Activation of Tumor-Cell STING Primes NK-Cell Therapy

Abstract

Activation of the stimulator of interferon genes (STING) pathway promotes antitumor immunity but STING agonists have yet to achieve clinical success. Increased understanding of the mechanism of action of STING agonists in human tumors is key to developing therapeutic combinations that activate effective innate antitumor immunity. Here, we report that malignant pleural mesothelioma cells robustly express STING and are responsive to STING agonist treatment ex vivo. Using dynamic single-cell RNA sequencing of explants treated with a STING agonist, we observed CXCR3 chemokine activation primarily in tumor cells and cancer-associated fibroblasts, as well as T-cell cytotoxicity. In contrast, primary natural killer (NK) cells resisted STING agonist-induced cytotoxicity. STING agonists enhanced migration and killing of NK cells and mesothelin-targeted chimeric antigen receptor (CAR)-NK cells, improving therapeutic activity in patient-derived organotypic tumor spheroids. These studies reveal the fundamental importance of using human tumor samples to assess innate and cellular immune therapies. By functionally profiling mesothelioma tumor explants with elevated STING expression in tumor cells, we uncovered distinct consequences of STING agonist treatment in humans that support testing combining STING agonists with NK and CAR-NK cell therapies.

Figure 1:. STING is highly expressed in immune-exhausted MPM

A, STING immunohistochemistry (IHC) quantified using QuPath software. MPM = malignant pleural mesothelioma; SCLC = small-cell lung carcinoma; NSCLC = non-small cell lung carcinoma. Scale bar = 100 μm. B, Immune cell flow cytometry analysis of MPM specimens (n = 29). C and D, Flow cytometry from freshly resected MPM specimens with number of samples specified. Blood for NK-cell analysis was collected from patients with head & neck squamous cell carcinoma or oral proliferative verrucous leukoplakia. Mann-Whitney test: ***p< 0.001. Flow cytometry antibody details provided in Supplementary Table 3 TIM-3 = T-cell immunoglobulin and mucin domain-containing protein 3; PD-1 = programmed cell death protein 1; LAG3 = lymphocyte activation gene 3; EMRA = effector memory re-expressing CD45 RA; EM = effector memory; CM = central memory.

Figure 2:. STING agonists promote antitumor immunity in MPM

A, Schematic for generation of patient-derived organotypic spheroids (PDOTS) by size filtration and representative flow cytometry profiling (n=16) for MPM case #35. B, Hoechst/propidium iodide live/dead IF from MPM case #35 after 6-day treatment with 50 μM ADU-S100 (ADU) or dH20 and 100ng/mL IFNβ controls. Scale bars = 100 μm. Cell area and percent live/dead quantification of each stain. T-test vs. dH20 control: **p<0.01, ****p<0.0001. C, Summary of percent change in live cell area (Hoechst/acridine orange) and cell death (propidium iodide) in 35 MPM patient specimens treated for 6 days with 50 μM ADU-S100 or dH20 control. Response shown by criteria for live cell area (≥30% decrease) and ≥20% increase in cell death, with p<0.05 by t-test between treated triplicate wells, ADU-S100 vs. dH20 control. D, Waterfall plot from the 35 patient specimens treated for 6 days with 50 μM ADU-S100 or dH20 control and analyzed in C. R = response by reduced live cell area, E = epithelioid MPM, B = biphasic MPM, Y/N = yes/no neoadjuvant treatment, M/F = male/female. E, Percent live/dead for MPM case #34 after 6-day treatment with 50 μM ADU-S100 or dH20 control and CD8 blocking antibody (αCD8). T-test vs. dH20 control: **p<0.01. F, Waterfall plot from 13 MPM patient specimens treated for 6 days with 50 μM ADU-S100, TAK-676, or dH20 control. G, Percent live/dead for MPM case #26 after 6-day treatment with 50 μM ADU-S100, 1 μM TAK-676, or dH20 control and CD8 blocking antibody (αCD8). T-test vs. dH20 control: *p<0.05.

Figure 3:. STING agonists activate tumor cells and fibroblasts

A, Combined UMAP plots from broad clustering of scRNA-seq of MPM case #26 after 24 hours of treatment with dH20 control, 10 μM ADU-S100 or 50 μM ADU-S100. B, Combined UMAP plot for CXCR3 ligands (CXCL9/CXCL10/CXCL11) and mesothelin (MSLN). C, Violin plots for select ISG transcripts from combined broad clustering split by treatment condition/dose of ADU-S100. D, UMAP plots from combined samples overlayed with contour plots showing the density of cells in each individual sample, normalized to number of cells per sample. Fraction bar graph for each cluster by treatment, normalized to number of cells per sample.

Figure 4:. STING agonists are toxic to T cells but not NK cells

A, Immune flow cytometry in MPM case #12 after 3-day treatment with dH20 control, 50 μM ADU-S100, or 100ng/mL IFNβ. B, Cell-titer glow proliferation after 24-hours of treatment of MPM case #12 with dH20 control, 50 μM ADU-S100, or 100ng/mL IFNβ in the presence of primary T cells or BCMA CAR T cells. T-test vs. dH20 control: ****p<0.0001. C, Flow cytometry after 72-hour treatment of MPM case #12 with 50 μM ADU-S100, 10 μM TAK-676, or dH20 control +/− 200 U/mL IL2 in the presence of TILs or NK cells, gating for live cells out of 10,000 total events expressing CD8 or CD56. Batch 3 primary NK cells expanded from PBMCs and batch 3 TILs. One-way ANOVA p<0.01 with corrected pairwise comparisons: **p<0.01, ***p<0.001. D, Western blot for STING, phospho-TBK1 (pTBK1), TBK1, phospho-STAT1 (pSTAT1), STAT1, and beta-actin loading control in TILs and NK cells from C treated for 3 or 6 hours with 50 μM ADU-S100 or dH20 control in triplicate.

Figure 5:. STING agonists enhance NK-cell therapies

A, NK-cell therapy schematic. Quantification of Hoechst/propidium iodide live/dead IF from PDOTS sample #36 treated for 6 days with 50 μM ADU-S100 or dH20 control −/+ primary NK cells (cNK). One-way ANOVA p<0.001 with corrected pairwise comparisons: **p<0.01. B, Representative live/dead IF and quantification from sample #37 treated for 6 days with 50 μM ADU-S100 or dH20 control −/+ primary NK cells and/or 5 μg/mL CXCR3 blocking antibody (αCXCR3). T-test: **p<0.01. Scale bar = 50 μm. C, Quantification of percent change in live cell area (Hoechst IF) in PDOTS from MPM case #32 after 6- or 10-day treatment with 50 μM ADU-S100, cNK, or MSLN CAR-NK cells. One-sample t-test with expected difference of zero: *p<0.05. D, Representative live/dead IF from MPM case #32 after 10-day treatment. Scale bar = 50 μm.

Figure 6:. STING agonists enhance adoptive NK-cell migration and killing

A, Representative overlayed IF and brightfield images of labeled primary NK cells (blue) migrating toward H226 MPM cells with propidium iodide dead stain (red) after 4-day treatment with 50 μM ADU-S100 or dH20 control. Scale bar = 100 μm. Quantification of triplicate NK-cell migration (T-test: *p<0.05) and cell death (one-way ANOVA p<0.01 with corrected pairwise comparisons: *p<0.05). B Quantification of triplicate NK-cell migration towards H226, focused on the center of the chamber after 4-day treatment with 50 μM ADU-S100 +/− 5 μg/mL CXCR3 blocking antibody (αCXCR3). T-test: *p<0.05. C, Representative immunofluorescence modeling NK-cell migration through vasculature over 24-hours of STING agonist treatment (ADU-S100 or TAK-676) to reach H226 tumor cells. Scale bar = 200 μm. Quantification of triplicate migration +/− vasculature during 24-hour treatment with 50 μM ADU-S100, 1 μM TAK-676, or dH20 control. T-test: **p<0.01, ***p<0.001. D, Flow cytometry for annexin V and live/dead with H2591 MPM cells in co-culture with control NK (cNK) or MSLN CAR-NK cells (E:T 2:1) following 6-hour treatment with 50 μM ADU-S100 or dH20 control. Quantification of four NK-cell donors, graphed as percent change with ADU-S100 vs. dH20 control treatment. One-sample t-test with expected difference of zero: *p<0.05, **p<0.01.