Homing of CD8+ T lymphocytes to the tumor microenvironment is an important step for mounting a robust antitumor immune response. TGFβ is responsible for CD103 (αEβ7) integrin induction in activated intraepithelial CD8+ T lymphocytes. However, the interplay between TGFβ and CD103 and their contribution to T-cell infiltration and antitumor activity remain unknown. Here, we used viable human lung tumor slices and autologous tumor antigen-specific T-lymphocyte clones to provide evidence that CD103 is directly involved in T-lymphocyte recruitment within epithelial tumor islets and intratumoral early T-cell signaling. Moreover, TGFβ enhanced CD103-dependent T-cell adhesion and signaling, whereas it inhibited leukocyte function-associated antigen (LFA)-1 (αLβ2) integrin expression and LFA-1-mediated T-lymphocyte functions. Mechanistic investigations revealed that TGFβ bound to its receptors (TGFBR), which promoted the recruitment and phosphorylation of integrin-linked kinase (ILK) by TGFBR1. We further show that ILK interacted with the CD103 intracellular domain, resulting in protein kinase B (PKB)/AKT activation, thereby initiating integrin inside-out signaling. Collectively, our findings suggest that the abundance of TGFβ in the tumor microenvironment may in fact engage with integrin signaling pathways to promote T-lymphocyte antitumor functions, with potential implications for T-cell-based immunotherapies for cancer. Cancer Res; 76(7); 1757–69. ©2016 AACR.
Adequate positioning of tumor antigen-specific T cells into tumor tissues plays a central role in an effective antitumor immune response. Hence, the number of CD8+ T cells found in contact with tumor cells is usually associated with good outcome in cancer patients (1). Recruitment of T lymphocytes within the tumor involves multiple sequential steps regulated by molecular interactions with several stromal cell types and coordinated by chemotactic agents, cell surface adhesion molecules, and extracellular matrix (ECM) proteins. Integrins are one of the major families of adhesion molecules that mediate cell–ECM and cell–cell interactions (2). Adhesion of T cells to ECM proteins or cellular integrin ligands is fundamental for lymphocyte development, migration, extravasation, and activation (3, 4). Among the integrin family members, LFA-1 (αL/β2, CD11a/CD18) is essential for lymphocyte homing in lymphoid and nonlymphoid tissues, transmigration, and antigen-specific T-cell responses (3). On naive T lymphocytes, LFA-1 has weak affinity for its ligands, the intercellular adhesion molecules (ICAM). However, T-cell stimulation through T-cell receptor (TCR) or chemokine receptors triggers an “inside-out” signal that results in LFA-1 activation by inducing integrin-extended conformation and clustering, thereby increasing its affinity for its ligands (5). Firm adhesion of LFA-1 to ICAM-1 (CD54) initiates an “outside-in” signal that has costimulatory functions in TCR signaling, thus contributing to T-cell stimulation, spreading, and cytotoxicity (6, 7).
The αE(CD103)β7 integrin, hereafter referred to as CD103, plays a crucial role in TCR-mediated cytokine secretion and cytotoxic activity toward epithelial tumors by interacting with its ligand, epithelial cell marker E-cadherin, on target cells (8, 9). This integrin is essential in controlling CD8+ tumor-infiltrating lymphocyte (TIL) activities, not only by promoting effector T-cell adhesion to tumor cells, but also by triggering intracellular signaling events that costimulate TCR signals (10). A role for CD103 in T-cell homing into epithelia has been suggested (11, 12); however, accumulating evidence indicates a role for CD103 in retention of T-lymphocyte subpopulations in epithelial tissues (13). Consistently, CD103 mediates arrest of T lymphocytes by interacting with E-cadherin on epithelial tumors (9) and recruiting CCR5 at the immune synapse (IS) with cancer cells (14). Moreover, an enhanced CD103+ TIL subset correlates with improved patient survival in ovarian (15) and lung (16) carcinomas and increased intraepithelial lymphocyte infiltration, suggesting that CD103 promotes recruitment of TIL within epithelial tumor islets. However, the true contribution of CD103 to promoting T-cell infiltration into tumor regions, and the influence of TGFβ, abundant within the tumor microenvironment and responsible for ITGAE gene expression in TCR-engaged CD8+ T lymphocytes (8, 17, 18), on integrin functions remain poorly understood. Using viable human tumor slices (19) and autologous tumor antigen-specific CTL clones, we show here that CD103 promotes T-cell recruitment within epithelial tumor regions and enhances intratumoral T-cell early signaling. We also show that TGFβ upregulates CD103-dependent T-cell effector functions by triggering TGFBR1 (activin receptor-like kinase-5, ALK5)-mediated phosphorylation of ILK and its subsequent binding to CD103, thus initiating inside-out signaling leading to activation of the integrin and strengthening of CD103-E-cadherin adhesion.
Materials and Methods
Tumor cell lines, T-cell clones, and freshly isolated TIL
IGR-Heu and IGR-Heu-ICAM-1 tumor cell lines were derived in one of our laboratories in 1996 as previously described (8). Heu171 (CD103+/LFA-1+) and H32-22 (CD103−/LFA-1+) T-cell clones were generated from autologous TIL and peripheral blood lymphocytes (PBL), respectively. Cells were regularly tested and authenticated by immunofluorescence analysis and cytotoxic assay including in this work. Fresh CD8+ TIL were isolated using a FACS ARIA cell sorter.
Recombinant molecules, antibodies, and chemical inhibitors
Recombinant (r) E-cadherin-Fc (rE-cadh-Fc), rICAM-1-Fc and rFibronectin-Fc molecules were provided by R&D Systems. rCXCL12 and rTGFβ was purchased from PeproTech and Abcys, respectively. Anti-CD103 and anti-β2 mAb were purchased from Beckman Coulter, and PE-conjugated anti-CD103 from eBiosciences. Anti-CD3 (UCHT1) was provided by Ozyme and anti-CD8 by BioLegend. Anti-ILK and anti-phosphorylated (phospho)-ILK (Ser-246) mAb were purchased from Millipore. Anti-TGFBRI and secondary mAb were provided by Santa Cruz Biotechnology. Anti-AKT/PKB and anti-phospho-AKT were purchased from Cell-Signaling. Anti-KLRG1 was a gift from H Pircher. TGRBR1 kinase inhibitor SB-431542 (SB) and PI3K inhibitor Wortmannin were supplied by Sigma-Aldrich. Phospho-AKT inhibitor MK-2206 was purchased from Millipore and ILK inhibitor QLT-0267 (QLT) from Quadra Logic Technology.
Confocal microscopy analyses
For T-cell adhesion, poly-l-lysine slices were coated with rE-cadh-Fc, rICAM-1-Fc, or rFibronectin-Fc (5 μg/mL) overnight at 4°C. T cells, nonstimulated or stimulated with rCXCL12 (100 nmol/L) or rTGFβ (2 ng/mL), were incubated on precoated slices as described (8, 9). Cells shape index was calculated as the ratio of the longer to the shorter axis measured via Image J software. For protein polarization, lymphocytes, untreated or pretreated with SB (1 μmol/L), QLT (20 μmol/L), or with rTGFβ were preincubated with IGR-Heu cells, or rE-cadh-Fc-coupled protein G-Dynabeads, at a 2:1 effector to target cell (E:T) ratio and then plated on poly-(l-lysine)-coated coverslips. Cells were stained with mouse anti-CD103, rabbit anti-phospho-ILK, or goat anti-TGFBRI mAb, followed by anti-mouse-Alexa Fluor-546, anti-rabbit-Alexa Fluor-647 or anti-goat-Alexa Fluor-488 Ab, respectively. Coverslips were analyzed using a fluorescence microscope with ×20 (adhesion) or ×63 (spreading and protein relocalization) lenses as described (9).
Tumor slices and T-cell recruitment experiments
Xenograft tumor samples were embedded in low-gelling-temperature agarose as reported (19). Lymphocytes were stained with SNARF and added onto the cut surface of each slice. In some experiments, T cells were preincubated with neutralizing anti-CD103 (anti-αE, 10 μg/mL), anti-β2 (10 μg/mL), or anti-CD3 (1 μg/mL), or treated with rTGFβ. Slices were then stained with anti-E-cadherin mAb followed by a secondary Ab (Alexa Fluor-488). Images were acquired with a ×4 or ×10 (S fluor; Nikon) objective and MetaVue imaging software (Universal Imaging).
Single-cell calcium measurement
The calcium response was measured in the same tumor slice with T cells loaded with 1 μmol/L Fura-2AM (Molecular Probes). One of the two clones was also labeled with 5-chloromethylfluorescein diacetate (CMFDA; 5 nmol/L). Fura-2AM-loaded lymphocytes and infiltrated into tumor slices were alternatively excited every 10 s at 350 and 380 nm on an inverted Eclipse TE300 microscope equipped with a ×20 objective and a Metafluor imaging system. Emissions at 510 nm were used for the analysis of Ca2+ responses. Values were represented as a ratio: fluorescence intensity at 350 nm/fluorescence intensity at 380 nm.
Western blot and immunoprecipitation experiments
Lymphocytes, untreated or pretreated with rTGFβ, were unstimulated or stimulated with plastic-coated rE-cadh-Fc. Equivalent amounts of protein extracts were separated by SDS-PAGE and transferred to a nitrocellulose membrane as described (10). Blots were then probed with rabbit anti-phospho(Ser-246)-ILK, anti-phospho(Ser-473)-AKT/PKB, anti-AKT/PKB or anti-β-actin-peroxidase followed by secondary HRP-conjugated Ab.
For immunoprecipitation, T-cell extracts were incubated for 2 h at 4°C with cross-linked anti-TGFBRI or anti-CD103 beads, or protein G-Sepharose beads prebound with control goat or mouse IgG. Beads were then washed and eluted with reducing agent at 95°C.
Flow cytometric analysis and cytotoxicity assay
For granule exocytosis, T cells, untreated or pretreated with inhibitors or cultured with rTGFβ, were stimulated with a combination of rE-cadh-Fc or rICAM-1-Fc (2.5 μg/mL) and UCHT1 mAb (0.4 μg/mL) in the presence of anti-CD107 mAb and monensin A (10). A high concentration of anti-CD3 (10 μg/mL) was used as a control. Cytotoxicity was evaluated with a conventional 51Cr-release assay (8).
Statistical analyses were performed using the two-tailed Student t-test. Two groups were considered as significantly different if P < 0.05.
Regulation of CD103-dependent T-cell adhesion and motility by TGFβ
Thus far, little is known about the involvement of CD103 in T-cell migration and regulation of integrin activities by TGFβ. Hence, experiments were performed to compare the effect of TGFβ on T-cell adhesion and the migratory behavior of the CD103+/LFA-1+ TIL clone Heu171 (8) on immobilized human rE-cadh-Fc or rICAM-1-Fc. The rFibronectin-Fc was included as a negative control. rCXCL12, rCCL5, and rCCL20 chemokines, the respective receptors (CXCR4, CCR5, and CCR6), of which are expressed on the T-cell clone surface, were used as positive controls. Indeed, chemokines can trigger integrin-dependent adhesion of lymphocytes by initiating an inside-out signaling leading to integrin activation (20, 21). Results indicated that T lymphocytes stimulated with rCXCL12 (Fig. 1A), rCCL5, or rCCL20 (Supplementary Fig. S1A), adhered more efficiently to both rE-cadh-Fc and rICAM-1-Fc than untreated cells. Of note, incubation of T cells for a short period of time (30 min) with the three chemokines had no effect on LFA-1 and CD103 integrin expression levels (data not shown). Remarkably, treatment of T lymphocytes with rTGFβ induced a sharp increase in T-cell adhesion to rE-cadh-Fc starting from 24 h of culture whereas it inhibited lymphocyte adhesion to rICAM-1-Fc (Fig. 1A). Treatment with rTGFβ of CD8+ TIL freshly isolated from three independent tumors also induced an increase in T-cell adhesion to E-cadherin-Fc whereas it inhibited adhesion to rICAM-1-Fc (Fig. 1B). Opposing effects of TGFβ on the adhesion strength of CD103+/LFA-1+ TIL clone and CD103−/LFA-1+ PBL counterpart (clone H32-22) to autologous E-cadherin+/ICAM-1− (IGR-Heu) and E-cadherin+/ICAM-1+ (IGR-Heu-ICAM-1) tumor cells, respectively, were also observed under flow conditions (Supplementary Fig. S1B). These results correlated with upregulation of CD103 expression and, in contrast, downregulation of LFA-1 following T-cell incubation with the cytokine (Supplementary Fig. S2A). Notably, T-cell clones, untreated or pretreated with rTGFβ, failed to express detectable levels of the E-cadherin inhibitory receptor KLRG1 (data not shown).
Chemokines generate both pro- and anti-adhesive intracellular signaling events, whose equilibrium is relevant to cell movement (21). Therefore, we analyzed morphological changes in CD103+/LFA-1+ lymphocytes cultured on recombinant molecule-coated surfaces in the absence or presence of rCXCL12, rCCL5, or rCCL20. Compared to untreated T cells, rCXCL12-treated cells overlaid on rE-cadh-Fc or rICAM-1-Fc displayed more active motility behavior characterized by greater polarized cell morphology (Fig. 1C). Indeed, an increase in the percent of cells with larger protrusion sizes was obtained when T lymphocytes were stimulated with rCXCL12 and plated on either of the molecules. Similar results were observed with rCCL5 and rCCL20 (Supplementary Fig. S2B). In contrast, starting from 24 h of treatment, TGFβ induced a strong increase in lymphocyte locomotion behavior when T cells were exposed to rE-cadh-Fc, but not to rICAM-1-Fc (Fig. 1C). In addition, live-microscopy imaging revealed that motility was enhanced when T cells were pretreated with rTGFβ before culturing them on rE-cadh-Fc, whereas it was compromised when the same pretreated cells were added onto rICAM-1-Fc (Supplementary Fig. S2C and Movies). These results suggest that TGFβ promotes T-cell adhesion and cytoskeleton rearrangement by strengthening the interaction of CD103 with E-cadherin.
TGFβ enhances CD103-mediated T-cell recruitment in epithelial tumor islets
Next, experiments were undertaken to assess the influence of CD103 on T-cell recruitment in tumor regions via our previously described approach using tumor slices (19) obtained from E-cadherin+ lung tumor cell line, untransfected or transfected with ICAM-1, engrafted into immunodeficient NOD-SCID mice. Fluorescent-labeled autologous CD103+/LFA-1+ and CD103−/LFA-1+ CTL were overlaid onto fresh tumor slices generated from either E-cadherin+/ICAM-1− or E-cadherin+/ICAM-1+ tumors. Slices were stained with anti-E-cadherin mAb to distinguish epithelial from stromal regions (Supplementary Fig. S3A), and then lymphocytes that infiltrated epithelial tumor regions and the stroma were counted. Results indicate that the CD103+/LFA-1+ TIL clone was much more efficiently recruited within E-cadherin+/ICAM-1− tumor islets than the CD103−/LFA-1+ PBL clone, which was concentrated within stromal regions (Fig. 2A). Moreover, treatment of the CD103+/LFA-1+ clone with TGFβ enhanced T-cell recruitment in tumor regions starting from 24 h of preincubation. In contrast, treatment of the CD103−/LFA-1+ PBL clone with TGFβ alone was unable to induce CD103 expression (8) and thus it had no effect on T-cell recruitment within tumor islets (data not shown). Recruitment of the TIL clone in tumor regions was dependent on CD103 and TCR engagement, because neutralizing anti-CD103 and anti-CD3 mAb dramatically blocked lymphocyte migration by inhibiting T-cell adhesion to E-cadherin+ target cells and peptide–MHC–I complex recognition by TCR, respectively. In contrast, neutralizing anti-β2 mAb, used as a negative control, had no effect on T-cell distribution (Fig. 2A).
We then analyzed the location of both T-cell clones plated on E-cadherin+/ICAM-1+ tumor slices and the effect of TGFβ on this process. Transfection of tumor cells with ICAM-1 only slightly increased recruitment of the CD103+/LFA-1+ clone in tumor islets, whereas it strongly enhanced CD103−/LFA-1+ clone migration (Fig. 2B). Moreover, anti-β2 and anti-CD3 mAb strongly inhibited PBL clone recruitment, whereas anti-CD103, anti-β2, or a combination of both mAb blocked TIL clone migration toward tumor regions. Notably, pertussis toxin, a bacterial toxin that blocks signaling from G protein–coupled receptors including chemokine receptors had no effect on T-cell distribution (data not shown). In contrast, treatment of the CD103−/LFA-1+ clone with TGFβ resulted in profound inhibition of T-cell recruitment within E-cadherin+/ICAM-1+ tumor islets, in particular after 96 h of preincubation (Fig. 2B). These results strongly suggest that CD103 is directly involved in T-lymphocyte recruitment in tumor islets and that TGFβ upregulates its functions, whereas it has opposing effects on LFA-1.
TGFβ potentiates intratumoral early signaling of CD103+ lymphocytes
Imaging of the intracellular calcium increase is a powerful means of monitoring the dynamics of T-lymphocyte activation. We therefore evaluated the role of CD103 and the influence of TGFβ on the Ca2+ response triggered in CTL CD103+/LFA-1+ and CD103−/LFA-1+. T-cell clones, untreated or pretreated with rTGFβ, were labeled with Fura-2 (red) to monitor calcium response. The CD103−/LFA-1+ clone was also labeled with CMFDA fluorescent dye (green; thus appearing in yellow) in order to identify each T-cell population on the same slice. The two clones were then added simultaneously to tumor slices, untransfected or transfected with ICAM-1. Results indicated that CTL CD103+/LFA-1+, added to the E-cadherin+/ICAM-1− slice and recruited within tumor islets, displayed a much stronger calcium response than CTL CD103−/LFA-1+ (Fig. 3A). Moreover, neutralizing anti-CD103 induced a decrease in the calcium level of the TIL clone, whereas anti-β2 mAb, used as a negative control, had no effect. Remarkably, treatment of the TIL clone with rTGFβ induced a sharp increase in the T-cell calcium response. An increase in CD103+/LFA-1+ Ca2+ response was also observed when the T-cell clone was pretreated with TGFβ and plated on a monolayer of E-cadherin+/ICAM-1− IGR-Heu tumor cells, which was inhibited when T lymphocytes were preincubated with neutralizing anti-CD103 mAb (Supplementary Fig. S3B). In contrast, treatment of the CD103−/LFA-1+ PBL clone with TGFβ alone failed to induce CD103 expression and thus a marginal calcium response was observed when T cells were added to E-cadherin+/ICAM-1− tumor cells (data not shown).
We then monitored Ca2+ signaling of both clones plated on E-cadherin+/ICAM-1+ tumor slices. As expected, a strong calcium response was obtained with the CD103+/LFA-1+ clone, which was inhibited by both anti-CD103 and anti-β2 blocking mAb. Transfection of tumor cells with ICAM-1 induced a strong Ca2+ increase in the CD103−/LFA-1+ clone, which was blocked by neutralizing anti-β2. Importantly, treatment of the PBL clone with TGFβ strongly inhibited early T-cell signaling (Fig. 3B). These results suggest a crucial role of CD103 in regulating TCR-mediated activities within a TGFβ-rich tumor ecosystem and that TGFβ is involved in the CD103 signaling pathway.
TGFβ regulates CD103 activation via ILK phosphorylation
The signaling pathway of CD103 and its contribution to T-cell migration and functions within epithelial tissues are unknown. ILK, a serine/threonine kinase, has been reported to mediate integrin signaling and to play a role in leukocyte recruitment (22) and T-cell chemotaxis (23). Moreover, a role for TGFβ in regulating ILK expression and activation has been reported (24). We therefore analyzed the involvement of ILK in CD103 signaling and the effect of TGFβ on integrin activation in CD103+/LFA-1+ CTL. Initial experiments indicated that TGFβ upregulated ILK gene transcription starting from 6 h of T-cell clone treatment (Supplementary Fig. S3C). Moreover, stimulation of the clone with immobilized rE-cadh-Fc and/or rTGFβ triggered phosphorylation of ILK at Ser 246 (Fig. 4A), which was strongly inhibited by the ILK inhibitor QLT, the TGFBR1 inhibitor SB, the PI3K inhibitor Wortmannin (Fig. 4B), and anti-CD103 mAb (Supplementary Fig. S3D). Notably, inhibition of ILK phosphorylation resulted in a marked decrease in CTL adhesion on immobilized rE-cadh-Fc (Supplementary Fig. S3E). Stimulation of the TIL clone with rE-cadh-Fc also induced phosphorylation of the PKB/AKT at Ser-473, which was strongly inhibited by the AKT inhibitor MK-2206 and QLT, SB, and Wortmannin (Fig. 4C). Of note, incubation of lymphocytes for a short period of time (1 h) with all inhibitors had no effect on integrin expression level and T-cell viability, as checked by CD103 and annexin V labeling, respectively (data not shown).
Next, we evaluated the consequences of CD103 engagement with its ligand on ILK localization in CD103+/LFA-1+ TIL plated on a rE-cadh-Fc-coated slide. Results indicate that the CD103–E-cadherin interaction resulted not only in T-cell spreading, but also in recruitment of phospho-ILK, together with CD103, at the leading edge of migrating CTL presenting a high actin cytoskeleton rearrangement (Fig. 5A). Moreover, ILK and TGFBR1 inhibitors induced a marked decrease in phospho-ILK recruitment at the leading edge. Together, these results suggested that TGFBR1 serine kinase catalyzed, even in the absence of its ligand, phosphorylation of ILK through a PI3K-dependent pathway and that phospho-ILK and phospho-AKT were involved in CD103 signaling.
ILK recruitment and phosphorylation by TGFBR1 triggers binding to CD103
Because the interaction of ILK with β-integrins is important for intracellular signaling, we first monitored the localization of phospho-ILK and TGFBR1 in conjugates formed between CD103+/LFA-1+ T cells and E-cadherin+/ICAM-1− tumor cells. Results revealed polarization of phospho-ILK, together with CD103 and TGFBR1, at the IS, which was inhibited by QLT and SB (Fig. 5B), as well as siRNA targeting E-cadherin (Supplementary Fig. S4A). Consistently, shRNA targeting ILK also inhibited CD103 polarization (Supplementary Figs. S4B and S4C) and AKT phosphorylation (data not shown). In contrast, treatment of the TIL clone with rTGFβ induced an increase in synaptic relocalization of CD103, phospho-ILK, and TGFBR1 (Supplementary Fig. S4D).
To explore a potential link between TGFBR1, phospho-ILK, and CD103, we searched for physical interactions between these proteins. Therefore, we performed immunoprecipitation experiments using anti-CD103 or anti-TGFBR1 mAb and immunobloting with anti-ILK. Results revealed that ILK bound to CD103 (Fig. 5C) and TGFBR1 (Fig. 5D), and that TGFβ, and to a lesser extent rE-cadh-Fc, enhanced ILK linking to CD103. Notably, no direct interaction between CD103 and TGFBR1 was observed, because CD103 was detected only in the supernatant of immunoprecipitation with anti-TGFBR1 (Fig. 5E). These results further support the hypothesis that the interaction of TGFβ with its receptors induces recruitment and phosphorylation of ILK by TGFBR1 and its subsequent binding to the CD103 intracellular domain, resulting in AKT phosphorylation and thereby initiating integrin inside-out signaling.
TGFβ regulates CD103-dependent CTL activities via PI3K–ILK–AKT pathway
Experiments were then conducted to evaluate the contribution of PI3K–ILK–AKT pathway in CD103-mediated T-cell functions. Using the tumor slice system, results indicated that Wortmannin, SB, QLT, and MK-2206 inhibited T-cell recruitment in epithelial tumor islets (Fig. 6A). We then analyzed the effect of all the inhibitors and the influence of TGFβ on CD103-dependent T-cell clone degranulation. As shown in Fig. 6B, CD103+/LFA-1+ T cells cultured with rTGFβ and stimulated with a combination of anti-CD3 plus rE-cadh-Fc expressed higher surface levels of CD107a than T cells exposed to a combination of anti-CD3 and rICAM-1-Fc. Moreover, Wortmannin, SB, QLT, and MK-2206 induced a decrease in T-cell degranulation triggered by a combination of anti-CD3 plus rE-cadh-Fc, whereas they had no effect on CD107a expression induced by a high dose of anti-CD3. Consistently, SB and QLT strongly inhibited cytotoxic granule polarization at the contact area between CD103+/LFA-1+ CTL and rE-cadh-Fc-coupled beads (Supplementary Fig. S5A). In addition, although TGFβ inhibited granzyme B polarization at the contact area between the TIL clone and rICAM-Fc-coupled beads, it had no effect on serine protease relocalization at the contact zone with rE-cadh-Fc-coated beads.
We then evaluated the effect of TGFβ, PI3K, TGFBR1, phospho-ILK, and phospho-AKT inhibitors on T-cell clone-mediated cytotoxicity toward target cells. Incubation of the CD103+/LFA-1+ clone with rTGFβ enhanced E-cadherin+/ICAM-1− IGR-Heu tumor cell killing (Fig. 6C), which was inhibited in the presence of neutralizing anti-CD103 (Fig. 6D), whereas Wortmannin, SB, QLT, and MK-2206 inhibited target cell lysis (Fig. 6C). Notably, all inhibitors had no effect on T-cell clone-mediated CD3-redirected cytotoxicity of the P815 target (Supplementary Fig. S5B). In contrast, culture of the CD103−/LFA-1+ CTL clone in the presence of TGFβ dramatically inhibited E-cadherin+/ICAM-1+ IGR-Heu-ICAM-1 target cell lysis (Fig. 6E). Notably, TGFβ had no effect on granzyme B expression levels in both T-cell clones (data not shown). These results further emphasize the role of PI3K, TGFBR1, phospho-ILK, and phospho-AKT/PKB in CD103 signaling and the positive effect of TGFβ in antitumor CTL-mediated activities.
In this report, we identified CD103 and its ligand E-cadherin as important adhesion molecules stimulating recruitment and effector functions of activated CD8+/CD103+ TIL inside human lung tumors. In agreement with our previous studies, we also show that CD8+/CD103− T cells accumulate preferentially within the tumor stroma and display inefficient antitumor activities (19). Along the same line, it has been reported that mouse CD103−/− T cells failed to infiltrate allograft islets, suggesting that CD103 promotes intragraft migration of CD8+ effectors into epithelial islets (25). Our data are consistent with a key role for CD103 in promoting entry of tumor antigen-specific CD8+ T cells into epithelial tumor islets and in triggering antitumor T-cell functions. Recognition of tumor antigen by TCR also plays an important role, because anti-CD3 mAb inhibited T-cell recruitment into tumor regions. It is possible that CD103 contributes to intratumoral scanning of target cells for cognate antigen recognition.
CD103 and LFA-1 are the predominant integrins expressed by TIL and both are likely involved in T-cell adhesion and migration within and toward epithelial tumors, respectively. Indeed, we show here that LFA-1, through an interaction with ICAM-1, also induced T-cell infiltration within tumor nests. However, tumors often express low levels of ICAM (26) and produce high amounts of TGFβ (27), thus altering LFA-1 expression and functions (ref. 28 and this study). Indeed, this cytokine shifted the integrin expression profile of tumor-specific TIL from LFA-1high/CD103low to LFA-1low/CD103high. Thus, CD103 most likely corresponds to an important integrin for adjusting T-cell adhesion and migratory potential in the tumor ecosystem, where TGFβ is abundant. Along this line, it has been reported that murine CD103 determines cell shape and motility, and confers the ability to move in an actin-polymerization-dependent manner (29). These results suggest that CD103-mediated cell movement probably involves signaling interactions with components of the cytoskeleton. Likewise, we show here that human CD103 confers the capacity to CD8+ T cells to form protrusions/filopodia in an E-cadherin–dependent fashion. Moreover, TGFβ strongly optimizes T-cell adhesion and motility by enhancing CD103 expression levels and initiating intracellular signals, resulting in integrin activation.
TGFβ is known to inhibit cell growth in benign cells but promotes progression in cancer cells (30). Moreover, TGFβ signaling in peripheral T lymphocytes is essential for tolerance and homeostasis by restraining TCR activation-dependent Th1, cytotoxic, and NKT cell differentiation program (31, 32). This cytokine, produced by stromal and tumor cells, is also described as an immunosuppressive factor used by malignant cells to escape from the immune response and to compromise CTL activities (33). Paradoxically, TGFβ is involved in promoting antitumor CD8+ T-cell effector functions within epithelial tumors by inducing CD103 upon TCR engagement and by adjusting its expression levels. TGFβ also plays an essential role in the formation and long-term persistence of tissue-resident memory T cells (TRM), at least in part via induction of CD103 (34). Importantly, our data indicate that TGFβ also contributes to T-cell migration toward epithelial tumor islets and subsequent cancer cell destruction. This paradoxical effect of TGFβ on T-cell functions is most likely associated with the use in this study of CD103+/LFA-1+ lymphocytes whereas most of the previous studies were based on peripheral T cells, which generally express a CD103−/LFA-1+ phenotype. Consistently, we show here that TGFβ is responsible for arrest of LFA-1-dependent T-cell migration within stromal regions by preventing integrin expression and functions. It has been reported that TGFβ inhibits phosphorylation of Erk and, thereby, LFA-1–mediated T-cell adhesion and chemotaxis (28, 35). In contrast, this cytokine contributes to CD103 signaling by upregulating ILK expression and inducing its phosphorylation. TGFβ initiates signaling by binding to TGFBR, leading to formation of a hetero-tetrameric complex composed of TGFBR1 and TGFBR2 dimers, in which TGFBR2 catalyzes phosphorylation of the cytoplasmic domain of TGFBR1 (36). In addition to the Smad pathway, TGFβ utilizes multiple non-Smad pathways, including PI3K/AKT pathway, to regulate a wide range of cell functions (37). We show here that TGFBR1 recruits and phosphorylates ILK, which then binds to the αE subunit of CD103, leading to AKT/PKB phosphorylation and thus initiating integrin inside-out signaling. It has been reported that ILK associates with TGFBR2 and that ILK gene inactivation results in a decrease in TGFBR2 signaling (38).
ILK is a serine–threonine kinase that interacts with intracytoplasmic domains of β1 and β3 integrins and cytoskeletal-associated proteins (39). This kinase is associated with integrin activation and intracellular signaling (40), and has been reported to mediate actin filament rearrangement and tumor cell migration and invasion through PI3K/AKT/Rac1 signaling (41). Indeed, ILK directly binds to AKT/PKB and contributes to its phosphorylation, thereby regulating several signaling pathways including spreading and migration (42). Consistently, our results indicated that ILK is directly involved in CD103 intracellular signaling by promoting phosphorylation of AKT/PKB and that the CD103–E-cadherin interaction induced phosphorylation of ILK and its subsequent polarization at the leading edge of migrating T cells. Implication of ILK in both inside-out and outside-in integrin signaling pathways (43, 44) and in β2 integrin-mediated polarization of lytic granules toward the target have also been reported (45). Our data suggest that ILK functions as a key component of a protein complex formed upon CD103 clustering that may connect extracellular ligands with the T-cell intracellular signaling complex. We also show here that TGFβ corresponds to an important factor that not only activates αE (CD103) subunit expression, but also locally regulates integrin activation and signaling via the ILK and AKT/PKB pathway. This leads to strengthening of CD103–E-cadherin interaction, thereby initiating outside-in signaling that promotes CD103-mediated T-cell functions such as T-cell spreading, migration, and cytotoxicity.
Overall, our data expand our knowledge of CD8+ T-cell locomotion within epithelial tissues and provide more insight into mechanisms that promote CD103 signaling within a TGFβ–rich tumor microenvironment. Thus, CD103–E-cadherin adhesive interactions regulate multiple local T-cell activities, including adhesion, Ca2+ response, and recruitment within epithelial tumor regions. This integrin is also involved in T-lymphocyte proliferation, cytokine secretion, cytotoxic IS maturation, and killing of epithelial target cells (8, 9, 14, 46). These findings are consistent with a general role of CD103 in promoting CD8 T-cell immune responses within epithelial tumors and may contribute to optimizing current T-cell–based cancer immunotherapy strategies.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: M. Boutet, F. Mami-Chouaib
Development of methodology: N. Théret, E. Donnadieu, F. Mami-Chouaib
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M. Boutet, L. Gauthier, M. Leclerc, G. Gros, V. de Montpreville, E. Donnadieu, F. Mami-Chouaib
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Boutet, L. Gauthier, E. Donnadieu, F. Mami-Chouaib
Writing, review, and/or revision of the manuscript: M. Boutet, L. Gauthier, N. Théret, F. Mami-Chouaib
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): F. Mami-Chouaib
Study supervision: F. Mami-Chouaib
The authors thank S. Salomé-Desmoulez for her help with confocal microscopy.
This work was supported by grants from the INSERM, Association pour la Recherche sur le Cancer (ARC), Institut National du Cancer (INCa), and Ligue contre le Cancer. M. Boutet was a recipient of a fellowship from Cancéropôle IDF and Gustave Roussy (SIRIC-SOCRATE).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.