CXCL12 Regulates through JAK1 and JAK2 Formation of Productive Immunological Synapses

Adaptive immune responses are initiated through interaction between naive Ag-specific T cells and Ag-bearing dendritic cells (DC) in the lymph node. A specialized structure, termed the immune synapse (IS), is formed at the T cell–APC contact site. This occurs immediately after TCR activation and is implicated in the induction of T cell proliferation, cytokine production, and lytic granule release (1). The IS is characterized by protein segregation, signaling compartmentalization, and bidirectional information exchange through soluble and membrane-bound transmitters. The TCR and associated proteins congregate in the central area (central supramolecular activation cluster; cSMAC), whereas integrins reorganize in a surrounding external ring called the peripheral SMAC (pSMAC) (2). IS formation also requires participation of the actin cytoskeleton, which forms a ring in the pSMAC.

Although TCRs are essential for IS organization, a number of other surface receptors such as adhesion molecules and costimulatory receptors also have important roles in SMAC formation and in cell cross-talk. Interaction between LFA-1 and the ICAMs precedes mature synapse development and facilitates signaling events that promote T cell activation (3, 4).

Chemokine receptors are also recruited to the IS (5, 6), which suggests that chemokines can help T cells to interpret the context of Ag presentation. During T cell activation, the chemokines CCL5 and CXCL12 enhance T cell proliferation and cytokine production (5, 7, 8), which indicates that chemokines might act as soluble immunotransmitters at the IS and help to activate T cells. Signals relayed from the active TCR trigger integrin activation at the IS, and there is evidence that chemokines cooperate in this process. Chemokine-mediated integrin activation is essential for cell extravasation (9, 10), and chemokines in the IS can reinforce T cell–APC attraction and prevent premature separation, thus facilitating T cell activation (11, 12). Chemokine-induced LFA-1 activation leads to rapid recruitment of microtubule-organizing center (MTOC) and mitochondria to the APC, which amplifies TCR-induced Ca²⁺ signaling at the nascent immunological synapse (13).

Regulation of integrin activation depends on a wide variety of signaling proteins (14). Signaling through small GTPases is currently the best-studied mechanism of chemokine-induced integrin activation (15, 16). Recent evidence nonetheless indicates that JAKs also participate in the control of chemokine-induced integrin activation (10, 17) and might modulate the activation hierarchy of the small GTPases.

Using CXCL12 and its receptor CXCR4 as a model, we assessed the precise role of chemokine-mediated signaling in T cell activation during IS formation. After downregulation of CXCR4 expression, after treatment with the CXCR4-specific binding site inhibitor AMD3100 or pertussis toxin (PTx), or after drug inhibition or downregulation of JAK1/2 expression, OT-II CD4⁺ T cells showed defective IS formation when cocultured with OVA peptide–loaded bone marrow–derived DC (BM-DC). These T cells also showed reduced F-actin accumulation as well as defects in MTOC polarization to the contact site and in IS formation. Time-lapse videomicroscopy analysis showed significantly reduced mean contact duration, reflected as T cell activation defects. Our results indicate that whereas T cell–APC interactions are Ag dependent (18, 19), these Ag-specific interactions coincide with Ag-independent, chemokine-promoted adhesive contacts between these cells, which help to build a productive immune synapse.

Male and female 3- to 5-mo-old C57BL/6 mice were purchased from Harlan Laboratories. OVA-specific TCR-transgenic mice (OT-II) were donated by Dr. C. Ardavín (Centro Nacional de Biotecnología, Madrid, Spain) and Cxcl12^Gagtm mice by Dr. F. Arenzana (Institut Pasteur, Paris, France). The phenotype of OT-II mouse offspring was confirmed by flow cytometry using anti–Vα2 TCR Ab (BD Pharmingen). Cxcl12^Gagtm/wt and Cxcl12^Gagtm/Gagtm mice were genotyped by PCR amplification on genomic DNA with specific primers (20). Mice were housed in pathogen-free conditions at the animal facility at the Centro Nacional de Biotecnología/Consejo Superior de Investigaciones Cientificas. All animal experiments were approved by the appropriate ethics committees and carried out according to national and European Union guidelines.

For Western blot assays, we used the following Abs: anti-JAK1, -Tyk2 (Santa Cruz Biotechnology), -JAK2 (Upstate Biotechnology), –Vav-1 mAb (Cell Signaling Technology), –ZAP-70 mAb (BD Pharmingen), –P-Tyr (Promega), –β-tubulin mAb, and -CXCL12 mAb (Sigma-Aldrich). HRP-labeled anti-mouse and -rabbit Ig (DakoCytomation) were used as secondary Abs. Cell purity was assessed by flow cytometry using anti–B220-FITC and –Gr1-PECy7 (Beckman Coulter), –CD4-PECy5 and –CD3-APC mAb (BioLegend), and –CD11c-PE (eBioscience). Surface marker expression was determined by flow cytometry using anti–CD69-PE and –CD25-PE mAb (BioLegend) and anti–CXCR4-biotin mAb (BD Pharmingen). CXCL12 was purchased from PeproTech. CellTrace CFSE and seminaphtharhodafluor-1 were obtained from Molecular Probes. Ruxolitinib and tofacitinib were donated by Prof. P. Cohen (MRC Protein Phosphorylation Unit, University of Dundee, Dundee, Scotland).

CD4⁺ naive T cells were obtained from OT-II mouse spleen and lymph nodes and purified by negative selection with a mouse T cell negative isolation kit (MACS; Miltenyi Biotec); T cell purity was routinely >97%. Freshly isolated murine OT-II CD4⁺ T cells were nucleofected (mouse T cell Nucleofector kit; Amaxa) with a mixture of four individual small interfering RNA (siRNA) duplexes (ON-TARGETplus SMARTpool siRNA; Dharmacon) for the target sequences on JAK1 (5′-GAAAAUGAAUUGAGUCGAU-3′; 5′-GAAAUCACCCACAUUGUAA-3′; 5′-CGCAUGAGGUUC UACUUUA-3′; 5′-GCACAGGGACAGUAUGAUU-3′), JAK2 (5′-AAUAGGAGACUUCGGAUUA-3′; 5′-GAAUUGUAACUGUCCAUAA-3′; 5′-GAACUUAGCUCAUUAAAAG-3′; 5′-GAAUUUAUGCGAA UGAUUG-3′), Tyk2 (5′-UGACAGAAUUCGUAGAACA-3′; 5′-GAACCUCUAUUUUCGAAUG-3′; 5′-CAAGGACCAUGUGUAUGA-3′; 5′-CCUGAUGGAUCCGUUUGUA-3′), and CXCR4 (5′-GAACCGA UCAGUGUGAGUA-3′; 5′-GUGUAAGGCUGUCCAUAUC-3′; 5′-GUGUUUCAAUUCCAGCAUA-3′; 5′-AAACGUCCAUUUCAAUAGG-3′) or siRNA control (ON-TARGETplus Nontargeting Pool D-001810-10; Dharmacon) to ensure silencing specificity. Nucleofection efficiency was controlled with fluorescence-labeled siRNA duplex siGLO Green indicator (Dharmacon) as a positive control. Nucleofected naive OT-II CD4⁺ T cells were incubated (24 h, 37°C) in the medium supplied with the Nucleofector kit and murine IL-7 (25 U/ml; Cell Signaling Technology).

BM-DC were obtained from bone marrow cell suspensions treated with erythrocyte lysis buffer and then cultured in 150-mm petri dishes in complete RPMI 1640 medium supplemented with 10% FCS and 20 ng/ml rGM-CSF (PeproTech) (21). Immature BM-DC were collected at day 8, and purity was evaluated by flow cytometry using anti–CD11c-FITC and –MHCII-PE Ab (BD Pharmingen). Batches with >95% purity were used for maturation with LPS (1 μg/ml, 12 h, 37°C).

Cells were plated in V-bottom 96-well plates (2.5 × 10⁵ cells/well) and incubated (30 min, 4°C) with 50 μl/well anti–CD3-allophycocyanin and –CD25-PE or anti–CD69-PE (all from BioLegend) for determination of T cell activation. When needed, anti-CXCL12 mAb, followed by goat anti-mouse IgG-Cy3 or anti–CXCR4-Biotin mAb, followed of avidin-PE were used. Cell-bound fluorescence was determined in a Gallios flow cytometer (Beckman Coulter).

IS formation between mouse BM-DC and OT-II CD4⁺ T cells was induced as reported previously (22, 23). Briefly, matured BM-DC (1 μg/ml LPS, 12 h, 37°C) were loaded with OVA_323–339-peptide (5 μM, 30 min, 37°C; GenScript) and mixed in complete medium (RPMI 1640 medium, 10% FCS) with CD4⁺ T cells (1:5 DC:CD4⁺ T cells). Cells were centrifuged (120 × g, 2 min) in a conical tube and incubated in complete medium (15–30 min, 37°C) to foster IS formation. When necessary, T cells were pretreated with PTx (0.2 μg/ml, 120 min, 37°C; Sigma-Aldrich), AMD3100 (10 μM, 30 min, 37°C; Sigma-Aldrich), ruxolitinib (1 μM, 60 min, 37°C), or tofacitinib (1 μM, 60 min, 37°C) before mixing with BM-DC. For AMD3100, ruxolitinib, and tofacitinib treatments, drug levels were maintained throughout the experiment.

Conjugation was scored visually by counting under a confocal microscope, with at least 50 random T cells scored as conjugates only if they contacted no more than one CMAC-labeled BM-DC and no other T cell contacted the scored T cell. F-Actin accumulation at the IS was detected by staining with phalloidin–Alexa 488, and images were analyzed with ImageJ software (National Institutes of Health, Bethesda, MD). Using region of interest of the same area for all measurements, we quantified the signal generated by 1) the BM-DC/T cell contact area (IS), 2) the area of BM-DC membrane not in contact with the T cell (B), 3) an area of the T cell membrane not in contact with the BM-DC (T), and 4) the background. We subtracted the background signal from other measurements, then measured F-actin accumulation at the IS relative to the rest of the T cell using the formula (IS − B)/T. Results were expressed as the percentage of T cells with accumulated F-actin at the contact with DC. Statistical significance was calculated using a two-tailed Student t test.

Cell conjugates were observed under a confocal laser scanning unit (TCS SP5; Leica) attached to an inverted epifluorescence microscope (DMI6000; Leica) fitted with an HCX PL APO ×40/1.40-0.6 oil objective. Images were acquired and processed with confocal software (LCS; Leica) or WCIF ImageJ (http://rsbweb.nih.gov/ij/) to obtain a three-dimensional analysis and maximal projections of T cell–bead contact. Figures were composed with Adobe Photoshop.

OT-II CD4⁺ T cells were nucleofected as above with siRNA pools for JAK1+JAK2 (siJAK1,2) or siRNA pool control (Dharmacon). These cells, alone or treated with AMD3100 (10 μM, 30 min, 37°C) or PTx (0.2 μg/ml, 120 min, 37°C), were conjugated to anti-CD3–coated latex beads or to isotype control–coated latex beads (Sigma-Aldrich) for 20 min, alone or with CXCL12 (50 nM). Cells were then fixed and stained with appropriate Abs, and the IS was analyzed by confocal imaging. We assumed that background signal is homogeneous in the latex beads. To quantify the fluorescence distribution ratio, individual IS were processed with the “Synapse Measures” plugin for ImageJ (http://rsbweb.nih.gov/ij/) (for a detailed description of Synapse Measures, including the algorithms used, see Ref. 24). To analyze talin and CXCR4 distribution at the IS-like complex, three-dimensional reconstructions of the area were generated with the Leica software. Mean fluorescence intensity (MFI) per area was calculated with a Matlab routine for pSMAC and cSMAC regions from three independent experiments (at least 24 cells were measured per experimental condition). Mean values from the pSMAC/cSMAC ratio of MFI per area were calculated.

OT-II CD4⁺ T cells were labeled with CellTrace Violet (0.5 μM, 30 min, 37°C; Molecular Probes) and added to plates containing OVA peptide–loaded BM-DC (3 × 10⁴ DC/6 × 10⁴ T cells). Cells were cocultured in RPMI 1640 medium with 10% FCS (24, 48, or 72 h), and proliferation was determined by flow cytometry using dye dilution evaluation in a Gallios flow cytometer (Beckman Coulter). The percentage of dividing cells was calculated using Flowjo (Tree Star). T cell activation was evaluated in parallel at indicated time points using anti–CD3-allophycocyanin and –CD25-PE or anti–CD69-PE mAb in flow cytometry, as above.

CXCL12 (50 nM)-activated OT-II CD4⁺ T cells (5 × 10⁶) were lysed in 200 μl detergent buffer (1% Nonidet P-40, 50 mM Tris-HCl [pH 8], 150 mM NaCl, 0.5 mM EDTA, 10 mM sodium pyrophosphate, 1 mM PMSF, 10 μg/ml aprotinin, 10 μg/ml leupeptin, and 10 mM sodium orthovanadate; 30 min, 4°C) and immunoprecipitated using anti-PTyr Ab (1 μg/sample; Promega), and cell extracts were analyzed in Western blot, as described previously (25). When needed, densitometry analyses were performed using ImageJ.

Rho GTPase activity was determined in T cell lysates using RhoA activation assay kits (BK124; Cytoskeleton). T cells were stimulated with CXCL12 (50 nM; various times), and the reaction was terminated by adding ice-cold PBS and centrifugation (470 × g, 5 min, 4°C); lysis and GTPase activity were measured following G-LISA kit protocols.

IL-2 mRNA levels were analyzed by quantitative real-time PCR with specific primers. Cells (3 × 10⁵) were lysed, and RNA was extracted using the RNeasy micro kit (Qiagen) with DNase treatment to digest residual genomic DNA. Equal RNA amounts were reverse transcribed using a reverse transcription system (Promega). cDNA was determined by semiquantitative real-time PCR with the LightCycler480 SybrGreen I Master kit (Roche Diagnostics) using specific primer pairs (IL-2, 5′-CACTTCAAGCTCTACAGCGGA-3′ and 5′-AAAATTTGAAGGTGAGCATCC-3′; β-actin, 5′-GGCACCACACCTTCTACAATG-3′ and 5′-TGGATGGCTACGTACATGG CTG-3′). Samples were analyzed in duplicate and normalized to β-actin using ABI 7900HT SDS 2.3 software.

The conjugates (5 × 10⁴ OVA peptide–loaded BM-DC/2.5 × 10⁵ OT-II CD4⁺ T cells) were plated on coverslips precoated with poly-l-lysine (20 μg/ml, overnight, 4°C; Sigma-Aldrich) and cultured (30 min, 37°C). Cells were washed in cold TBS and fixed with 2% paraformaldehyde (10 min, 37°C) and then permeablized with 0.2% Triton X-100 in TBS (10 min, 37°C). After washing with TBS with 1% BSA, 0.1% goat serum, and 0.05% Tween 20 to avoid nonspecific binding, cells were stained with anti-Ptyr mAb (Millipore) or –γ-tubulin (Sigma-Aldrich) (60 min, 37°C), followed by Cy3-goat anti-mouse IgG (3.5 μg/ml, 45 min, 37°C; Jackson ImmunoResearch Laboratories). When necessary, cells were stained with Alexa 488–phalloidin (60 min, 37°C; Sigma-Aldrich). Coverslips were mounted on slides with Fluoromount G medium (Southern Biotechnology Associates), and fluorescence was evaluated using the Olympus IX81 microscope with a PLAPON 40 × 03 objective (aperture 1:40) and FV10-ASW 1.6 software. When needed, MFI was determined using ImageJ64 software; staining intensity is shown as a percentage of total MFI versus MFI at the cell–cell contact site.

OVA peptide–loaded BM-DC (5 × 10⁴ cells) cells were added to fibronectin-coated glass chambers (10 μg/ml, 4°C, overnight; Nunc Lab-Tek), placed on a 37°C stage of a laser-scanning fluorescence microscope (AMI 6000B; Leica), and allowed to settle (60 min). We then added 2.5 × 10⁵ OT-II CD4⁺ T cells. Images were acquired every 30 s before, during, and after T cell addition. Videos were analyzed with image analysis software (Adobe Photoshop CS5).

Statistical analysis was performed with Prism software (GraphPad) using the non-parametric Student t test. Multiple comparisons were analyzed with a one-way ANOVA, followed by the Tukey posttest or Dunn’s posttest, where needed.

CXCR4 is recruited to the pSMAC at the IS (6), and its ligand, CXCL12, is secreted by bone marrow stroma cells and endothelial cells (26), among other cell types (27, 28). Using FACS and immunofluorescence techniques, we detected CXCL12 bound to the surface of BM-DC but not on naive OT-II T cells (Supplemental Fig. 1A). To assess the effect of CXCR4/CXCL12 blockade on IS formation, naive OT-II CD4⁺ T cells were pretreated with AMD3100 (10 μM, 30 min, 37°C) or PTx (0.2 μg/ml, 120 min, 37°C) and used to form conjugates with OVA peptide–loaded BM-DC (ratio 5:1, 30 min, 37°C). Phalloidin staining showed that whereas 87.5 ± 3.5% of untreated OT-II CD4⁺ T cells showed cytoskeletal rearrangement at the IS contact zone, F-actin accumulation was greatly reduced in T cells treated with AMD3100 (39.7 ± 2.5%) or PTx (20.3 ± 1.5%) (Fig. 1A). Because MTOC translocation toward the contact zone is a hallmark of IS formation (29), we tested for γ-tubulin in the BM-DC/T cell conjugates (Supplemental Fig. 1B). Quantitative analysis showed that AMD3100 or PTx treatment significantly reduced MTOC translocation (Fig. 1B).

To confirm these observations, we used nucleofected OT-II CD4⁺ T cells with an siRNA pool for CXCR4 (siCXCR4) or control (siControl). Flow cytometry analysis using anti-CXCR4 mAb showed a marked reduction in CXCR4 expression at the cell surface (∼85%; Supplemental Fig. 1C). We thus formed conjugates using siCXCR4 and siControl cells and OVA peptide–loaded BM-DC and stained for γ-tubulin as above. The results indicated significant alteration in MTOC translocation to the contact zone in CXCR4-deficient cells (Fig. 1C).

We next determined the effect of AMD3100 and PTx treatments on BM-DC/T cell conjugation dynamics time-lapse videomicroscopy. The majority of untreated T cells formed medium-lived (5–15 min) and long-lived (>15 min) contacts, whereas drug treatments significantly increased the proportion of short-lived (<5 min) contacts (Fig. 1D). Mean contact duration of untreated OT-II CD4⁺ T cells with BM-DC (14 ± 1 min) was significantly longer than with AMD3100- or PTx-treated T cells (4 ± 0.3 and 4.5 ± 0.2 min, respectively; Fig. 1E, 1F, Supplemental Videos 1–3).

TCR engagement induces phosphorylation and relocation of specific signaling proteins that interact with the TCR-CD3 complex (30). To assess the role of chemokines in DC-triggered signaling events during IS formation, we analyzed phosphotyrosine staining as a T cell activation marker. Untreated or AMD3100- or PTx-treated naive OT-II CD4⁺ T cells were incubated with OVA peptide–loaded BM-DC for various times; cells were fixed, permeabilized, stained for PTyr, and evaluated by FACS. Treatment with either drug clearly reduced anti-PTyr staining in CD4⁺ T cells (Fig. 2A).

We then analyzed in detail the contribution of CXCL12 to the IS structure. OT-II CD4⁺ T cells were conjugated with anti–CD3-coated latex beads (31), alone or in the presence of soluble CXCL12 (50 nM), and cells were stained for specific IS markers (CD3ζ for cSMAC; talin and CXCR4 for pSMAC) (Fig. 2B, 2C, Supplemental Fig. 1D). We found partial colocalization of talin and CXCR4 in a ring-shaped structure, as shown in three-dimensional reconstitution (Fig. 2B); this colocalization was reinforced by CXCL12 addition, as indicated by the coincidence of red and blue lines in the histograms in Fig. 2B. Talin and CXCR4 redistribution at the pSMAC was clearly increased in the presence of CXCL12 (pSMAC/cSMAC ratio > 1; graphs in Fig. 2C). CXCL12 treatment similarly increased the anti–CD3-triggered IS localization ratio of CD3ζ (Supplemental Fig. 1D). These CXCL12-mediated effects were reduced by treatment of OT-II CD4⁺ T cells with AMD3100 or PTx (Fig. 2B, 2C, Supplemental Fig. 1D).

We compared the in vitro activation of drug-treated CD4⁺ cells in the OVA peptide–loaded BM-DC system. Flow cytometry analysis showed that AMD3100 and PTx treatment reduced the stimulatory effect on both CD25 and CD69 expression markers (Fig. 3A, Supplemental Fig. 2A). We therefore used untreated and AMD3100- or PTx-treated live cell tracker–labeled OT-II CD4⁺ T cells to form conjugates with OVA peptide–loaded BM-DC and determined their ability to proliferate. Flow cytometry detection of cell tracker dilution after 48-h coculture showed lower proliferation of drug-treated T cells (Fig. 3B). These results indicate that CXCL12 binding to CXCR4 in T cells is an important element for IS formation and that blockade of this interaction alters conjugate formation and stability and thus T cell activation.

To determine whether the CXCL12 effect requires chemokine binding to glycosaminoglycans (GAG) on the BM-DC surface, we prepared BM-DC from Cxcl12^Gagtm/Gagtm mice, which bear a mutated CXCL12 gene (Cxcl12Gagtm) (Supplemental Fig. 2B) that precludes CXCL12 interaction with heparin sulfate structures but does not affect CXCR4-dependent CXCL12 cell signaling (20). Conjugates of OVA peptide–loaded Cxcl12^Gagtm/Gagtm BM-DC with OT-II CD4⁺ T cells (ratio 1:5, 30 min, 37°C) showed a small but significant reduction in MTOC translocation toward the contact zone that was not detected when we used BM-DC from Cxcl12^Gagtm/wt mice (Fig. 3C). In addition, time-lapse videomicroscopy analysis of conjugation dynamics showed similar levels of short-, medium-, and long-lived contacts compared with those for wild-type (wt) BM-DC (Fig. 3D). These results concur with those using anti-CD3–coated beads and soluble CXCL12, suggesting that in vitro CXCL12–mediated IS stabilization does not require chemokine binding to GAG.

Integrin activation is an essential step in IS stabilization (32); it is triggered by TCR engagement and chemokine binding to their receptors (33, 34), which require ZAP-70 phosphorylation and Rho small GTPase activation (16, 35). In static adhesion assays, CXCL12 stimulation induced OT-II CD4⁺ T cell adhesion to ICAM-1 and potentiated that triggered by anti-CD3 stimulation (Supplemental Fig. 2C). To evaluate the CXCL12-mediated signaling cascades involved, we used OT-II CD4⁺ T cells, untreated or treated with AMD3100 (10 μM, 30 min, 37°C) or PTx (0.2 μg/ml, 120 min, 37°C). CXCL12 promoted rapid ZAP-70 phosphorylation and RhoA activation; AMD3100 treatment notably decreased both processes, which were reduced to a lesser extent by PTx (Fig. 4A, top, 4B). The nucleotide exchange factor (GEF) Vav is also associated with CXCL12-mediated LFA-1 activation (16). We found that, in CXCL12-activated CD4⁺ T cells, Vav-1 was phosphorylated. This activation was abolished by AMD3100 treatment and was not blocked in PTx-treated cells (Fig. 4A, middle). A recent report suggests G protein–independent CXCL12 mediated Vav-1 phosphorylation, which links Rho small GTPase and integrin activation in a pathway that entails rapid CXCL12-mediated JAK activation (17). We thus tested the effect of specific JAK inhibitors on IS formation, including ruxolitinib, which blocks JAK1 and JAK2 activity (36) and tofacitinib, which inhibits JAK3 (37). We calibrated drug concentrations in a CXCL12-triggered OT-II CD4⁺ T cell migration assay. Whereas ruxolitinib treatment (1.0 μM, 60 min, 37°C) reduced cell migration by 50%, tofacitinib had no effect at any concentration tested (0.3, 1.0, or 3.0 μM, 60 min, 37°C) (Supplemental Fig. 3A). We thus formed conjugates using untreated, ruxolitinib (1 μM)-, or tofacitinib (1 μM)-treated OT-II CD4⁺ T cells, and OVA peptide–loaded BM-DC. γ-Tubulin staining showed significantly less MTOC translocation to the contact zone only in ruxolitinib-treated cells (Fig. 4C).

To rule out nonspecific effects because of the use of chemical inhibitors, we performed the remaining experiments using nucleofected OT-II CD4⁺ T cells with siRNA pools for JAK1/2 (siJAK1,2) or controls (siControl) (10). Western blot analysis confirmed a sharp reduction in protein levels for both kinases (>60%; Supplemental Fig. 3B) in siJAK1,2 T cells. We also observed abrogation of CXCL12-mediated Vav-1 phosphorylation (Fig. 4D) and reduced Rho activation (Fig. 4E) in the siJAK1,2 T cells.

To confirm the role of JAK in chemokine-mediated stabilization of the IS, we formed conjugates using siJAK1,2 OT-II CD4⁺ T cells and OVA peptide–loaded BM-DC (30 min, 5:1 ratio, 37°C). Whereas phalloidin staining showed F-actin accumulation cytoskeletal rearrangement at the contact zone in 94.7 ± 2.5% of siControl OT-II CD4⁺ cells, F-actin accumulation was reduced in JAK-deficient CD4⁺ T cells (41.3 ± 3.0%) (Fig. 5A). γ-Tubulin staining indicated a significant alteration in MTOC translocation to the contact zone in JAK-deficient cells (Fig. 5B). Finally, time-lapse videomicroscopy analysis (Supplemental Videos 4, 5) showed that JAK1/2 deficiency significantly increased the percentage of short-lived cell contacts (<5 min; Fig. 5C) and reduced mean contact duration (4 ± 0.3 min; Fig. 5D, 5E). To control biological specificity, we repeated the experiments using OT-II CD4⁺ T cells nucleofected with a siRNA pool for Tyk2 (siTyk2). These cells showed >90% Tyk2 protein in Western blot with a specific Ab (Supplemental Fig. 3C). siTyk2 cells showed no alteration in MTOC translocation to the contact zone when conjugated with OVA peptide–loaded BM-DC (Fig. 5F).

Detailed analysis of IS structure on siControl and JAK-deficient T cells showed that JAK1 and JAK2 are essential for correct IS formation. siJAK1,2 and siControl OT-II CD4⁺ T cells were allowed to conjugate with stimulating anti-CD3–coated beads plus soluble CXCL12 (50 nM) (Fig. 6). In the three-dimensional reconstruction of the IS-like area, we observed an increase in talin and CXCR4 colocalization at the pSMAC after CXCL12 addition to the anti-CD3–coated beads (histograms in Fig. 6A). Lack of siJAK1,2 nonetheless prevented correct localization of these molecules; their redistribution at the pSMAC was abolished by JAK1/2 knockdown in the presence of CXCL12 and anti-CD3 (pSMAC/cSMAC ratio < 1; graphs in Fig. 6A, 6B). Similarly, CXCL12 plus anti-CD3–mediated protein kinase Cθ redistribution to the contact site was abolished in siJAK1,2 T cells (Supplemental Fig. 4A). JAK1,2-deficient naive OT-II CD4⁺ T cells conjugated with BM-DC showed a marked reduction in intracellular PTyr staining when evaluated by flow cytometry and by immunofluorescence (Fig. 7A, 7B). These defects led to reduced T cell activation, as shown by lower IL-2 mRNA (Fig. 7C) as well as CD69 levels (Fig. 7D, Supplemental Fig. 4B). The results indicate that, through both Gi and JAK signaling pathways, CXCL12 regulates integrin activation and IS stabilization.

For T cell activation, cell membrane and cytoplasmic proteins must be rearranged to form the immunological synapse (32). IS formation depends on TCR-mediated signals that, in concert with costimulatory signals, cause the T cell cytoskeleton, membrane receptors, and certain signaling effectors to polarize toward the APC/T cell interface. Synapse-associated signaling also leads to spatial segregation into organized clusters of the TCR, CD28, LFA-1, and other surface molecules such as chemokine receptors (2). Some reports regard TCR recognition of specific Ags as the starting point of organelle redistribution and IS formation (38–40), and contact duration is thought to control efficient T cell activation in vivo (41). Although guidance of leukocyte trafficking is the principal role of chemokines, they also influence T cell recognition of Ag. Chemokines direct T cell tethering to APC, which facilitates scanning and formation of stable Ag-dependent interactions (34). In addition, chemokines promote cell polarization, which increases sensitivity to Ag (42) and costimulate T cell function (5, 43).

In this study, we show that chemokines also influence IS formation and T cell activation. Membrane CXCR4 concentrates in the IS periphery; drebrin has a function in this relocation and in F-actin reorganization (6). When we blocked CXCL12 binding to CXCR4 or its signaling pathways, IS structure and duration as well as T cell activation were altered. We detected CXCL12 at the BM-DC surface but not on OT-II CD4⁺ T cells. Our in vitro experiments using anti-CD3–coated beads nonetheless indicated that soluble CXCL12 also influences IS formation. The anti-CD3–coated bead model is often used for T cell activation analysis as it provides the geometry of the cell–cell contact with the APC and, because of the control of the stimuli used, permits dissection of the molecular pathways involved (44, 45). Although these experiments were performed in the absence of integrin ligands, we detected CXCR4 and talin in the pSMAC. Anti-CD3 Ab promotes cytoskeletal organization through Vav-1 and talin complexes (46), and Vav-1 is also involved in CXCL12-mediated talin interaction with β₁ integrins (16). Talin binding to the cytosplasmic tails of the integrins is thus one of the inside–outside mechanisms that activates integrins and enables their interaction with adhesion molecules.

IS were also formed between BM-DC prepared from CXCL12^Gagtm/Gagtm mice and T cells. These mice have a mutant CXCL12 that cannot interact with GAG (20), impeding its retention at the cell surface. In this mouse model, we nonetheless detected a significant but small defect in MTOC translocation to the contact site, suggesting that soluble CXCL12 is less efficient than GAG-bound CXCL12. Although chemokines act as soluble mediators in vitro, in vivo they are thought to bind GAG at the cell surface (47, 48). These results coincide with data that suggest various roles for chemokines in T cell recognition of APC. Whereas some dominant chemokines such as CCL19 or CCL21 increase T cell motility on DC monolayers (49) and mediate direction and tethering (34), others like CXCL12 might stabilize cell–cell contacts. IS stability is impaired in T cells bearing warts, hypogammaglobulinemia, infections, and myelokathexis syndrome–mutant CXCR4 because of defects in receptor recruitment to the cell–cell contact area (50).

We found defective MTOC translocation in CXCR4-deficient T cells or those with altered CXCR4 function. The MTOC translocation mechanism underlies IS formation, sustains T cell signaling (29), and increases mitochondrial localization at the cell–cell contact site (51). MTOC positioning during T cell activation facilitates correct effector functions such as cytokine secretion (52) and target cell killing (53, 54). T cell treatment with the inhibitors AMD3100 or PTx or with reduced JAK1 and JAK2 expression levels significantly diminished conjugate duration; these defects might explain the IL-2 production defect as well as the reduced T cell activation and proliferation in these conditions. In T cells, CXCR4 interaction with the TCR–CD3 complex induces ZAP-70 and SLP-76 recruitment, ERK pathway activation, and cytokine synthesis (55–57). Blockade of CXCL12-mediated signaling in T cells during IS formation correlated with less cell activation, as indicated by reduced levels of IL-2 mRNA and of activation markers.

LFA-1 is reported to enhance T cell activation (58–60). Cell adhesion amplifies TCR signaling via several mechanisms, including prolonged inositol phospholipid hydrolysis (61), replenishment of intracellular Ca²⁺ stores (61), and cAMP increase (62). Chemokine-mediated integrin activation can thus contribute to IS stabilization. We nonetheless observed that PTx treatment did not completely block CXCL12-mediated RhoA activation and Vav-1 phosphorylation in a key signaling pathway involved in integrin activation by chemokines (16), which suggests G protein–dependent and –independent mechanisms. An earlier study indicated that although chemokines increase basal T cell adhesion to DC, the effect on Ag-pulsed T cells is not PTx sensitive (63). In primary T lymphocytes, we previously showed that chemokines control LFA-1– and VLA-4–mediated cell adhesion, both key processes in T cell homing (10) that involve chemokine-mediated JAK activation. A recent report implicates JAKs in CXCL12-mediated integrin activation and lymphocyte trafficking, via RhoA, Rac1, and Rap1 (17). Our findings now link JAK activation to CXCL12-mediated IS stabilization through the Vav-1/RhoA pathway. We cannot entirely rule out a JAK3 function because our experiments to evaluate the role of this kinase were only performed using a chemical inhibitor, tofacitinib. Our results nonetheless show that JAK1 and JAK2 inhibition or knockdown in T cells altered IS formation, reduced conjugate number, contact duration, and T cell activation.

Adhesion molecules, chemokines, and costimulatory and coinhibitory receptors all take part in the complex process of tuning TCR signaling and T cell activation thresholds. Whereas initiation of integrin-mediated cell adhesion is Ag dependent, our data suggest that through Gi and JAK signaling, CXCL12 helps to maintain the integrin activation at the pSMAC needed for continuous TCR signaling (Fig. 8), a finding with broad implications for cell contact–dependent communication in adaptive immunity.

We thank Drs. P. Cohen for the JAK inhibitors and F. Arenzana-Seisdedos for Cxcl12^Gagtm mice and C. Bastos for secretarial and C. Mark for editorial assistance.

The authors have no financial conflicts of interest.