained. In addition, as expected and as a negative control, no colocalization of FMNL1 and anti-phospho-Ser PKC substrate was observed in FMNL1-interfered cells (Suppl. Fig. S11 and Suppl. Video 7, second column). Moreover, FMNL1 and anti-phospho-Ser PKC also colocalized using super resolution STED microscopy, that greatly reduces xy resolution limits down to 30 nm (Suppl. Fig. S12). Although all these data do not yet allow us to infer that FMNL1β is phosphorylated at the IS due to the resolution limits of super resolution microscopy, the results are compatible with the idea that both endogenous FMNL1 and YFP-FMNL1βWT are specifically phosphorylated at the cIS.

Figure 8

S1086 phosphorylation of FMNL1β regulates exosome secretion

The aforementioned results support that FMNL1β and its phosphorylation at S1086 are involved in MTOC/MVB polarized traffic to the IS, most probably via control of F-actin rearrangement at the IS. MVB transport and fusion to the plasma membrane are necessary for exosome secretion (46) (22) and PKCδ controls F-actin depletion at the cIS (25), thus, it is conceivable that FMNL1β may control subsequent exosome secretion at the IS. To directly analyse FMNL1β contribution to exosome secretion, we used an exosome secretion reporter assay. We transiently cotransfected C3 clone with the exosome reporter GFP-CD63 expression plasmid (2) (47) and the different shFMNL1-YFP constructs. Subsequently, IS were formed between the cotransfected C3 clone and SEE-pulsed Raji cells, exosomes were purified from cell culture supernatants (2), lysed, and exosome lysates were analysed by anti-CD63 WB. This secretion assay excludes the detection of exosomes released by Raji cells in the coculture, that otherwise may mask the exosomes released by Jurkat cells (2) (22). In parallel, we determined nanovesicle concentrations in coculture supernatants by using Nanoparticle Tracking Analyses (NTA) (Fig. 9A). Anti-CD63 WB analysis of exosome secretion is shown in Fig. 9B. We normalized exosomal GFP-CD63 signal to the transfection efficiency of the exosome reporter (measured by GFP-CD63 signal from cell lysates) and the number of viable exosome-secreting cells (Fig. 9B). As shown in Fig. 9, interference with all FMNL1 isoforms not only reduced exosomal GFP-CD63 levels, but also reduced SEE-stimulated nanovesicle concentration increase. These results mirrored those obtained in cells in PKCδ-interfered cells (25). Moreover, interference with all FMNL1 isoforms and YFP-FMNL1βWT expression recovered exosome secretion to the levels produced by control YFP^– cells. However, YFP-FMNL1βS1086A expression did not recovered exosome secretion, whereas YFP-FMNL1βS1086D expression rescued exosome secretion to the values observed in control YFP^– or YFP-FMNL1βWT-expressing cells (Fig. 9B). These results demonstrate IS-induced exosome secretion, but do not reveal the presence of exosomes at the synaptic cleft. Therefore, we have performed STED super-resolution imaging of the immune synapses made by control and FMNL1-interfered cells. Nanosized (100–150 nm) CD63⁺ vesicles can be found in the synaptic cleft between the APC and control YFP^– cells with polarized MVB, whereas we could no detect these vesicles in the cleft from FMNL1-interfered cells that maintain unpolarized MVB (Fig. 10). Taken together, these results on exosome secretion endorse the outcomes regarding FMNL1 interference and FMNL1β-mediated control of cortical F-actin reorganization at the IS and their role on MVB polarized secretory traffic.

Figure 9

Figure 10

Discussion

Our previously published results suggest that PKCδ-dependent phosphorylation of FMNL1β may regulate its function in F-actin reorganization at the IS and hence affect MTOC polarization (25) (26). Here, we have identified for the first time a positive regulatory role of FMNL1β phosphorylation at S1086 in cortical F-actin regulation that governs MTOC/MVB polarization and exosome secretion at the IS in Th lymphocytes. Moreover, the fact that the phosphomimetic YFP-FMNL1βS1086D expression in PKCδ and FMNL1-interfered cells, did not recover the MTOC PI observed in C3 control cells (Fig. 5) supports the idea that FMNL1β phosphorylation at S1086 is necessary but not sufficient for MTOC polarization, at least in cells lacking PKCδ. Thus, these results support that another signal, apart from PKCδ activation, or another PKCδ-controlled pathway not involving FMNL1β phosphorylation, evoked upon TCR triggering at the IS, is also implicated in MTOC/MVB polarization to the IS.

Our results on FMNL1β phosphorylation are certainly related to what was found in formin homology domain protein 1 (FHOD1) regulation. FHOD1 is grouped into Dia-related formins (48). FHODi requires active Rac to be recruited to the plasma membrane, but its recruitment appears insufficient for FHODi activation, which is achieved only after being phosphorylated by Rho-dependent protein kinase (ROCK) at three serine and threonine sites included in a polybasic arginine-rich region inside the C-terminal DAD (49), leading to disruption of FHODi autoinhibitory state and F-actin stress fiber formation (49). This activation process is similar to what we have described in this article for FMNL1β. Moreover, the PKC-dependent mode of FMNL2 activation during integrin activation (32) and filopodia formation (33) is comparable to what we have found in FMNL1β during polarized traffic to the IS. Furthermore, FHOD3 formin is also autoinhibited by an intramolecular interaction between its C- and N-terminal domains. Phosphorylation of the three highly conserved residues (S1406, S1412, and T1416) within the polybasic, C-terminus of FHOD3 by ROCK1/2 is sufficient for its activation and is crucial in regulating myofibrillogenesis in cardiomyocytes (50). Thus, with the results provided here for FMNL1β, it has been reported that at least four different formins are activated through phosphorylation by three different kinases at the C-terminal polybasic regions, which are included in a common DAD. This underscores the evolutionary importance of this regulatory mechanism in formin activation, ultimately leading to very diverse cellular responses. However, although we do not have conclusive evidence supporting that PKC directly phosphorylates FMNL1β, the presence of a conserved S1086 within a RRSV/AR motif that contains a potential target for classical PKCδ (32, 33, 51), along with the recognition of the phosphorylated residue by a validated anti-Phospho-Ser PKC substrate antibody (32), certainly support this possibility.

With regards to potential regulatory pathways involved in formin activation and subcellular localization, it is remarkable that the most abundant formins in T lymphocytes, Dia1 and FMNL1 (29), are constitutively inactive in the cytoplasm due to intramolecular DAD-DID binding, which blocks formin ability to nucleate and elongate actin filaments (7). Regarding a possible control of formin subcellular location by PKC, it has been shown that both relocation of FMNL2 from the plasma membrane to the endosomal compartment and activation are dependent on C-terminal S1072 phosphorylation by PKCα (32). However, FMNL2 location in filopodia seems to be independent of the phosphorylation state (33). We have previously described PKCδ-dependent phosphorylation of FMNL1β (26), and here we have here demonstrated that this phosphorylation occurs at S1086, leading to F-actin reorganization at the IS, and resulting in MTOC polarization. However, in contrast to FMNL2 relocation, which is dependent on PKCα, we have shown that transient YFP-FMNL1βWT relocation from cytosol to the IS occurred in the absence of PKCδ (Fig. 6), in conditions where neither Factin reorganization (25) nor MTOC polarization (Fig. 5) occurred. Moreover, both YFP-FMNL1βS1086A and YFP-FMNL1βS1086D translocated to the IS (Suppl. Video 4 and Suppl. Fig. S8), demonstrating that S1086 phosphorylation is not required for IS location. However, as it happens during S1072 phosphorylation in FMNL2 (32), it is possible that blocking phosphorylation at S1086 through the introduction of single point mutations (YFP-FMNL1βS1086A or YFP-FMNL1βS1086D) may not block FMNL1β translocation to the IS per se, but may alter the kinetics of the process. For this reason, in future approaches, a thorough study on the kinetic considerations on S1086 PKCδ-mediated phosphorylation contribution to FMNL1β translocation to the IS will need to be addressed.

Jurkat T cells may not fully recapitulate the regulation of polarization in primary T cells, and the evidences from activating surfaces acting as artificial synapses show distinct actin cytoskeleton behavior (presence or not of F-actin foci) in primary versus immortalized T cells, at least in CD4⁺ cells, that may reflect mechanistic distinctions in the T-cell models systems utilized to study cytoskeletal actin dynamics (52) (53). However, to date, no clear evidence has been obtained regarding the existence of such an F-actin foci network in cell-cell immune synapse models based on primary T lymphocytes or Jurkat cells. F-actin foci have been exclusively observed in primary T cells engaged with artificial activating surfaces (planar lipid bilayers or coated glass), which does not apply to our data obtained from cell-cell synapses. However, to solve any potential differences between primary versus immortalized T cells, we have performed experiments analyzing the subcellular localization of FMNL1 in CD19/CD22 dual-targeted CAR T cells forming synapses with CD19⁺/CD22⁺ Raji cells used as CTL targets and T lymphoblasts forming synapses with SEE and SEB-pulsed Raji cells. We observed, in fixed-cell images, the presence of endogenous FMNL1 colocalizing with F-actin-enriched areas in the IS region of both CAR T cells and primary T lymphoblasts, which also exhibited MTOC polarization towards the IS (Suppl. Fig. S7). The presence of FMNL1 at the IS developed by primary T lymphoblasts was transient, as occurred in Jurkat cells (not shown). These observations extend our results on FMNL1 localization at the IS in the Jurkat-Raji plus SEE synapse model to the IS developed by primary human T lymphocytes. Regarding the effect of the diverse FMNL1β variants expression on MTOC/MVB polarization and exosome secretion in primary T lymphocytes, the data from activating surfaces clearly shows that the synaptic actin architecture of the IS from primary CD8⁺ T cells is essentially indistinguishable and thus unbiased from that of Jurkat T cells, but different to that of primary CD4⁺ cells (41). Thus, all our data in Jurkat T cells can be extrapolated to the synaptic architecture of primary CD8⁺ cells, although more experiments involving primary CD4⁺ are necessary to extend all our results to the last cell type.

Although the mechanism(s) involved in FMNL1β recruitment to the IS remains unclear, it is probably that it involves Rac1 and Cdc42 small G proteins (30) (29) (54), which are activated at the IS upon TCR triggering (55). However, we cannot exclude the participation of another PKC isoform or a different protein kinase that may induce FMNL1β phosphorylation in different residues upon previous FMNL1β relocation to the IS primed by small G protein, as was shown for FMNL2 relocation (32). Further experiments are needed to establish this important point.

Taken together, our results support that IS-induced, PKCδ-dependent phosphorylation of S1086 in FMNL1β DAD, most probably executed by PKCδ itself, activates FMNL1β, which in turn regulates cortical actin at the IS and hence controls MTOC/MVB polarization and exosome secretion. The behaviour of non-phosphorylatable YFP-FMNL1βS1086A (equivalent to FMNL2-S1072A described in (32)) confirms this hypothesis, since its expression inhibits F-actin reorganization at the IS, MTOC polarization and exosome secretion. In this respect, it has been reported that the phosphorylation of FMNL2 by PKCα in S1072 is necessary for the secretion of the prometastatic factor Angiopoietin-like 4 (ANGPTL4) in breast cancer cells (56), which constitutes the first description for an activated formin involvement in secretion during cell invasion. In addition, ANGPTL4 can be secreted into exosomes by lung cancer cells (57) (58) and FMNL2 controls exosome secretion (59). Thus, although it is not clear whether ANGPTL4 is released in exosomes or in soluble form (56), both FMNL2 and FMNL1β appear to control PKC-regulated exosome secretion. However, the extracellular signals evoking exosome secretion are strikingly different in the various cell types, since in tumor cells hypoxia and radiation can amplify the otherwise constitutive and multidirectional exosome secretion, whereas T lymphocytes constitutively secrete very low levels of exosomes and none in a polarized manner unless activated trough TCR or synaptic contact (22). Taken together, these considerations support a more general role of the formin-actin cytoskeleton axis in directional exosome secretion. Given all these considerations, it will be interesting to analyse the role of FMNL2 in inducible exosome secretion at the IS in T lymphocytes, but also the role of FMNL1β in exosome secretion by cancer cells during cell invasion. This is particularly relevant as exosome secretion is required for directionally persistent and efficient in vivo movement of cancer cells (60), and it is known that FMNL1 stimulates both leukemia cell proliferation and migration (54) (61).

The fact that FMNL1- and PKCδ-interfered cells have a similar phenotype to that of FMNL1- or PKCδ-mterfered cells (26) (25) (this paper), supports the notion that PKCδ and FMNL1β participate in the same F-actin regulatory pathway controlling MTOC/MVB polarization. Additional experiments involving in vitro measurements of F-actin reorganizing/severing activities of the diverse FMNL1β variants, phosphorylated or not in vitro by PKCδ, will be necessary to address whether PKCδ directly phosphorylates FMNL1β at S1086.

Our findings show that FMNL1 colocalizes with F-actin at the IS edges which are part of the dSMAC (Figs. 6, 8). This data are aligned with the previous observation that Dia1 and FMNL1 are intrinsically inactive because they undergo intramolecular folding and remain in the cytoplasm (7) and that after IS formation, they are enriched at the tips of F-actomyosin spikes at the outer edge of the dSMAC, giving rise to the actomyosin arcs at the pSMAC in the IS formed by Jurkat cells (41) (7). Dia1 is not a target of PKC, whereas FMNL1 is phosphorylated by PKC activation (26). Thus, in this paper we have focused on FMNL1. High resolution, spatio-temporal analysis of FMNL1β localization at the different subsynapic locations, together with superresolution techniques (41) (43) may provide new clues regarding the subsynaptic F-actin network targeted by phosphorylated FMNL1β. Even although the nature of the downstream effectors and consequences of FMNL1β phosphorylation on synaptic F-actin architecture can only be speculated at this point, concentric actomyosin arcs generated by FMNL1 propel TCR microclusters towards the IS and reinforce cytotoxicity (41) (7). Thus, high-resolution spatiotemporal analysis of actomyosin arcs structure and function during TCR microcluster movement in FMNL1-interfered cells, and/or cells expressing the FMNL1β mutants described here, will provide important clues on this important matter. It is noteworthy that a majority of the image analysis of IS organization and F-actin dynamics is based on planar synapses obtained using artificial APC substitutes, by either coating stimulatory and co-stimulatory molecules on glass or plastic surfaces or by embedding these molecules into lipid bilayers (62) (7) (63). These artificial synapses can indeed be imaged and analysed with the highest possible resolution (64) (41) (43). However, the cell-cell synapse model analysed here mimics the complex interactions and irregular, 3D stimulatory surface of a physiological synapse better than artificial synapses do (62) (7). In this important sense, the data presented here are indeed closer to a physiological situation. In addition, since actomyosin arcs arise from active, formin-dependent nucleation sites at the outer edge of the dSMAC (41), it will be interesting to analyse, using superresolution techniques, whether phosphorylated FMNL1β is located at these sites. Indeed, actomyosin arc network spanning from dSMAC along the pSMAC is the most likely source of T cellbased force that augments cytotoxicity by straining the target cell plasma membrane (7). Thus, it will be interesting to study the effect of FMNL1β mutants on actomyosin arcs formation and analyse functional deficiencies (i.e., decreased T cell-APC adhesion frequency, IS size and IS formation efficiency) in IS formed by cells interfered in FMNL1β and/or expressing the FMNL1β variants described here. Visualization of actin and actomyosin networks, TCR microclusters and secretion granules dynamics in “real” live T cell-APC conjugates at the superresolution level (7) is clearly needed to advance in this area.

Remarkably, although it has already been shown that FMNL1 is important for MTOC polarization in Jurkat cells and for target cell lysis in CTL (29), our data provide the first molecular basis underpinning this essential FMNL1 function. Here, we demonstrate that FMNL1β and its phosphorylation at S1086 participate in F-actin-controlled, polarized MVB traffic secreting effector exosomes (2, 17), which deliver proapoptotic signals to target cells (19) (65) (20). In addition, FMNL1 promotes proliferation and migration of leukemia cells (54) and mediates posterior perinuclear actin polymerization to promote T lymphocyte effector cell migration to inflammatory sites to enable T cell-mediated autoimmunity (61). PKCδ is crucial for directional T cell migration (66) (67) and this process requires MVB polarized traffic and exosome secretion in the direction of migration (60). For that reason, it will be interesting to analyse the effects of the FMNL1β mutants we have described here during T lymphocyte migration and tumorigenesis, as well as to study the role of FMNL1β phosphorylation in several autoimmune disorders. In addition, improving our understanding of the molecular bases underlying the traffic events involved in polarized secretion of pro-apoptotic exosomes as we have performed here will provide clues to modify crucial immune functions involving apoptosis, such as cytotoxicity by CTLs and immunoregulatory AICD and their associated pathologies, as well as it may allow to design therapeutic strategies to modify CAR T lifespan and/or polarized secretory function, in order to improve their efficiency for the treatment of certain cancers (17) (68) (2) (18) (25) (69) (70) (71).

Materials and Methods

Cells

Raji B cell line was obtained from the ATCC. Cell lines were cultured in RPMI 1640 medium containing L-glutamine (Invitrogen) with 10% heat-inactivated FCS (Gibco) and penicillin/streptomycin (Gibco). Jurkat control (C3 and C9) and PKCδ-interfered (P5 and P6) stable clones derived from JE6.1 clone have already been described and were cultured with puromycin (25). C3 and C9 control clones are used, instead of canonical JE6.1 cells, since these puromycin-resistant control clones (containing a scramble shRNA) were isolated by limiting dilution together with the PKCδ-interfered clones (25), thus are the best possible controls for P5 and P6 clones. The P5 and P6 clones exhibited a comparable defect in MVB/MTOC polarization when compared with the control clones (25)(26). In addition, when interference-resistant GFP-PKCδ was transiently expressed in all the PKCδ interfered clones, MTOC/MVB polarization was recovered to control levels (25). Therefore, the deficient MTOC/MVB polarization in these clones is exclusively due to the reduction in PKCδ expression (25), and thus clonal variation cannot be responsible of the results in stable clones. Since the FMNL1 interference and YFP-FMNL1 variants reexpression experiments are performed in transient assays (2–4 days after transfection), there is no chance for any clonal variation in these short-time experiments. Peripheral mononuclear cells from healthy donors (La Paz University Hospital, Madrid, Spain) were stimulated with Phytohemagglutinin (PHA-M, 1 μg/ml) for 3 days. Dual CAR T cells recognizing CD19/CD22 were produced as previously shown (72) and were provided by Department of Pediatric Hemato-Oncology, La Paz University Hospital (Madrid, Spain). Subsequently, both T lymphoblasts and CAR T cells were expanded in the presence of recombinant human IL-2 (20 ng/ml, Roche).

Plasmids and transient transfection

pEGFP-C1CD63 (expressing GFP-CD63) was provided by G. Griffiths. CD63 is an enriched marker in intraluminal vesicles (ILVs) contained into MVB and exosomes and has been widely used to study MVB secretory traffic in both living and fixed cells (2, 17). The FMNL1 interfering vector (shFMNL1-HA-YFP) and FMNL1-interfering, YFP-FMNL1βWT expressing vector (shFMNL1-HA-YFP-FMNL1βWT) were previously described (35) and generously provided by Dr. Billadeau. shFMNL1-HA-YFP-FMNL1βS1086A and shFMNL1-HA-YFP-FMNL1βS1086D mutants were generated by site-directed mutagenesis, as previously reported (73). Briefly, overlap extension PCR was used to introduce the desired mutations into a 900 bp Sa1I-NotI fragment, which was then used to replace the corresponding wild type sequence in shFMNL1-HA-YFP-FMNL1β-WT. All amplifications were carried out with Platinum SuperFi DNA polymerase (ThermoFisher). Primers used were: Sa1I-FMNL1β_F (CCAGAGCCTGGATGCGCTGTTGG), FMNL1 p—S1086A_F (CAGACACAGGCCGCCGCGCTGCCCGTCGGCGTCCC), FMNL1β—S1086A_R (GGGACGCCGACGGGCAGCGCGGCGGCCTGTGTCTGC), FMNL1 p—S1086D_F (CAGACACAGGCCGCCGCGATGCCCGTCGGCGTCCC), FMNL1 p—S1086D_R (GGGACGCCGACGGGCATCGCGGCGGCCTGTGTCTGC), and NotI-FMNL1β_R (GCGAGCTCTAGGGCCGCTTGCG). Presence of the desired mutations, and absence of undesired ones, was confirmed by Sanger DNA sequencing (Eurofins Genomics, LightRun). Both interference and expression of the different chimeric molecules were assessed in single cells level by immunofluorescence and/or bulk cell populations by WB with an anti-FMNL1 antibody, which recognizes all FMNL1 isoforms (see below). Jurkat clones were transiently transfected with 20–30 μg of the plasmids as described (17). To analyse the effects of shFMNL1 constructs on exosome secretion by Jurkat cells expressing exosome reporter GFP-CD63, Jurkat cells were transiently co-transfected with a three-fold molar excess of the different shFMNL1-YFP plasmids with respect to the GFP-CD63 exosome-reporter plasmid, to favor that the exosomes analysed were mainly those secreted by Jurkat cells containing both the GFP-CD63 plasmid and the shFMNL1 construct (2) (18).

Antibodies and reagents

Rabbit monoclonal anti-human PKCδ EP1486Y for WB that does not recognize mouse PKCδ (Abcam). Rabbit monoclonal anti-PKCδ EPR17075 for WB that recognizes both human and mouse PKCδ (Abcam). Mouse monoclonal anti-human CD3 (clone UCHT1) for cell stimulation and immunofluorescence (BD Biosciences and Santa Cruz Biotechnology). Mouse monoclonal anti-FMNL1 clone C-5 for WB and immunofluorescence and mouse monoclonal anti-FMNL1 clone A-4 for immunoprecipitation (Santa Cruz Biotechnology), recognizing all FMNL1 isoforms. Mouse monoclonal anti-γ-tubulin for immunofluorescence (SIGMA, clone GTU-88). Rabbit polyclonal anti-pericentrin ab4448 for immunofluorescence (Abcam). Mouse monoclonal anti-human CD63 (clone TEA3/18) for immunofluorescence (Immunostep). Mouse monoclonal anti-CD63 clone NKI-C-3 for WB (Oncogene). Rabbit polyclonal Phospho-Ser PKC substrate antibody for WB and immunofluorescence (Cell Signalling Technology, #2261). Fluorochrome-coupled secondary antibodies (goat-anti-mouse IgG AF488 A-11029, goat-anti-rabbit IgG AF488 A-11034, goat-anti-mouse IgG AF546 A-11030, goat-anti-mouse IgG AF647 A-21236) for immunofluorescence (ThermoFisher). Horseradish peroxidase (HRP)-coupled secondary antibodies (goat anti-mouse IgG-HRP, sc-2005 and goat anti-rabbit IgG-HRP, sc-2004) for WB (Santa Cruz Biotechnology). CellTracker™ Blue (CMAC) and phalloidin (ThermoFisher). Staphylococcal enterotoxin E (SEE) (Toxin Technology, Inc). Phytohemagglutinin (PHA-M) was from SIGMA. Recombinant human IL-2 was from Roche.

Immunoprecipitation

Immunoprecipitation from cell lysates was performed by using Protein A/G Magnetic Beads (Pierce, ThermoScientific) following the instructions provided by the company. Briefly, 0.5 ml lysates corresponding to 2–4×10⁶ transfected Jurkat cells, stimulated or not with PMA (100 ng/ml, 30 min at 37 °C), were incubated with anti-FMNLl (clone A4, 5 μg) for 2 h at 4° C. Subsequently, 15 μl of magnetic beads suspension was added and incubated for 3 h at 4° C. Beads were washed 5× with lysis buffer and the antigens were eluted with 2 M glycine pH =2 and then neutralized. Eluates were run on 6,5 % SDS-PAGE gels and proteins transferred to PVDF membranes.

Isolation and quantification of exosomes

To analyse the exosomes produced by cells transfected with the exosome reporter GFP-CD63, a similar protocol to the originally described for exosome isolation (74) was performed. We have included a 48 h post-transfection Ficoll gradient purification step to remove dead cells and cell debris, which otherwise could contaminate exosome preparations with micro and nanoparticles (2, 18). Using these standard protocols, culture supernatants from Jurkat cells transfected with GFP-CD63 or cotransfected with GFP-CD63 and with a 3-fold molar excess of the different shFMNL1-YFP plasmids, cocultured with SEE-pulsed Raji cells, were centrifuged in sequential steps to eliminate cells and cell debris/apoptotic bodies (74). Subsequently, the exosomes were recovered by ultracentrifugation (100,000×g for 12 h at 4° C) as described (16), lysed and then analysed by WB. CD63 is characteristically present in MVB, ILVs and hence in exosomes, but also in secretory lysosomes and the plasma membrane (Suppl. Video 8). Plasma membrane CD63 localization is produced by degranulation of MVB and diffusion of CD63 from the limiting membrane of MVB to the plasma membrane upon MVB fusion (2, 18) (Suppl. Video 8). Approximately 1–2×10⁶ transfected Jurkat cells were challenged with SEE-pulsed Raji cells and WB signals in exosome lysates were normalized by the cell expression levels of GFP-CD63 among different transfections, cell groups and stimuli, in the WB corresponding to the cell lysates (2, 47) (18). It has been established that the exosomal CD63 WB signal in this synapse model correlates with exosome number obtained by flow cytometry (75), by electron microscopy (76) and by nanoparticle concentration analysis (nanoparticles/ml), using NTA (18). WB analysis of CD63 protein and GFP-tagged CD63 in isolated exosomes has been used as a bona fide method to specifically determine changes in exosome production (17, 47, 77) in transiently-transfected Jurkat cells forming synapses with SEE-pulsed Raji cells without the detection of contaminating exosomes produced by MHC-II-stimulated Raji cells or constitutively secreted by Raji cells (78) (2, 17, 74) (18) (25) (22). No significant differences in the GFP-CD63 levels (i.e. Fig. 9B) were observed in the lysates of transfected C3 Jurkat cells, stimulated or not with SEE-pulsed Raji cells, at the end of the cell coculture period for exosome secretion, showing that the GFP-CD63⁺ exosomes were produced by an equal number of viable, GFP-CD63-expressing cells. Moreover, to quantify exosome concentration and to analyse their size distribution, the cell culture supernatant collected just before the ultracentrifugation step was diluted (1/5) in Hank’s balanced salt solution (HBSS) and analysed by NTA using a NANOSIGHT equipment (LM10, Malvern) that was calibrated with 50 nm, 100 nm and 400 nm fluorescent calibration beads (Malvern). The hydrodynamic diameter measured by NTA, although apparently higher to that originally described for exosomes using electron microscopy (50–100 nm) (Fig. 9A), certainly corresponds to the real size of canonical, unfixed exosomes in solution, as described (79). The NTA measurements of exosome concentration (particles/ml) were normalized by the transfection efficiency and exosome-producing Jurkat cell number, by referring exosome concentration to GFP-CD63 signals in the WB of the cell lysates (Fig. 9B).

Western blot analysis

Cells were lysed in Triton^TM X100-containing lysis buffer supplemented with both protease and phosphatase inhibitors. Approximately 50 μg of cellular proteins were recovered in the 10,000×g pellet from 10⁶ cells. Cell lysates and neutralized, acid-eluted IPs were separated by SDS-PAGE under reducing conditions and transferred to Hybond^TM ECL^TM membranes (GE Healthcare). Membranes were incubated sequentially with the different primary antibodies and developed with the appropriate HRP-conjugated secondary antibody using enhanced chemiluminescence (ECL). When required, blots were stripped following standard protocols prior to reprobing them with primary and HRP-conjugated secondary antibodies. For exosome secretion studies, cells and isolated exosomes were lysed in RIPA lysis buffer containing protease inhibitors. Approximately 5 μg of exosomal proteins were recovered in the 100,000×g pellet from 1–2×106 cells. Exosomes were resuspended in 60 μl of RIPA lysis buffer and 20 μl of exosomal or cell lysate proteins were separated by SDS-PAGE and transferred to Hybond^TM ECL^TM membranes (GE Healthcare). For CD63 detection, proteins were separated under nonreducing conditions as described (17). For WB analysis of exosomes, each lane contained the total exosomal protein that was recovered from the culture medium coming from the same number of cells, untreated or treated with stimuli. Blots were incubated with mouse anti-CD63 (clone NKI-C-3, Oncogene) and developed with the appropriate HRP-conjugated secondary antibody using ECL. Autoradiography films were scanned and the bands were quantified using Quantity One 4.4.0 (Bio-Rad) and ImageJ (Rasband, W.S., ImageJ, National Institutes of Health, Bethesda, Maryland, USA, http://rsb.info.nih.gov/ij/, 1997–2004) software.

Time-lapse microscopy, immunofluorescence and image analysis

In the experiments requiring IS formation, we challenged the transfected Jurkat clones with SEE-pulsed Raji cells, a well-established synapse model (80). Raji cells were attached either to ibiTreat microwell culture dishes (ibidi) pretreated with fibronectin (0.1 mg/ml, for paraformaldehyde-PFA-fixation) or glass bottom microwell culture dishes (ibidi) pretreated with poly-L-lysine (0.02 mg/ml, for PFA and acetone fixation). Next, they were labelled with CMAC (10 μM) and pulsed with 1 μg/ml SEE and mixed with Jurkat clones transfected with the different expression plasmids, at 24–48 h posttransfection. The resulting IS were analysed as described (80) (2) (44) (26). PFA followed by acetone fixation was required for a clean FMNL1 staining and was compatible with phalloidin labelling (81). Immunofluorescence of fixed synapses was performed as previously described (82), and additional blocking and fixations steps were performed between each primary antibody and subsequent fluorochrome-coupled secondary antibody or phalloidin staining, to exclude any potential cross-reaction of secondary antibodies (i.e., Fig. 5).

For FMNL1β relocalization experiments in living cells, control (C3) and PKCδ-interfered (P5), YFP-FMNL1β-expressing Jurkat clones were challenged with CMAC-labelled, SEE-pulsed Raji cells as described above. Wide-field, time-lapse microscopy was performed using an OKO-lab stage incubator (OKO) for sample ambient on a Nikon Eclipse TiE microscope equipped with a Prime BSI (Photometrics) digital camera, a PlanApo VC 60×/1.4NA OIL objective (Nikon) and with Perfect Focus System, allowing distinct, appropriate offsets per each XY field. Wide-field fluorescence of fixed synapses was performed by capturing 30–40 Z sections (0.3–0.4 μm thickness) using the same microscope and NIS-AR software (Nikon), and maximum intensity projection (MIP) images of all channels were generated using NIS-AR software.

Time-lapse acquisition during the indicated times and analysis were performed using NIS-AR software (Nikon). Subsequently, in some experiments epifluorescence images were improved by Huygens Deconvolution Software from Scientific Volume Image, using the “widefield” optical option as previously described (43) (44). For quantification, digital images were analysed using NIS-AR (Nikon) or ImageJ software. The quantification and analysis of YFP MFI at the synapse area (YFP-FMNL1β MFI IS) and at the whole cell (YFP-FMNL1β MFI Cell) in time-lapse experiments, were performed within floating regions of interest (ROI) (i.e., ROI changing XY position over time) by using NIS-AR software (Fig. 6A and Suppl. Fig. S8).

Confocal microscopy imaging in fixed synapses was performed by using an SP8 Leica confocal microscope (63×/1.40 Oil DIC objective), with sequential acquisition, bidirectional scanning and the following laser lines: UV (405 nm, intensity: 33.4%), supercontinuum visible (633 nm, intensity: 15.2%), supercontinuum visible (550 nm, intensity: 20.8%), supercontinuum visible (488 nm, intensity: 31.2%). STED fluorescence microscopy was performed by using a Stellaris 8 Tau STED (Leica) equipped with a 775 nm depletion laser and a 100x/1.4 Oil STED white Objective. Deconvolution of confocal images and widefield images was performed by using Huygens Deconvolution Software from Scientific Volume Image with the “confocal” optical option or “widefield” option, respectively. 2D colocalization analyses were accomplished by using JACoP plugin from ImageJ, whereas interface colocalization and the corresponding videos to generate the IS interface were performed using the “colocalization” tool, followed by the “volume view” and then “movie maker” tools from NIS-AR. For interface colocalization analyses, the original emission colors of the different fluorochromes were changed to Red, Green or Blue, since NIS-AR requires and RGB color code for this analyses (83). In particular, F-actin and Phospho-Ser PKC substrate acquired in magenta were changed to blue, and YFP acquired in yellow was changed to green (i.e. Fig. 8, Suppl. Fig. S4, Suppl. Fig. S5 and Suppl. Fig. S11). Profile analyses of MFI corresponding to each separate channel and the colocalization pixels at the IS interface (ZX axes) were performed by using “Intensity Profile” tool in NIS-AR and then Excel, as reported (83). To measure the relative size of the F-actin low area at the cIS, which quantifies the decrease in F-actin density at the cIS, we used the last frame from the former IS interface videos (Suppl. Video 5 and Fig. 7), generated as described (83). The areas of the F-actin-low region at cIS (Fact-low cIS area) (yellow line) and the synapse (IS area) (white line) (i.e., Fig. 7) were defined in ImageJ using the ROI manager and applying an appropriate, manually-defined threshold to the F-actin channel. The definition of the ROI to measure the Fact-low cIS area and the IS area was performed manually or by using automated algorithm “auto-detect ROI/segmentation” from NISAR (25) (26). As for the Fact-low cIS area (yellow line) and the IS area (white line) (i.e., Fig. 7), they were defined in ImageJ using the ROI manager and applying a default threshold or, in cases where precise detection was not achieved, an appropriate, manually-defined threshold to the F-actin channel. In both cases, we have found that both manual and automated values were quite similar. The areas included in these ROIs were measured and limited to the defined thresholds, and the relative area of the F-actin-low region at the cIS (Fact-low cIS area / IS area), which is independent of both cell and synapse size, was calculated and represented (Fig. 7). Regarding the possibility that actin cytoskeleton of Raji cells can also contribute to the measurements of synaptic F-actin, it is important to remark that MHC-II-antigen triggering on the B cell side of the Th synapse does not induce noticeable F-actin changes along the synapse (i.e. F-actin clearing at the cIS), in contrast to TCR stimulation on T cell side (84) (85) (3). In addition, we have observed that majority of F-actin at the IS belongs to the Jurkat cell (83). Thus, the contribution to the analyses of the residual, invariant F-actin from the B cell is negligible using our protocol (83).

To compare MTOC and MVB polarization in synapses, MTOC and MVB PI were calculated, as previously described (37) (25) (26) (86) (Suppl. Fig. S2A), using MIP, both for deconvoluted, wide-field, and confocal-acquired images. In the MIP, the positions of the cell geometric center (Cell^C) and MTOC and MVB center of mass (MTOC^C and MVB^C, respectively) were used to project MTOC^C (or MVB^C) on the vector defined by the Cell^C-synapse axis. Then the MTOC (or MVB) PI was calculated by dividing the distance between the MTOC^C (or MVB^C) projection and the Cell^C (“B” or “A” distance, respectively) by the distance between the Cell^C and the synapse (“C” distance) (Suppl. Fig. S2A, Fig. 4 and Fig. 5). Cell^C position was taken as the origin to measure distances, thus those “A” and “B” values in the opposite direction to the synapse were taken as negative. Thus, PI (PI = A/C or B/C) ranked from +1 (fully polarized) to -1 (fully antipolarized). Therefore, PI values were normalized by cell size and shape (Fig. 4) (25) (86) and measured the relative ability of MTOC and MVB to polarize towards the IS. Remarkably, one important feature of the IS consists of both the onset of the initial cellcell contacts and the establishment of a mature, fully productive IS, are intrinsically stochastic, rapid and asynchronous processes (87, 88) (43). Thus, the score of the PI corresponding to the distance of MTOC/MVB with respect the IS (42) may be contaminated by background MTOC/MVB polarization, in great part due to the stochastic nature of IS formation (87). In order to circumvent this caveat, a substantial number of IS of each cell group/condition were analysed to obtain statistically significant results, as reported by numerous authors (89) (37) (90) (87) (25) (26). Image analysis data correspond to at least three different experiments, analysing a minimum of 30 synapses from 15 different, randomly selected, microscopy fields per experiment. ANOVA was performed for statistical significance of the results using Excel and IBM’s SPSS Statistics software.

Abbreviations:

• AICD: activation-induced cell death

• ANGPTL4: Angiopoietin-like 4

• APC: antigen-presenting cell

• BCR: B-cell receptor for antigen

• ^C: center of mass

• CAR T: chimeric antigen receptor-bearing T lymphocytes

• cIS: central region of the immune synapse

• CMAC: CellTracker™ Blue (7-amino-4-chloromethylcoumarin)

• cSMAC: central supramolecular activation cluster

• CTL: cytotoxic T lymphocytes

• DAD: diaphanous autoinhibitory domain

• Dia1: Diaphanous-1

• DID: diaphanous inhibitory domain

• dSMAC: distal supramolecular activation cluster

• ECL: enhanced chemiluminescence

• F-actin: filamentous actin

• Fact-low cIS: F-actin-low region at the center of the immune synapse

• FasL: Fas ligand

• FHOD1: formin homology domain protein 1

• FMNL1: formin-like 1

• fps: frames per second

• HBSS: Hank’s balanced salt solution

• HRP: horseradish peroxidase

• ILVs: intraluminal vesicles

• IS: immune synapse

• IPs: immunoprecipitates

• MFI: mean fluorescence intensity

• MHC: major histocompatibility complex

• MIP: maximum intensity projection

• MVB: multivesicular bodies

• MTOC: microtubule-organizing center

• MW: molecular weight

• NTA: Nanoparticle Tracking Analyses

• NS: not significant

• PFA: paraformaldehyde

• PHA: Phytohemagglutinin

• PKC: protein kinase C

• PKC8: protein kinase C 8

• PMA: phorbol myristate acetate

• PI: polarization index

• pSMAC: peripheral supramolecular activation cluster

• ROI: region of interest

• ROCK: Rho-dependent protein kinase

• SD: standard deviation

• SEE: Staphylococcal enterotoxin E

• SEB: Staphylococcal enterotoxin B

• SMAC: supramolecular activation cluster

• SMAPs: supramolecular attack particles

• TCR: T-cell receptor for antigen

• Th: T-helper

• TRANS: transmittance

• WB: western blot

• YFP: yellow fluorescent protein.

Acknowledgements

We are indebted and acknowledge Dr. D.D. Billadeau (Mayo Clinic, USA) for generous sharing of shFMNL1 and FMNL1 isoform rescue constructions. We acknowledge the excellent technical support from A. Sánchez. We acknowledge Dr. M.A. Alonso (CBM, CSIC) for reagents and scientific advice. We acknowledge the generous contribution of pre-graduate students Alejandro Martín, Irene Gascuña, Gregorio Pantoja, María Ruiz, Sofía Blázquez, Carlos del Hoyo, Pablo del Barrio and Elena Fernández to this work. Thanks to D. Morales (SIDI-UAM), A. Oña and G. D’Agostino (CNB, CSIC) and M. Martín (IIBM,CSIC) for their superb expertise with confocal microscopy. Work in the F.R.G-G lab was funded by grant PID2019-104941RB-I00 from the Spanish Ministry of Science and Innovation (MCIN/AEI/ 10.13039/501100011033). This work was supported by grants from the Programa Estatal de Investigación, Desarrollo e Innovación Orientada a los Retos de la Sociedad (Grant PID2020-114148RB-I00 from the Spanish Ministry of Science and Innovation MCIN/AEI/ 10.13039/501100011033) to MI.

Conflict of Interest

The authors report no conflict of interest.

Author Contributions Statement

V.C. and M.I. conceived and designed all the experiments. J.R-N. did most of the experiments, analysed data, and contributed to the writing of the manuscript. J.R-N., S.F, and I.S. contributed to the MTOC/MVB polarization experiments and IS image analyses. S.F and I.S. performed the FMNL1 immunoprecipitation experiments and contributed to the FMNL1 phosphorylation studies and image analyses, and J.R-N. contributed to timelapse studies and image analyses of area F-actin. A. N.-Z. and A. P.-M. developed the CART cells and CART synapses studies. P.B. and F.R.G-G. created and sequenced the FMNL1β mutants, and contributed to manuscript writing. M.I. conceptualized and coordinated the research, directed the study, analysed data, and wrote the manuscript. All the authors contributed to the planning and designing of the experiments and to helpful discussions.

Funding Statement

This work was supported by a grant from the Programa Estatal de Investigación, Desarrollo e Innovación, Modalidad Retos Investigación (Grant PID2020-114148RB-I00 funded by Spanish Ministry of Science and Innovation MCIN/AEI/ 10.13039/501100011033), and grant P2022/BMD-7225, funded by Consortia in Biomedicine of Comunidad de Madrid to MI.

Data Availability Statement

The materials described in the manuscript, including all relevant raw data, will be freely available to any researcher wishing to use them for non-commercial purposes, without breaching participant confidentiality.

Additional Declarations:

The authors declare no competing interests.

Supplementary figures

Supplementary Fig. S1

Supplementary Fig. S2

Supplementary Fig. S3

Supplementary Fig. S4

Supplementary Fig. S5

Supplementary Fig. S6

Supplementary Fig. S7

Supplementary Fig. S8

Supplementary Fig. S9

Supplementary Fig. S10

Supplementary Fig. S11

Supplementary Fig. S12