Abstract
CD8 T lymphocytes are able to eliminate nascent tumor cells through a process referred to as immunosurveillance. However, multiple inhibitory mechanisms within the tumor microenvironment have been described that impede tumor rejection by CD8 T cells, including increased signaling by inhibitory receptors. Lysophosphatidic acid (LPA) is a bioactive lysophospholipid that has been shown repeatedly to promote diverse cellular processes benefiting tumorigenesis. Accordingly, the exaggerated expression of LPA and LPA receptors is a common feature of diverse tumor cell lineages and can result in elevated systemic LPA levels. LPA is recognized by at least six distinct G protein–coupled receptors, several of which are expressed by T cells, although the precise function of LPA signaling in CD8 T-cell activation and function has not been defined. Here, we show that LPA signaling via the LPA5 receptor expressed by CD8 T cells suppresses antigen receptor signaling, cell activation, and proliferation in vitro and in vivo. Importantly, in a mouse melanoma model tumor-specific CD8 T cells that are LPA5-deficient are able to control tumor growth significantly better than wild-type tumor-specific CD8 T cells. Together, these data suggest that the production of LPA by tumors serves not only in an autocrine manner to promote tumorigenesis, but also as a mechanism to suppress adaptive immunity and highlights a potential novel target for cancer treatment. Cancer Immunol Res; 1(4); 245–55. ©2013 AACR.
Introduction
The adaptive immune system is able to detect and eliminate nascent tumors through a process referred to as immune surveillance and mediated in large part by cytotoxic CD8 T cells. However, this immune response to tumor may also contribute to tumorigenesis by providing selective pressure to which tumors adapt and eventually evade eradication, a process coined cancer immunoediting (1, 2). Tumors that evade the initial T-cell response create an immunosuppressive microenvironment from which variants arise, escape immune control, and grow without restraint. The mechanisms used by tumors to either evade the initial CD8 T-cell response or promote the tumor immunosuppressive environment are not fully defined. However, the recent success of immunotherapies that interfere with tumor-derived immune suppression (3–5) has underscored the importance of identifying the mechanisms by which tumors suppress CD8 T cells to escape immune control (6, 7).
The activation of cytotoxic CD8 T cells by either foreign or tumor antigen is initiated via signals transmitted by the T-cell antigen receptor (TCR; ref. 8). TCR signaling and the subsequent function of mature T cells can be regulated in a positive or negative manner by different surface coreceptors (9). Multiple inhibitory mechanisms within the tumor microenvironment that impede tumor rejection by tumor-infiltrating lymphocytes (TIL) have been described (6, 10), including the increased signaling by CD8 T-cell inhibitory receptors, such as the well-characterized CTLA-4 molecule (11).
Lysophospholipids comprise a small family of structurally simple lipids that include sphingosine-1-phosphate (S1P) and lysophosphatidic acid (LPA), and they induce diverse biologic and pathophysiologic effects by signaling through specific G protein–coupled receptors (GPCR; refs. 12, 13). The LPA lysophospholipid is recognized by six specific GPCR, LPA1–6 (14), and although T lymphocytes are known to express several LPA receptors (13, 15–17), the immune regulatory activities of LPA are not well understood. LPA concentration in the blood of healthy individuals has been reported to vary from high nanomolar to low micromolar levels (12, 18). In vivo, production of LPA results predominantly from the activity of autotaxin (ATX; ref. 19), an extracellular lysophospholipase D originally isolated and identified from a human melanoma as an autocrine motility factor (20). Since then LPA has been found aberrantly produced in a number of different malignant cell types (21–23), resulting in significantly increased systemic levels that can reach 60 μmol/L in malignant effusions (24–26). At these elevated levels, LPA has been shown to promote tumor progression by enhancing tumor migration, survival, metastasis, angiogenesis, and therapeutic resistance (27–31).
Previously LPA has been shown to modulate the activation of different cell types (17), and in this study, we investigated if LPA could influence CD8 T-cell activation. Here, we report that CD8 T cells express the LPA5 receptor and signaling by this GPCR inhibits CD8 T-cell receptor signaling, activation, and proliferation. Furthermore, we show that tumor-specific CD8 T cells lacking LPA5 can control the progression of established tumor progression more efficiently than the LPA5-sufficient tumor-specific CD8 T cells. Thus, our findings reveal a novel role for lysophospholipid-mediated protection of tumor from adaptive immunity.
Materials and Methods
Mice
C57BL/6 (CD45.2) and CD45.1 (B6.SJL-Ptprca Pepcb/BoyJ) were obtained from The Jackson Laboratory or bred in the Biological Resource Center (BRC) at National Jewish Health (NJH; Denver, CO). CD45.1 OT-I mice (a gift from Dr. Ross Kedl, University of Colorado, Denver, CO), Lpar2−/− mice (a gift from Dr. Jerold Chun, Scripps Research Institute, La Jolla, CA), Lpar5−/− mice, TCRα−/− mice (a gift from Dr. Philippa Marrack, NJH) were bred in the BRC at NJH. LPA5−/− mice were generated as described in the Supplementary Data. All mice used were 8 to 12 weeks of age, housed under specific pathogen-free conditions, and were maintained in accordance with the regulations of the Institutional Animal Care and Use Committee.
Calcium mobilization
Erythrocyte-lysed cells isolated from spleen, pooled lymph nodes, or purified CD8+ T cells were suspended at 20 × 106 cells/mL in RPMI-1640 medium supplemented with 2.5% fatty acid–free (faf) bovine serum albumin (BSA; Calbiochem) and loaded with Indo1-AM (Molecular Probes) as described in the Supplementary Data.
qPCR
CD8+ T cells were isolated from the spleens and lymph nodes of wild-type mice and LPA receptor expression measured by real-time quantitative reverse transcriptase (RT-PCR) using Platinum SYBR Green quantitative PCR (qPCR) SuperMix-UDG (Invitrogen Life Technologies). The details of qPCR analysis including specific forward and reverse LPA receptor primers are provided in the Supplementary Data.
Cytometry
All antibodies were purchased from eBiosciences, BD Pharmingen, Biolegend, or were produced in our laboratories. Cells were stained in 2% BSA–PBS+0.1% sodium azide with blocking Fc receptor antibody (2.4G2) on ice for 20 to 30 minutes. For viability assessment, 7-amino actinomycin D (7-AAD) was added 10 minutes before data acquisition. All flow-cytometric analysis was conducted on the LSRII flow cytometer (BD) and analyzed with FlowJo v8 (TreeStar) and GraphPad Prism software (v 5.0).
Lipid preparation
LPA (C16 LPA; Avanti Polar Lipids) was solubilized to 5 mmol/L concentration in 0.1% BSA–PBS, aliquoted, and stored at −20°C. Aliquots were diluted to 1 mmol/L in RPMI-1640 medium supplemented with 2.5% faf-BSA (Calbiochem) before use. Octodecenyl thiophosphate (OTP) was generated as previously described (32), stored as a powder, and solubilized in 95% methanol to create aliquots. Virgin glass tubes and caps were sterilized by autoclave for aliquots that were stored at −20°C. The concentration was confirmed by phosphorus assay. OTP was solubilized for experimental use by sonication with FBS- or BSA-containing culture media or vehicle (2% propanediol and 1% ethanol in PBS) for in vitro or in vivo usage, respectively. For in vitro experiments, OTP was solubilized to 50 μmol/L and passed through a 0.2-μm filter for further sterilization. For in vivo experimentation, solubilized OTP was transferred to siliconized Eppendorf tubes and animals were dosed at 5 mg/kg every 8 hours.
Generation of bone marrow–derived dendritic cells
Congenic gender-matched bone marrow–derived dendritic cells (BMDC) were generated by flushing of femur and tibia and culture at 106 cells/mL in RPMI-1640 with 20 ng/mL granulocyte macrophage colony-stimulating factor (GM-CSF), 10% FBS (Omega Scientific), penicillin–streptomycin, and GlutaMAX (Invitrogen). Media was refreshed on days 3 and 5. On day 7, BMDCs were harvested from culture and stimulated with 1 ng/mL LPS for 90 minutes and pulsed with peptide for the last hour of LPS treatment. BMDCs were washed five times to remove LPS and unbound peptide before transfer.
In vitro T-cell activation and proliferation
To determine how LPA affected antigen-specific activation of CD8 T cells, OT-I splenocytes were isolated, erythrocyte lysed, and labeled with carboxyfluorescein succinimidyl ester (CFSE; Invitrogen). For all CFSE labeling, cells were suspended at 15 × 106 cells/mL in PBS and CFSE was added to a final concentration of 2 μmol/L for 10 minutes and then washed in media. Splenocytes were pulsed with 1 μmol/L of the SIIGFEKL (G4; AnaSpec, Inc.) or SIINFEKL (a gift from Philippa Marrack) peptides for 4 hours or 90 minutes, respectively, in 5% faf-BSA RPMI, then washed. Cells were cultured in 96-well plates at 2.5 × 106 cells/mL in the presence or absence of 50 μmol/L OTP that was sterile-filtered before addition to culture. Cells were enumerated by flow cytometry, and the proportion of cells proliferated was calculated by FlowJo analysis. The mean fluorescence intensity (MFI) values of activation marker expression were normalized.
To assess in vitro cytokine production, OT-I effector T cells were generated by pulsing erythrocyte-lysed OT-I splenocytes with 1 μmol/L SIINFEKL and culture with interleukin-2 (IL-2) for 5 days. On day 5 of culture, target cells (EL4 cells) were pulsed with 1 μmol/L SIINFEKL and cultured at an effector to target ratio of 0.625:1 with OT-I effector T cells for 4 hours in the presence of Brefeldin A, in the presence or absence of sterile-filtered 50 μmol/L OTP.
In vivo T-cell transfer and antigen-specific stimulation
BMDCs were generated as described earlier. One day before BMDC transfer, CD8+ T cells were purified from OT-I spleen and lymph node cells with a CD8+ T-cell enrichment kit (Miltenyi) to a purity of 95% or more, and 106 CFSE-labeled CD8+ T cells were transferred to CD45 allotype-mismatched recipient C57BL/6 mice. SIINFEKL-BMDC (106) were suspended in PBS and transferred subcutaneously in the scruff of individual recipients. On day 3 postimmunization, animals were sacrificed and draining lymph nodes (dLN; axilary, brachial, and cervical), non-draining lymph nodes (ndLN; inguinal and mesenteric), and spleen were harvested. After erythrocyte lysis, cells were counted by Z2 Coulter Particle Count and Size Analyzer (Beckman-Coulter) and 10 × 106 cells were stained for flow cytometry. Cells were suspended in fluorescence-activated cell sorting (FACS) buffer and stained with 7-AAD for viability before analysis on the BD LSRII.
B16.cOVA tumor experiments
The OVA-transfected B16 tumor cell line (B16.cOVA) was kindly provided by Dr. Ross Kedl. Cells were maintained in RPMI-1640 (Cellgro), supplemented with 10% heat-inactivated FBS (Omega Scientific), GlutaMAX, penicillin–streptomycin, minimum essential medium (MEM) nonessential amino acids, sodium pyruvate (Invitrogen), and 0.75 mg/mL G418 sulfate selection (Cellgro).
To determine how LPA5−/− OT-I T cells responded to tumor, 105 B16.cOVA cells were transferred subcutaneously into the hind leg of recipients, and 5 days later, 0.5 × 106 CFSE-labeled OT-I T cells were transferred via the tail vein. Recipients were sacrificed 5 days later and tumor and dLN were harvested. Lymphoid organs were mashed through a cell strainer (100 μm; BD), erythrocyte lysed, and counted. Tumors were digested in 0.5 mg/mL collagenase D (Fisher Scientific) and 60 U/mL DNase (Sigma-Aldrich) for 30 minutes with perturbation every 10 minutes, and then neutralized with 5 mmol/L EDTA for 5 minutes. Tumor cells were homogenized by pipette tituration and passed through a 100 μm filter, erythrocyte lysed, and counted. Tumor diameter was measured with calipers and body weight and overall appearance were assessed every 2 days. Mice were sacrificed when tumor diameter exceeded 10 mm or at 8 days after T-cell transfer, whichever came first.
Statistical analysis
All statistical analysis was conducted with GraphPad Prism software (version 5.0) using a two-tailed unpaired Student t test.
Results
LPA inhibits T-cell antigen receptor-mediated calcium mobilization via LPA5
Both human and murine T cells express LPA receptors (15, 16), so we initially questioned whether LPA could influence TCR signaling. Specifically, we tested whether LPA could modulate TCR-induced intracellular calcium mobilization. These results revealed that 20 μmol/L of LPA potently inhibited intracellular calcium mobilization by CD8+ T cells in response to TCR stimulation achieved by antibody-mediated cross-linking of the TCR (Supplementary Fig. S1). In contrast, LPA treatment in the absence of antigen receptor stimulation did not affect intracellular calcium levels. LPA inhibition of CD8 TCR signaling was further shown to be dose dependent with a reduction of intracellular calcium levels achieved with LPA concentrations as low as 1 μmol/L (Fig. 1A), approximating the levels reported in the blood of healthy individuals (12, 18).
LPA signals via LPA5 to inhibit CD8+ T-cell TCR–mediated intracellular calcium release. A, intracellular calcium concentration was assessed in purified CD8+ T cells stimulated with 10 μg/mL of anti-CD3 in the absence (black line) or presence of the indicated titrated concentrations of LPA. B, expression of Lpar1-6 by isolated CD8+ T cells as measured by qPCR. C, LPA2−/− or LPA5−/− CD8+ T cells were isolated, and intracellular calcium mobilization was assessed in response to anti-CD3 stimulation in the presence (bold line) or absence (thin line) of 20 μmol/L LPA.
There are currently six validated LPA receptors: LPA1–6 (14) and qPCR analyses revealed that purified CD8 T cells expressed LPA2, LPA5, and LPA6 (Fig. 1B), consistent with previous findings (15, 16). To determine which LPA receptor(s) was responsible for LPA inhibition, we compared calcium mobilization in TCR-stimulated CD8+ T cells from LPA2−/− and LPA5−/− mice (LPA6−/− mice were not commercially available). These results revealed that although LPA inhibition was intact in the LPA2-deficient T cells, LPA inhibition was absent in LPA5-deficient T cells (Fig. 1C). These data show that LPA5 expression is required for the inhibition of TCR-mediated calcium mobilization.
LPA signaling inhibits proliferation and activation in vitro
Increased intracellular calcium concentration is an important early consequence of antigen receptor signaling that leads to the activation of distinct transcriptional programs important for T-cell activation and function (33). Thus, we next tested whether LPA suppression of TCR-induced calcium mobilization would lead to reduced CD8 T-cell activation in vitro. Initial experiments revealed that LPA was able to inhibit TCR-mediated activation of CD8+ T cells, as determined by reduced CD62L downregulation (34), however, only when T cells were stimulated with suboptimal concentration of anti-CD3 (Supplementary Fig. S2). To investigate whether LPA could inhibit antigen-specific stimulation of CD8 T cells, we used the well-characterized OT-I TCR transgenic mouse model that features CD8 T cells that express a TCR specific for the chicken ovalbumin peptide, SIINFEKL, in the context of the MHC molecule, H-2Kb (35). This system also allowed us to modulate T-cell activation with different stimulatory peptides (36). More specifically, the OT-I TCR displays relatively high affinity (KD ∼ 5 μmol/L) for the SIINFEKL ovalbumin peptide, but it also recognizes the altered peptide ligand, SIIGFEKL (G4), with relatively weaker affinity (KD ∼ 10 μmol/L; ref. 37). Thus with these peptides, we could compare the effects of LPA signaling after relatively high or low affinity TCR stimulation. Because LPA is degraded both in vitro and in vivo (38), we also used a metabolically stabilized LPA analog, OTP, to induce LPA signaling (39, 40). Of note, OTP is preferentially recognized by LPA5 relative to other LPA receptors and displays a much lower EC50 for LPA5 compared with other LPA receptors (41).
Using this system, we monitored CD8 T-cell proliferation and expression of activation markers after antigen-specific TCR stimulation in vitro in the absence or presence of LPA signaling achieved with OTP. Proliferation was monitored by CFSE dilution and showed that in the presence of OTP, CD8 T cells were acutely inhibited from proliferating in response to G4 peptide stimulation (Fig. 2A). Specifically, by day 3, control cells had typically started to divide with the majority having undergone at least two divisions by day 7, whereas the majority of OTP-treated cells remained undivided in the same period. OTP inhibition of CD8 T-cell activation was further evidenced by significantly reduced expression of both the CD25 and CD44 activation antigens after TCR stimulation (Fig. 2B). Likely as a consequence of inhibiting activation and proliferation, OTP treatment also resulted in a reduced accumulation of CD8+ T cells (Fig. 2C) without any significant effect on viability as compared with unstimulated controls (Fig. 2D). Importantly, OTP treatment did not affect the ability of antigen-presenting cells to present peptide to OT-I T cells as OTP-treated and -untreated antigen-presenting cells were equally able to induce CD25 upregulation (Supplementary Fig. S3). Similar to our initial experiment with anti-CD3 stimulation, we found that the ability of LPA signaling to dampen CD8 T-cell proliferation and activation was less effective after the TCR was stimulated with the higher affinity SIINFEKL peptide (Fig. 2E). In addition, OTP treatment did not seem to significantly influence SIINFEKL-stimulated TNF-α or IFN-γ cytokine production by in vitro–generated effector CD8+ T cells (Fig. 2F). Together, these findings show that LPA signaling potently suppresses TCR signaling, cell activation, and proliferation when T cells are stimulated by relatively weak-affinity antigens.