Implication of combined PD-L1/PD-1 blockade with cytokine-induced killer cells as a synergistic immunotherapy for gastrointestinal cancer

Abstract

Cytokine-induced killer (CIK) cells represent a realistic approach in cancer immunotherapy with confirmed survival benefits in the context of metastatic solid tumors. However, therapeutic effects are limited to a fraction of patients. In this study, immune-resistance elements and ideal combination therapies were explored. Initially, phenotypic analysis was performed to document CD3, CD56, NKG2D, DNAM-1, PD-L1, PD-1, CTLA-4, TIM-3, 2B4, and LAG-3 on CIK cells. Upon engagement of CIK cells with the tumor cells, expression of PD-1 on CIK cells and PD-L1 on both cells were up-regulated. Over-expression of PD-L1 levels on tumor cells via lentiviral transduction inhibited tumoricidal activity of CIK cells, and neutralizing of PD-L1/PD-1 signaling axis could enhance their tumor-killing effect. Conversely, blockade of NKG2D, a major activating receptor of CIK cells, largely caused dysfunction of CIK cells. Functional study showed an increase of NKG2D levels along with PD-L1/PD-1 blockade in the presence of other immune effector molecule secretion. Additionally, combined therapy of CIK infusion and PD-L1/PD-1 blockade caused a delay of in vivo tumor growth and exhibited a survival advantage over untreated mice. These results provide a preclinical proof-of-concept for simultaneous PD-L1/PD-1 pathways blockade along with CIK infusion as a novel immunotherapy for unresectable cancers.

Keywords: CIK; NKG2D; PD-L1/PD-1; gastrointestinal tumor; immunotherapy.

Conflict of interest statement

CONFLICTS OF INTEREST

The authors disclose no potential conflicts of interest

Figures

Figure 1. In vitro expansion and main phenotypes of CIK cells derived from donors (n = 10)

Differential expression of three main phenotypic subsets of CIK cells over in vitro culture are shown for identification of CIK cells, CD3/CD56 (A), T cell associated immune receptors, CD3/CD4/CD8 (B), and major activating receptors of CIK cells, CD3/NKG2D (C). Flow cytometric analysis was performed over expansion periods on day0, day7, day14, and day21. Each representative picture was accordingly shown on the right panel.

Figure 2. Stable variations in PD-L1 expression on the tumor cells via lentiviral transduction and corresponding influences upon tumoricidal activity of CIK cells

A. Gastric and colorectal cancer cell lines (MGC803 and RKO) were transduced with lentiviral vectors containing siRNA directed against PD-L1, whereas HGC27 and SW480 were transfected with PD-L1 cDNA. After 72 hours, transfected tumor cells were confirmed for PD-L1 expression by flow cytometry. filled isotype staining; bold line PD-L1-stained untransfected cell lines; dashed line PD-L1-stained transfected cell lines with siRNA (MGC803 and RKO) or PD-L1 cDNA (HGC27 and SW480). B. CIK cells exerting cytotoxicity against the tumor cells were assessed by a non-radioactive cytotoxicity assay. Comparative curves were drawn between the tumor cells transfected with siRNA or PD-L1 cDNA and with scrambled shRNA as negative control group. Results represent at least three independent experiments using CIK cells from the same donor and are shown as Mean±SEM. C. CIK cells was pre-incubated with above tumor cells in 2ml culture medium per well in a 6-well-plate at a E:T ratio of 10:1 overnight, and both suspending and adherent cells were harvested and analyzed for cell apoptosis assay. D. Concordantly, obvious lysis of representative tumor cells (MGC803 and RKO) was observed over time by microscopy.

Figure 3. PD-L1/PD-1 pathway blockade efficiently increases tumor-killing activity of CIK cells

A. The levels of PD-1 and PD-L1 on the CIK cells were observed after 12 hours co-culture with MGC803 or RKO. Data from three independent experiments were shown as Mean±SEM, and statistical significance was determined by Student t test (* means P<0.05, ** means P<0.01, and *** means P<0.001). Representative pictures are indicated on the right panel. B. PD-L1 levels on the tumor cells (MGC803) were detected by the co-culture, as well as those with the adding of IFN-γ (200U/ml). filled isotype staining; bold line PD-L1-stained MGC803. C. Correlation was performed between IFN-γ and PD-L1 mRNA levels in the gastric cancer tissues. Pearson's r and P value are displayed. -ΔCt indicates the difference in the threshold cycle between the target genes and β-actin. D. CIK cells exerting cytotoxicity against MGC803 or RKO, respectively, were assessed by a non-radioactive cytotoxicity assay, with or without PD-L1/PD-1 blockade using indicated antibodies (20μg/ml). Corresponding isotype antibody (20μg/ml) for PD-L1/PD-1 was used as a control group. Results represent at least three independent experiments using the CIK cells from the same donor and are shown as Mean±SEM.

Figure 4. In vivo cytotoxicity of CIK cells against murine gastric cancer model

Nude mice were subcutaneously implanted with an 8mm3 tumor fragment prepared beforehand with MGC803 cells injected into 5 mice. Three days after tumor implantation, mice received weekly intravenous infusions of 107 CIK cells. Anti-PD-1 antibody was pre-added to block PD-1 on CIK cells for 1 hour. A. An obvious tumor growth delay was observed in combined CIK with anti-PD-1-treated mice compared with the control group (P<0.001), whereas no significant difference was found in CIK alone-treated mice (P>0.05). B. Percentage survival of MGC803 tumor-bearing mice treated with CIK cells with or without anti-PD-1 antibody was significantly increased compared with untreated mice using Kaplan-Meier analysis (P<0.05), whereas no significant difference was found between CIK cells-treated groups with or without anti-PD-1 antibody (P>0.05). C. At the end of infusion, representative pictures of CIK cells infiltration at tumor sites were shown by IHC using antibodies against CD5, and expression of ki-67 was shown in tumors from CIK-treated mice and control, scale bar, 50μm. Tumor growth and survival data were shown independently with 6 to 10 mice per group.

Figure 5. Blocking of NKG2D binding with its ligands results in impaired cytotoxic activity of CIK cells against tumor cell lines (MGC803, HGC27, RKO, and SW480)

A. All the tumor cell lines were observed for MIC A/B expression. Besides, the lytic activity of CIK cells, with or without blocking its recognition receptor, NKG2D, was assessed by cytotoxicity assay. B. Meanwhile, all the above cell lines were detected for the levels of ULBP-2, another crucial ligand for NKG2D. Antibody-blockade of ULBP-2 was conducted to observe cytotoxicity of CIK cells against all the above cells that differentially express ULBP-2. filled isotype staining; bold line MIC A/B− or ULBP-2-stained tumor cell. All the data on the curves represent three independent experiments using CIK cells from the same donor and are shown as Mean±SEM.

Figure 6. Disrupting PD-L1/PD-1 binding using blocking antibodies markedly induced an increase in multiple immune effector molecules

A. IFN-γ and CD107a, as direct measurements of cytotoxicity of CIK cells against tumor cells, as well as granzyme B, were observed by the co-incubation at E:T ratio of 5:1, along with the detection for a potentially responsive phenotype in terms of the co-inhibitory molecules, CTLA-4 and LAG-3 (B), and the activating receptors, NKG2D and DNAM-1 (C). D. Gastric clinical specimens were examined for analyzing the difference in the mRNA levels of these functional receptors between the gastric cancer samples and matched adjacent normal samples, and statistical significance was determined by paired t test. E. Associated immune-promoting molecules, IFN-γ and CD107a, accompanied with NKG2D levels, were examined after PD-L1/PD-1 pathway blockade using indicated antibodies (20μg/ml), among the test groups and the control groups (indicated as untreated group and isotype antibody group). F. 5×105 CIK cells were co-cultured in triplicate with 1×105 tumor cells in 500 μl culture medium per well in a 48-well-plate. Anti-PD-L1 and anti-PD-1 together with isotype antibody (20μg/ml) were separately added into each well. After 48 h, the levels of IFN-γ in the supernatants were evaluated by ELISA. G. Using the same co-culture condition described above, changes of NKG2D levels were observed with the direct adding of IFN-γ cytokine. Results represent three independent experiments and are shown as Mean±SEM (* means P<0.05, ** means P<0.01, and *** means P<0.001).