Tumor-treating fields (TTFields) induce immunogenic cell death resulting in enhanced antitumor efficacy when combined with anti-PD-1 therapy

The cell lines used for this study were Lewis lung carcinoma (LLC-1), CT-26 murine colon carcinoma, HEPG2 (HB-8065) human hepatocellular carcinoma, H520 (HTB-182) human lung squamous cell carcinoma, and spontaneously transformed murine ovarian surface epithelial (MOSE-L).

TTFields were applied for 24–72 h using the Inovitro™ system (Novocure, Israel) as previously described [5, 8]. Cells were treated with TTFields at intensity of 175 V/m (RMS) and frequencies of 150 kHz (LLC-1, H520, and HEPG2) and 200 kHz (CT-26, MOSE-L).

Inhibition of cell growth was analyzed by quantitatively determining cell count using iCyt EC800 flow cytometer (Sony Biotechnology). The relative number of cells at the end of treatment was expressed as a percentage of untreated control cells.

Cell death was assessed by double staining of cells with FITC-conjugated annexin V (MEBCYTO^© 4700 Apoptosis Kit; MBL^©) and 7-Aminoactinomycin D (7-AAD; BioLegend^©) as per manufacturer’s instructions.

Cells were incubated with rabbit anti-mouse calreticulin antibody (1:200; Abcam), followed by incubation with donkey anti-rabbit Alexa Fluor 488 conjugated antibody (1:250; Jackson ImmunoResearch).

Cells were loaded with 1 μM quinacrine (Sigma-Aldrich^©). Cellular ATP levels were measured by quantification of the percentage of viable cells (7AAD-) that exhibited strong reduction in quinacrine staining (Quinacrine-/7AAD-). Alternatively, supernatants were used for quantification of ATP release using the ENLITEN^® ATP Assay System (Promega).

Thin sections (70 nm) were coated with carbon and visualized using Zeiss Ultra-Plus FEG-SEM equipped with transmission electron detector, at acceleration voltage of 30 kV.

Supernatants were collected and were used for quantification of HMGB1 by ELISA assay (IBL International GmbH).

Microtubule-associated protein 1 light chain 3 (LC3) detecting antibody (rabbit polyclonal, Novus) and Alexa Fluor 488-conjugated secondary antibody (Jackson ImmunoResearch) were used. Images were collected using LSM 700 laser scanning confocal system, attached to an upright motorized microscope with × 63/1.40 oil objective (ZeissAxio Imager Z2). For detection of T-cells, frozen sections (7 µm thick) were stained with anti-CD8 primary antibody (YTS169.4, abcam), followed by secondary Cy3-conjugated donkey anti-mouse antibody (Jackson ImmunoResearch). The whole slide image was collected using automatic slide scanner 250 Flash (3DHISTECH). For immunohistochemistry staining, lungs were embedded in paraffin and staining was performed on 4-µm sections using the Leica Bond max system (Leica Biosystems Newcastle Ltd, UK). Sections were dewaxed and pretreated with epitope retrieval solution (Leica Biosystems Newcastle Ltd, UK) followed by anti-CD45 antibody (1:600, ab10558 by Abcam).

Anti-LC3 was purchased from Novus, phospho-eIF2α (Ser 51) and eIF2α detecting antibodies from Cell signaling, anti-vinculin from Sigma-Aldrich, anti-GAPDH from Santa-Cruz, followed by HRP-conjugated secondary antibody (Abcam) and a chemiluminescent substrate (Millipore). Densitometric analysis was performed with the Image Studio Lite 5.2 software (LI-COR).

For orthotopic lung cancer model, 10–12-week-old male C57Bl/6 mice were injected directly into the lungs with LLC-1 cells. For heterotopic subcutaneous colon cancer model, 0.5 × 10⁶ CT-26 cells in 200 µL PBS were subcutaneously injected to the upper right flank of 10-week-old female Balb/c mice. Mice received an I.P. injection of anti-PD-1 (RMP1-14; 250 μg) or Rat IgG2a (2A3; 250 μg) (Bioxcell). Tumor size was assessed with Vernier calipers using the formula width² × length × 0.5.

For more details, see Figs. 5a, 6a, and S5.

The tumors were prepared as single-cell suspensions using the gentleMACS™ dissociator and the tumor dissociation kit (Miltenyi Biotec GmbH, Gladbach, Germany).

TILs were isolated by dissociating tumor tissue as described above. Single-cell suspensions were filtered and separated by Mouse Pan T magnetic bead selection (Invitrogen). Isolated cells were then cultured in vitro in the presence of T-Activator CD3/CD28 magnetic beads (Invitrogen) for 24 h.

For the generation of bone marrow-derived DCs (BMDCs), bone marrow cells were flushed from the femurs and tibias of 5–7-week-old C57BL/6 mice. For DC maturation assay, TTFields-treated cells were added at a ratio of 1:1 for 24 h. For phagocytosis assay, CellTracker™ Deep Red Dye (25 μM; Invitrogen) pre-stained cells were added at a ratio of 1:1 for 2 h.

10–12-week-old male C57Bl/6 mice were injected to the peritoneal cavity with 200 µL PBS or 200 µL PBS containing 2 × 10⁶ control, TTFields-treated cells, and freeze/thawed (F/T)-treated cells. Following 48 h, mice were euthanized. Infiltration of immune cells (CD45+) was evaluated using flow cytometry.

Fluorochrome-conjugated anti-mouse CD45 (clone 30-F11), CD8a (clone53-6.7), F4/80 (clone BM8), PD-L1 (clone 10F.9G2), CD4 (clone GK1.5), CD3 (clone 17A2), I-A/I-E (clone M5/114.15.2), CD40 (clone 3/23), CD80 (clone 16-10A1), CD11c (N418), CD11b (clone M1/70), Foxp3 (clone MF-14 and REA788), and IFN-γ (clone XMG1.2) were used. For intracellular FOXP3 and cytokines, staining was done following the manufacturer protocol using Foxp3 Fix/Perm buffer kit (Biolegend) or FoxP3 Staining Buffer Set (miltenyi Biotec). Mouse Fc block (anti CD16/CD32 clone 93) was used prior to staining with markers antibodies. Antibodies were purchased from BioLegend and Miltenyi Biotec. Viobility 405/452 Fixable Dye (Miltenyi Biotec) was used for the discrimination of dead cells. Data were acquired on Fortessa (BD) or MACSQuant Analyzer 10 (Miltenyi Biotec) flow cytometers and analyzed using FlowJo software (Ashland).

Data were analyzed with Graphpad Prism software (Graphpad), and P values of < 0.05 were considered to be statistically significant and indicated as *, P < 0.05; **, P < 0.01; and ***, P < 0.001.

To determine whether TTFields application promote ICD, murine Lewis lung carcinoma (LLC-1), murine colon carcinoma (CT-26), murine ovarian surface epithelial (MOSE-L), human hepatocellular carcinoma (HEPG2), and human lung squamous cell carcinoma (H520) cells were treated with TTFields for 24–72 h, each at its optimal frequency [9]. The tested cell lines demonstrated reduction in cell counts (Fig. 1a, Supplementary Fig. 1a, b) as well as induction of apoptosis as indicated by staining with annexin V and the vital dye, 7-Aminoactinomycin D (7AAD; Fig. 1b, Supplementary Fig. 1c–e). Since apoptosis is considered to be immunologically quiescent [10, 11], we then examined whether TTFields-induced cell death is accompanied by the release of the alarmin high-mobility group box 1 (HMGB1) protein. Indeed, upon treatment with TTFields the tested cell lines exhibited significant secretion of HMGB1, which, in accordance with cell death, was also dependent on treatment duration (Fig. 1c, Supplementary Fig. 1f–g).

We next analyzed the effects of TTFields on another prominent prerequisite of ICD, i.e., translocation of the chaperone CRT to the cell surface of dying cells [12]. A quantitative analysis of CRT surface exposure revealed that TTFields treatment induced enhanced surface exposure of CRT in all tested cell lines. Specifically, we saw a significant increase in both percentages of viable CRT positive cells (Fig. 2a, b, Supplementary Fig. 2a, b) and cell surface distribution of CRT, as evaluated by the differences in median fluorescence intensity (MFI; Fig. 2c, Supplementary Fig. 2c, d). As seen with the induction of apoptosis and the release of HMGB1, CRT surface exposure was also dependent on treatment duration. To better understand why TTFields-induced cell death activates the exposure of CRT, and in accord with the mandatory role of ER stress for this process, we examined whether TTFields treatment leads to ER stress response [13, 14]. Our results demonstrate that under TTFields application, the translation initiation factor eIF2α becomes phosphorylated (Fig. 2d). These results demonstrate that TTFields induce ER stress response, which may be the trigger to the observed CRT translocation to the cell surface [15].

Along with cell surface exposure of CRT and the release of HMGB1, the secretion of adenosine triphosphate (ATP) by dying cells constitutes one of the hallmarks of ICD [16]. To further explore the observation that TTFields application potentially induces ICD in replicating cells, we examined ATP secretion by treated cells. We performed flow cytometry measurements of quinacrine staining in viable cells as an indicator of intracellular ATP levels in cells exposed to TTFields [17]. Cells treated with TTFields exhibited reduced quinacrine staining (Fig. 3a upper panel, Supplementary Fig. 3). This reduction was also accompanied by secretion of ATP (Fig. 3a lower panel). Since autophagy is required for the optimal release of ATP from dying cells, we investigated whether TTFields affect autophagy in the treated cells [18, 19]. Transmission electron microscopy analysis revealed the accumulation of vacuoles with the typical morphological appearance of autophagic structures containing cytosolic materials (Fig. 3b; red arrows) in treated cells. To further confirm these findings, lipidated LC3 (LC3-II) levels were used to monitor autophagy using immunoblotting and immunofluorescence microscopy. Increased punctate distribution of LC3-II was observed following TTFields treatment (Fig. 3c). The observed LC3-II accumulation suggests either an increase in the number of autophagic events or a block in autophagosome degradation. To assess this, we performed immunoblot analysis of LC3-II in the presence or absence of the lysomotropic agent chloroquine (CQ) [20]. The addition of CQ led to increased LC3-II levels in TTFields-treated cells, demonstrating enhanced autophagic flux, which can explain the induction of ATP release (Fig. 3d).

We next evaluated the actual potential of cells that die during TTFields application to promote phagocytosis by DCs by challenging bone marrow-derived DCs (BMDCs) with LLC-1 cells that were treated with TTFields for 48 h. TTFields-treated cells were effectively phagocytosed by BMDCs, while untreated control cells did not show similar outcome (Fig. 4a, b). To assess the functional status of these BMDCs, we also analyzed the surface expression of the activation markers MHC class II (MHC II), CD40, and CD80. Lipopolysaccharide (LPS)-treated cells served as a positive control, and cells subjected to freeze–thaw cycles (F/T cells) served as negative control. After overnight culture with control, TTFields, or F/T-treated LLC-1 cell suspension, these markers were found to be upregulated in BMDCs that were co-cultured with TTFields-treated cells. This upregulation was observed for all the costimulatory molecules analyzed (Fig. 4c–e, Supplementary Fig. 4). Finally, to test the ability of TTFields-treated cells to attract immune cells in vivo, we evaluated leukocytes recruitment in mice following intraperitoneal injection of: live, F/T-treated, or TTFields-treated cells. Intraperitoneal injection of TTFields-treated cells induced significant recruitment of leukocytes (CD45+), as compared to control or F/T-treated cells (Fig. 4f). Together, these results indicate that TTFields-treated cells promote phagocytosis by DCs, DC maturation in vitro, and leukocyte recruitment in vivo.

To evaluate the effect of concurrent application of TTFields and anti-PD-1 therapy on normal lung tissue, non-tumor-bearing C57Bl/6 mice were treated with TTFields, anti-PD-1, or the combination of the two modalities. Histopathological analysis of the lungs determined that there were no pathological changes in the lungs from the different treatment groups and that the leukocytes level was also within the normal limits in all treatment groups (Supplementary Fig. 5). To further evaluate the effect of concurrent application of TTFields and anti-PD-1 therapy on tumors, C57Bl/6 mice orthotopically implanted with LLC-1 cells were treated with TTFields (Supplementary Fig. 6), anti-PD-1, or the combination of the two modalities (Fig. 5a). Mice treated with anti-PD-1 and TTFields monotherapies demonstrated decreased tumor volume as compared to the control group, although statistical significance was not reached (Fig. 5b). The combined treatment of TTFields and anti-PD-1 led to a significant decrease in tumor volume as compared to all the other groups. A significant increase in leukocyte infiltration (CD45+) was observed in both groups receiving anti-PD-1 injections (Fig. 5c). We next characterized the frequency of specific myeloid populations to the tumors. Specifically, we found a significantly higher frequency of macrophages (CD45+/CD11b+/F4/80+) and DCs (CD45+/CD11c+) in tumors from mice that were concomitantly treated with TTFields and anti-PD-1. There were no significant differences in the frequency of macrophages and DCs between mice treated with TTFields alone or anti-PD-1 alone and the control group. A trend toward increase in these cell populations was observed in mice treated with anti-PD-1 injections (Fig. 5d, e). We also examined whether PD-L1 expression levels, associated with response to anti-PD-1 therapy and adaptive immune resistance, had changed in these myeloid populations following the different treatments. The PD-L1 expression levels of tumor-infiltrating CD45+ cells were increased in tumors from mice treated with TTFields in combination with anti-PD-1 as compared to the control group, suggesting elevated inflammatory response in these tumors. No significant differences were observed between the other groups (Fig. 5f). Specifically, a significant upregulation of surface PD-L1 expression was demonstrated in macrophages and DCs in tumors from mice treated with anti-PD-1 and TTFields, suggesting an adaptive immune attempt to limit the inflammatory response elicited by the combined treatment (Fig. 5g, h) [21]. There were no significant differences in the PD-L1 levels of macrophages and dendritic cells between mice treated with TTFields or anti-PD-1 monotherapies and the control mice injected with the isotype antibody. Taken together, these results suggest that the combination of TTFields and anti-PD-1 augmented the immune response resulting in improved tumor control.

To directly address whether treatment with TTFields plus anti-PD-1 promotes antitumor immunity, we first tested how the presence of TILs was affected by treatment with the different treatment modalities. We found that treatment with TTFields or anti-PD-1 alone had no effect on the frequency of CD8+ and CD4+TILs, while treatment with TTFields plus anti-PD-1 showed a trend toward an increased number of these cells, but this varied between experiments and did not reach statistical significance (Fig. 5i–k). We did not observe any significant changes in the levels of regulatory CD4+Foxp3+TILs between the different treatment groups. TILs functionality was tested in tumor-isolated lymphocytes following in vitro stimulation with anti-CD3 and anti-CD28 (Fig. 5l, m). We found that the combined treatment of TTFields and anti-PD-1 led to a significant increase in IFN-γ production in cytotoxic CD8+TILs (Fig. 5l). The comparable results were obtained when we used the CT-26 tumor model (Fig. 6a), where mice treated with combination of TTFields and anti-PD-1 demonstrated decreased tumor volume as compared to all the other treatment groups (Fig. 6b). In addition, a significant increase in leukocyte infiltration was observed in tumors from mice receiving the combined treatment, as compared to all the other groups (Fig. 6c). The relative fraction of macrophages from tumor leukocytes did not change between the different treatment groups (Fig. 6d), and the relative fraction of DCs from tumor leukocytes has significantly decreased in tumors from mice receiving TTFields alone or the combined treatment (Fig. 6e), although their percentage of total tumor cells remain unaltered between the different treatments (data not shown). This can indicate that other populations of immune cells might contribute to the overall increase in CD45+ cells as observed in CT-26 tumors following treatment with TTFields and anti-PD-1. Furthermore, the PD-L1 expression levels decreased in macrophages from tumors treated with TTFields in combination with anti-PD-1 and TTFields alone (Fig. 6g). No significant differences were observed in DCs PD-L1 levels between the different treatment groups (Fig. 6h). While we did not observe significant change in the infiltration pattern of TILs from LLC-1 tumor, we identified significant increase in both CD8 and CD4 T-cells in the combined treatment tumors, and CD8 in tumors treated with anti-PD-1 alone (Fig. 6i, j, Supplementary Fig. 7). Similarly, to the LLC-1 model, we did not observe any effect on the frequency of CD4+Foxp3+TILs in any of the treatment groups (Fig. 6k).