Tipping the immunostimulatory and inhibitory DAMP balance to harness immunogenic cell death

Abstract

Induction of tumor cell death is the therapeutic goal for most anticancer drugs. Yet, a mode of drug-induced cell death, known as immunogenic cell death (ICD), can propagate antitumoral immunity to augment therapeutic efficacy. Currently, the molecular hallmark of ICD features the release of damage-associated molecular patterns (DAMPs) by dying cancer cells. Here, we show that gemcitabine, a standard chemotherapy for various solid tumors, triggers hallmark immunostimualtory DAMP release (e.g., calreticulin, HSP70, and HMGB1); however, is unable to induce ICD. Mechanistic studies reveal gemcitabine concurrently triggers prostaglandin E₂ release as an inhibitory DAMP to counterpoise the adjuvanticity of immunostimulatory DAMPs. Pharmacological blockade of prostaglandin E₂ biosythesis favors CD103⁺ dendritic cell activation that primes a Tc1-polarized CD8⁺ T cell response to bolster tumor rejection. Herein, we postulate that an intricate balance between immunostimulatory and inhibitory DAMPs could determine the outcome of drug-induced ICD and pose COX-2/prostaglandin E₂ blockade as a strategy to harness ICD.

Fig. 1. The release of hallmark DAMPs from gemcitabine-induced dying cancer cells.

Representative peptide chromatogram illustrating differential enrichment of a CRT, b HSP70, c HSP90aa1, and d PDIA3 on the cell surface, as well as e ANXA and f HMGB1 released into the cultured media by cancer cells (murine G69 bladder cancer cells) treated with gemcitabine (n = 2 independent mass spectrometry proteomic profiling experiments). ANXA1 annexin A1, CRT calreticulin, gemCTx gemcitabine chemotherapy-treated, HMGB1 high-mobility group protein box 1, HSP heat-shock protein, PDIA3 protein disulfide isomerase A3. Extracted chromatograms for a representative peptide of each protein was plotted using the Skyline software; precursor (blue color), precursor [M + 1] (purple color), precursor [M + 2] (red color) represent the three isotopic peaks for each indicated peptide.

Fig. 2. Hallmark DAMP release is insufficient to induce immunogenic cell death.

a, b Flow cytometry analysis and validation of DAMPs (i.e., CRT and HSP70) on the cell surface of human T24 and murine G69 bladder cancer cells treated with gemcitabine in vitro (representative plot shown with two technical replicates of n = 3 independent experiments; example gating depicted in Supplementary Fig. 3). c, d Western blot of HMGB1, an extracellular DAMP, released into cultured media that were collected from gemcitabine-treated human T24 cells and murine G69 cells, respectively. e A schematic depiction of the in vivo vaccination assay treatment schedule for assaying immunogenic cell death. f Tumor volume and g corresponding tumor-free survival Kaplan–Meier plot (n = 3 mice per vaccination group; tumor volume calculation as indicated in Methods section) resulting from mice vaccinated with dying G69 cells via various chemotherapy and subsequently challenged with live G69 murine bladder cancer cells. cis cisplatin, gem gemcitabine, mitox mitoxantrone. Statistics: two-tailed, unpaired T test (a; p = 0.0013 [top] and p = 0.0006 [bottom]) and (b; p = 0.0044 [top] and p = 0.0123 [bottom]); two-tailed, two-way ANOVA-Tukey’s multiple comparisons test (f; p < 0.0240); Kaplan–Meier survival analysis using Mantel–Cox test (g; p = 0.0123); and where appropriate, data are presented as mean values ±SEM.

Fig. 3. Prostaglandin E₂ release hinders ICD-induction by gemcitabine.

a Western blot analysis of COX-2 and corresponding ELISA of PGE₂ from G69 cells treated with gemcitabine in vitro (representative plot shown with two technical replicates of n = 3 independent experiments). b A schematic depicting the proposed interventional strategy that exploits the immunostimulatory/inhibitory DAMP balance to harness immunogenic cell death. c, d Flow cytometry of cell surface CRT and HSP70 of single cells (example gating depicted in Supplementary Fig. 3), as well as western blot of HMGB1 from G69 and Panc02 cells treated with gemcitabine plus celecoxib to achieve iDAMP blockade (representative plot shown with two technical replicates of n = 3 independent experiments). e, f Western blot of COX-2 and ELISA of PGE₂ from murine G69 and Panc02 cells to demonstrate the efficacy of celecoxib in abrogating gemcitabine-induced PGE₂/ iDAMP release in vitro (representative plot shown with two technical replicates of n = 3 independent experiments). g Temporal changes in tumor volume and h tumor-free survival of mice after challenge (n = 9 per treatment group) resulting from a gold-standard vaccination assay. i Temporal changes in tumor volume and j tumor-free survival of mice after challenge (n = 5 per treatment group) resulting from a gold-standard vaccination assay using the murine Panc02 PDAC model. celex celecoxib, CTx chemotherapy, gem gemcitabine. Statistics: two-tailed, unpaired T test (a; p = 0.0107); two-tailed, one-way ANOVA-Tukey’s multiple comparisons test (c; p = 0.002 [top] and p = 0.008 [bottom]), (d; p = 0.0011 [top] and p = 0.0007 [bottom]), (e; **p = 0.0076 and **p = 0.0028), (f; p = 0.0033), (g; **p = 0.015 and ***p = 0.0003), and (i; *p < 0.0486); Kaplan–Meier survival analysis using Mantel–Cox test (h; p = 0.0001) and (j; p = 0.004); and where appropriate, data are presented as mean values ±SEM.

Fig. 4. An inhibitory signal 0 counteracts DAMPs to mitigate immunogenic dendritic cell maturation.

a A schematic depicting the workflow utilized to assay CD103⁺ BMDC activation: CD103⁺ BMDCs were incubated with cultured media from cancer cells pre-treated with gemcitabine ± iDAMP blockade in vitro. PGE₂ neutralizing antibody or celecoxib were implemented as independent approaches to block PGE₂ action. b Heat-map derived from Fluidigm Biomark™ analyzing surrogate genes representing CD103⁺ BMDC activity that were collected from a. Log2 values were calculated from normalized ct values of non-treated CD103⁺ BMDCs as baseline control. c qPCR validating genes associated with immunogenic versus tolerogenic dendritic cells. Relative mRNA expression was normalized to Gapdh and to gemCTx-treated CD103⁺ BMDCs (representative plot shown with two technical replicates of n = 3 independent experiments). d Representative flow cytometry histogram plots of CD103⁺ BMDCs 24 h post-cultured media treatment (n = 3 independent experiments; example gating depicted in Supplementary Fig. 7). mAb monoclonal antibody, NS statistically non-significant. Statistics: two-tailed, one-way ANOVA-Tukey’s multiple comparisons test (H2-k; p < 0.0001), (Cd40; ***p = 0.0002 and ****p < 0.0001), (Il-12b; *p = 0.0475 and **p = 0.0013); (Il-2; *p = 0.323 and **p = 0.0069), (Ifng; **p = 0.0017 and ****p < 0.0001), (Tnfa; *p = 0.0347 and **p = 0.0024), (Arg1; p < 0.0001), (Ido1; p < 0.0001), (Pd-l1; *p = 0.0103 and ***p = 0.0001), (Tim3; p < 0.0001); and where appropriate, data are presented as mean values ±SEM.

Fig. 5. COX-2/PGE₂ blockade promotes priming of a CD8⁺ Tc1-mediated immune response.

a A schematic depiction of the experimental procedures implemented to analyze vdLN (i.e., popliteal) and PB-circulating CD8⁺ T cells. b Representative gating-strategy implemented to analyze/sort CD8⁺ T cells from vdLN and PB by flow cytometry. c Quantification of PB-circulating CD8⁺ T cells post-vaccination (n = 3 per treatment group). d qPCR analysis of genes associated with functionally activated Tc1-polarized CD8⁺ T cells. e An illustration depicting the adapted in vivo vaccination assay to interrogate vdLN CD8⁺ T cells. f–i Flow cytometry analysis and quantification of vdLN CD8⁺ T cells 5-days post-vaccination (non-dLN n = 4; vdLN n = 5 per group). PB peripheral blood, PLN popliteal lymph node. Statistics: two-tailed, two-way ANOVA-Tukey’s multiple comparisons test (c; *p = 0.387); two-tailed, one-way ANOVA (d, T-bet; p = 0.0483), (d, Tnfa; p = 0.0080), (d, Ifng; p = 0.0002), (d, Gzmb; p = 0.0379), (f, T-bet; *p = 0.0136 and **p = 0.0089), (f, IFNg; ***p = 0.0008 and ****p < 0.0001), (f, CD107a; **p = 0.0071 and ***p = 0.0004), (h, GATA3; p = 0.0115), (h, RoRyt; ****p < 0.0001), and (h, Foxp3; *p = 0.0103 and ***p = 0.0008); and where appropriate, data are presented as mean values ±SEM.

Fig. 6. CD8⁺ T-cell-mediated immune response drives tumor rejection upon challenge.

Corresponding vaccination assay resulting from two methodologies of iDAMP blockade: pharmacological (i.e., celecoxib) and genetic (CRISPR/Cas9 KO). a Tumor volume and b tumor-free survival from the same vaccination assay comparing two methods to deplete PGE₂/iDAMP and their efficacies in affecting drug-induced immunogenic cell death. Corresponding aCD8 mAb treatment results as shown in c, d Vaccine groups: gem (n = 7); gemCelex (n = 8); gemCox-2^−/− (n = 15); gem/aCD8 (n = 4); gemCelex/aCD8 (n = 4); gemCox-2^−/−/aCD8 (n = 5). Statistics: two-tailed, two-way ANOVA-Tukey’s multiple comparisons test (a; **p < 0.0028, ***p < 0.0006, and ****p < 0.0001) and (c; p = 0.0452); Kaplan–Meier survival analysis using Mantel–Cox test (b; p < 0.0001); and where appropriate, data are presented as mean values ±SEM.

Fig. 7. schematic depiction of the proposed immunostimulatory/inhibitory DAMP balance model in modulating drug-induced immunogenic versus tolerogenic cell death fate.

a The modulation of immunogenic cell death via iDAMP blockade. b Signal 0 skews the maturation phenotype of dendritic cells. c Resulting polarization and antitumoral CD8⁺ T cell response.