PI3Kδ is essential for tumor clearance mediated by cytotoxic T lymphocytes

Abstract

Background: PI3Kδ is a lipid kinase of the phosphoinositide 3-kinase class 1A family and involved in early signaling events of leukocytes regulating proliferation, differentiation and survival. Currently, several inhibitors of PI3Kδ are under investigation for the treatment of hematopoietic malignancies. In contrast to the beneficial effect of inhibiting PI3Kδ in tumor cells, several studies reported the requirement of PI3Kδ for the function of immune cells, such as natural killer and T helper cells. Cytotoxic T lymphocytes (CTLs) are essential for tumor surveillance. The scope of this study is to clarify the potential impact of PI3Kδ inhibition on the function of CTLs with emphasis on tumor surveillance.

Principal findings: PI3Kδ-deficient mice develop significantly bigger tumors when challenged with MC38 colon adenocarcinoma cells. This defect is accounted for by the fact that PI3Kδ controls the secretory perforin-granzyme pathway as well as the death-receptor pathway of CTL-mediated cytotoxicity, leading to severely diminished cytotoxicity against target cells in vitro and in vivo in the absence of PI3Kδ expression. PI3Kδ-deficient CTLs express low mRNA levels of important components of the cytotoxic machinery, e.g. prf1, grzmA, grzmB, fasl and trail. Accordingly, PI3Kδ-deficient tumor-infiltrating CTLs display a phenotype reminiscent of naïve T cells (CD69(low)CD62L(high)). In addition, electrophysiological capacitance measurements confirmed a fundamental degranulation defect of PI3Kδ-/- CTLs.

Conclusion: Our results demonstrate that CTL-mediated tumor surveillance is severely impaired in the absence of PI3Kδ and predict that impaired immunosurveillance may limit the effectiveness of PI3Kδ inhibitors in long-term treatment.

Figure 1. Diminished reaction of PI3Kδ−/− CTLs to allogeneic mixed lymphocytes, but unaltered proliferation and Ca²⁺-response.

A. WT and PI3Kδ−/− splenocytes were CFSE-labeled and cultured in the absence and presence of allogeneic (BALB/c), mitomycin C-treated splenocytes. At the indicated time points, cells were harvested and proliferation of responding CTLs was assessed by flow cytometry. Percentages of proliferating CFSE+CD8+ T cells with and without the stimulus of mixed lymphocytes are illustrated. Proliferating CD8+ T cells were discriminated from undivided T cells by the reduced levels of CFSE in daughter cells. B. WT splenocytes were CFSE-labeled and cultivated in analogy to A. Pharmacological inhibition of PI3Kδ was achieved by treatment with indicated concentrations of CAL-101 during the experimental procedure. DMSO-treatment served as negative control. C. Proliferation of WT and PI3Kδ−/− CTLs in response to aCD3ε treatment was assessed in a CFSE proliferation assay. D. Proliferation of WT and PI3Kδ−/− aCD3-activated T cells was assessed under standard T cell medium conditions (in the presence of IL-2) and after deprivation from IL-2 by performing an [³H]-thymidine incorporation assay over 48 hours (with IL-2: WT: 12097±491cpm; versus PI3Kδ−/−: 12413±501cpm; without IL-2: WT: 1392±381cpm; versus PI3Kδ−/−: 1140±160cpm, n = 6, values represent mean±SD). E., F. WT and PI3Kδ−/− splenocytes were stained with 1 µM Indo-1 AM. Ca²⁺ flux in response to aCD3ε followed by crosslinking with streptavidin (E) or thapsigargin (F) was measured in CD8+ T cells using flow cytometry. Treatment with ionomycin served as positive control. Three independent experiments were carried out and one representative experiment is shown, respectively.

Figure 2. Reduced expression of cytotoxic components in PI3Kδ−/− CTLs.

WT and PI3Kδ−/− splenocytes were activated for 3 days with aCD3ε and cultured in T cell medium. A. mRNA expression of grzmA (WT: 0.097±0.036; versus PI3Kδ−/−: 0.014±0.01, n = 6, p = 0.0003), grzmB (WT: 1±0.076; versus PI3Kδ−/−: 0.419±0.075, n = 6, p<0.0001) and prf1 (WT: 0.0433±0.004; versus PI3Kδ−/−: 0.013±0.003, n = 6, p<0.0001) was quantified by qRT-PCR and normalized to the house-keeping gene gapdh. B. Similarly, under standard culturing conditions mRNA levels of trail (WT: 0.0049±0.0012; versus PI3Kδ−/−: 0.0026±0.0007, n = 4, p = 0.0178) and fasl (WT: 0.0138±0.0015; versus PI3Kδ−/−: 0.0039±0.0001, n = 4, p<0.0001) were measured. C. Ifng mRNA was quantified by qRT-PCR under standard culturing conditions (WT: 0.019±0.0005; versus PI3Kδ−/−: 0.004±0.001, n = 6, p<0.0001) and after stimulation with 5 ng/ml IL-12 for 4 h (WT: 0.241±0.06; versus PI3Kδ−/−: 0.087±0.016, n = 6, p = 0.0002). D. To quantify IFN-γ protein levels, WT and PI3Kδ−/− splenocytes were stimulated with ConA. After 48 h supernatants were harvested and IFN-γ release was measured by ELISA (WT: 6515±2061 pg/ml; versus PI3Kδ−/−: 1359±1147 pg/ml, n = 3, p = 0.0193). IFN-γ release of unstimulated controls of WT and PI3Kδ−/− splenocytes was below detection limit of the assay (<10 pg/ml). Statistics were calculated with an unpaired Student’s t-test, and values represent mean±SD. One out of two independently performed experiments with comparable results is shown.

Figure 3. PI3Kδ is indispensable for CTL degranulation.

A. Using the whole cell patch clamp technique, the cellular capacitance of aCD3-activated and in T cell medium cultivated WT and PI3Kδ−/− CTLs was determined. Membrane capacitances before (ctr) and after stimulation with PMA and ionomycin (PMA/iono) at the single cell level are summarized from two representative experiments (WT: ctr: 6.3±1.2pF, PMA/iono: 9±1.8pF, n = 10, p = 0.01; versus PI3Kδ−/−: ctr: 6.7±1.7pF, PMA/iono: 6.6±1.8, n = 9, p = 0.4; paired t-test; values represent mean±SEM). B. Fold increase in membrane capacitance due to stimulation with PMA/iono is calculated (WT: 1.46±0.13, n = 10, p = 0.0064; versus PI3Kδ−/−: 0.94±0.06%, n = 9, p = 0.3). Additionally, WT CTLs were pre-incubated with the PI3Kδ isoform-specific inhibitor IC-87114 (1 µM, 4 µM, 1 hour), DMSO-treatment served as control (DMSO: 2.14±0.18, n = 11, p<0.0001; 1 µM IC-87114: 0.98±0.03, n = 11, p = 0.49; 4 µM IC-87114: 0.98±0.01, n = 9, p = 0.18; values representing mean±SEM, One sample t-test). C. In vitro cultivated WT and PI3Kδ−/− CTLs were challenged with PMA and ionomycin. The percentage of CD107a+ T cells was measured via flow cytometry before and after the stimulus (WT: ctr: 15.5±2.4%, PMA/iono: 41.2±1.3%, p<0.0001; versus PI3Kδ−/−: ctr: 24.6±2.7%, PMA/iono: 29.3±1.7%, p = 0.085; n = 8, paired t-test, values represent mean±SEM). D. Accordingly, the degranulation of WT CTLs upon pharmacological inhibition of PI3Kδ using CAL-101 was tested (ctr: 6±0.7%; +PMA/iono: DMSO: 23.3±2.4%, 0.1 µM: 18.1±2%, 0.5 µM: 16.6±2.1%, 1 µM: 13.6±1.3%, 5 µM: 11.7±1%, n≥6, values represent mean±SEM. One-Way ANOVA revealed p<0.0001, Tukey’s Post-Hoc test was significant p<0.01 for 1 µM and highly significant p<0.001 for 5 µM compared to DMSO control). E. WT and PI3Kδ−/− CTLs were cultivated in T cell medium and their expression of CD107a was measured via flow cytometry under basal conditions and after electrostimulation or electrostimulation plus the degranulation inhibitor Concanamycin A, respectively (WT: ctr: 11.5±1.2% CD107a+ cells, electrostimulation: 41±3.3% CD107a+ cells, p = 0.0001; versus PI3Kδ−/−: ctr: 11±1.3% CD107a+ cells, electrostimulation: 12±0.8% CD107a+ cells, p = 0.54, paired t-test, values represent mean±SEM). All experiments were performed at least two times independently and summarized in the depicted graphs.

Figure 4. PI3Kδ-deficiency leads to severely impaired CTL cytotoxicity in vivo and in vitro.

A., B. WT and PI3Kδ−/− animals were immunized with SIINFEKL and CpG. Seven days later mice received CFSE-labeled targets and peptide-specific CTL activity in draining inguinal lymph nodes was analyzed by flow cytometry. A. A representative FACS histogram for each genotype is depicted. B. SIINFEKL-specific target cell killing was calculated from three independent experiments as described in Materials and Methods (WT: 80±6% antigen-specific lysis; versus PI3Kδ−/−: 52±10% antigen-specific lysis, n = 12; values represent mean±SD, unpaired t-test, p<0.0001). No specific killing was observed in control mice (data not shown). C., D. WT and PI3Kδ−/− animals were immunized with SIINFEKL peptide and CpG and boosted 7 days later. To generate effector cells, splenocytes were isolated at day 14 and co-cultured for 5 days with irradiated SIINFEKL-pulsed splenocytes. To determine peptide-reactive CTL cytotoxicity in vitro, CFSE-labeled EL4 (C) and OVA-expressing EG7 (D) target cells were co-cultured with effectors in ratios of 30∶1, 15∶1, 5∶1 and 1∶1. Specific in vitro target cell killing was quantified by flow cytometry (EG7, E:T = 30∶1: WT: 63% specific lysis; versus PI3Kδ−/−: 15% specific lysis). One representative experiment out of three is shown.

Figure 5. Impaired activation of PI3Kδ−/− CTLs by tumor cells.

A. 1×10⁶ MC38 colon adenocarcinoma cells were injected subcutaneously into the flanks of WT and PI3Kδ−/− animals. Fifteen days later tumor weights were analyzed. Depicted is the summary of 3 independent experiments (WT: 0.19±0.19 g; versus PI3Kδ−/−: 1.16±0.33 g, n = 12, values represent mean±SD). B. Representative pictures of MC38 solid tumors grown in WT and PI3Kδ−/− animals. C. After removal, tumors were minced and digested with collagenase D and DNase I. Tumor-infiltrating CD3+CD8+ CTLs were quantified using flow cytometry (WT: 1.9±0.5%; versus PI3Kδ−/−: 1.8±0.7%, n = 8, values represent mean±SD). D., E. Representative histograms showing expression levels of CD44, CD45RB, CD62L and CD69 on splenic CD3+CD8+ T cells from untreated WT and PI3Kδ−/− mice (D) and on MC38 tumor-infiltrating CD3+CD8+ CTLs of WT and PI3Kδ−/− mice (E). F. Summary of the expression levels of CD44 (WT: median = 19040, IQR = 16670–24220; versus PI3Kδ−/−: median = 15053, IQR = 12750–26440, n = 8), CD45RB (WT: median = 3222, IQR = 2907–4142; versus PI3Kδ−/−: median = 3945, IQR = 2579–5833, n = 8), CD62L (WT: median = 194, IQR = 146–219; versus PI3Kδ−/−: median = 228, IQR = 187–333, n = 8, Mann-Whitney test: p = 0.0572) and CD69 (WT: median = 2673, IQR = 1825–7606; versus PI3Kδ−/−: median = 1428, IQR = 1334–1698, n = 8, Mann-Whitney test: p = 0.0037) on tumor-infiltrating WT and PI3Kδ−/− CD3+CD8+ CTLs.