Retinoic acid receptor activation reprograms senescence response and enhances anti-tumor activity of natural killer cells

Abstract

Cellular senescence can exert dual effects in tumors, either suppressing or promoting tumor progression. The senescence-associated secretory phenotype (SASP), released by senescent cells, plays a crucial role in this dichotomy. Consequently, the clinical challenge lies in developing therapies that safely enhance senescence in cancer, favoring tumor-suppressive SASP factors over tumor-promoting ones. Here, we identify the retinoic-acid-receptor (RAR) agonist adapalene as an effective pro-senescence compound in prostate cancer (PCa). Reactivation of RARs triggers a robust senescence response and a tumor-suppressive SASP. In preclinical mouse models of PCa, the combination of adapalene and docetaxel promotes a tumor-suppressive SASP that enhances natural killer (NK) cell-mediated tumor clearance more effectively than either agent alone. This approach increases the efficacy of the allogenic infusion of human NK cells in mice injected with human PCa cells, suggesting an alternative therapeutic strategy to stimulate the anti-tumor immune response in "immunologically cold" tumors.

Keywords: AP-1; NK-Killing; RAR; SASP; adapalene; allogenic infusion; immunotherapy; metabolism; prostate cancer; senescence.

Graphical abstract

Figure 1

Identification of RAR agonists as pro-senescence compounds (A) PC3 cell proliferation fold change with specified compounds, normalized to time 0. (B) Fold change in PC3shTIMP1 cell proliferation treated with specified compounds, normalized to time 0. (C) Fold change in LNCaP cell proliferation treated with specified compounds, normalized to time 0. (D) Fold change in 22RV1 cell proliferation treated with specified compounds, normalized to time 0. (E) SA-β Gal quantification in treated PC3 cells. (F) SA-β Gal quantification in treated PC3shTIMP1 cells. (G) SA-β Gal quantification in treated LNCaP cells. (H) SA-β Gal quantification in treated 22RV1 cells. (I) RT-qPCR analysis of senescence markers in PC3 cells. (J) RT-qPCR analysis of senescence markers in PC3shTIMP1 cells. (K) RT-qPCR analysis of senescence markers in LNCaP cells. (L) RT-qPCR analysis of senescence markers in 22RV1 cells. (M) Experimental design (upper panel) and western blot quantification (lower panel) of treated PC3shTIMP1 cells, referred to Figure S1J. (N) SA-βGal quantification of PC3shTIMP1 cells treated with p21 inhibitor (p21i) UC2288 and adapalene. (O) Fold change in proliferation normalized to time 0 of PC3shTIMP1 cells treated with the p21 inhibitor (p21i) UC2288 and adapalene (left panel) and cristal violet assay quantification (right panel). Data presented in (E–N) and (O, right) are shown as the mean ± SD; Data presented in (A–D) and (O, left) are presented as mean ± SEM. Data represent three to five independent experiments. Data presented in (M) represents one experiment. Statistical test used in data presented in (E–H): one-way ANOVA followed by Tukey’s test ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, ^∗∗∗∗p < 0.0001. Statistical test used in data presented in (A–D) and (I–O): two-way ANOVA ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, ^∗∗∗∗p < 0.0001. Also see Figure S1.

Figure 2

The retinol metabolism and pathway are impaired during PCa progression (A) Venn diagrams depicting the number of differentially expressed genes in each RA-associated signature. (B) Principal component analysis (PCA) of pan-prostate cancer transcriptomes obtained from the indicated studies of normal prostate (green), and primary (purple), castration-resistant (CRPC, orange), and neuroendocrine prostate cancer (NEPC, yellow). (C) PCA showing the activation of RAR signaling pathway during PCa disease progression. (D) PCA showing the expression of RAR downstream targets during PCa disease progression. (E) PCA showing the expression of RA metabolism related genes during PCa disease progression. (F) Plot representing the Pearson correlation of RA-related genes from RNA-seq data of patients. Gray dots represent genes not related to the RAR pathway, blue dots represent genes related to RA metabolism, green dots represent genes that are RAR targets, and pink dots represent genes related to the RAR signaling. (G) Barplot showing the NES of RAR downstream targets, RA-associated metabolism and RAR signaling from RNA-seq of PC3shTIMP1 cells upon adapalene treatment compared to untreated cells. (H) Plot representing the Pearson correlation of the RA-related genes from RNA-seq data of patients and PC3shTIMP1 cells treated with adapalene. Gray dots represent genes not related to the RAR pathway, black dots represent genes downregulated upon adapalene treatment, and red dots represent genes upregulated upon adapalene treatment. (I) Experimental design of (J). (J) Fold change in ROH conversion products (atRA and 13-cisRA) in a non-transformed prostate cell line, RWPE-1, and a cancer cell line, 22RV1 upon ²H-ROH exposure. Data presented in (J) are shown as the mean ± SD representing one experiment. Statistic test used in (J): one-way ANOVA followed by Tukey’s test ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, ^∗∗∗∗p < 0.0001. See also Figure S2.

Figure 3

RARs activation enhances chemotherapy by reprogramming the SASP (A) Heatmap showing the results of the synergistic proliferation assay performed on PC3shTIMP1 cells treated with the combination of either docetaxel and adapalene or docetaxel and bexarotene (48 h of treatment). (B) Quantification of SA-β Gal assay performed in PC3shTIMP1 cells after two days of treatment with docetaxel, adapalene, and their Combo (left), and docetaxel, bexarotene, and their Combo (right). (C) RT-qPCR analysis for the senescence markers in PC3shTIMP1 cells treated with docetaxel, adapalene, and the Combo. (D) Fold change in PC3shTIMP1 cells proliferation normalized on day 7 in continuum treatment with the indicated drugs for 12 days (media refreshed every three days). (E) Pathway analysis performed on the RNA-seq of PC3shTIMP1 cells, showing the upregulated and downregulated pathways in Combo vs. docetaxel and docetaxel vs. vehicle. Pathway reported with FDR <0.05. (F) Cytokine array of PC3ShTIMP1 cells treated with the indicated drugs and normalized on the vehicle-treated cells. (G) Experimental design of (H and I). (H) Quantification of Boyden chamber migration assay performed on PC3shTIMP1 cells exposed to the cm of docetaxel, adapalene, or the Combo, expressed as fold change in OD590. (I) Representative images of Boyden chambers stained with cristal violet. (J) Experimental design of (K and L). (K) Percentage of wound density of PC3shTIMP1 parental cells treated with conditioned media (cm) from docetaxel, adapalene, or Combo-treated PC3shTIMP1 cells. (L) Representative pictures of K (scale bar: 400 μm). (M) Experimental design of (N and O). (N) Proliferation of TdTomato⁺ RapidCaP cells cocultured with senescent non-labeled RapidCaP cells, quantified as fold change in total integrated intensity (RCU x μm²/image). (O) Representative images of the proliferation assay in N (scale bar: 400 μm, inserts 200 μm). Data presented in (B), (C), and (H) are shown as the mean ± SD. Statistic test used for (H): one-way ANOVA followed by Tukey’s test ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, ^∗∗∗∗p < 0.0001. Data presented in (D), (K), and (N) are shown as the mean ± SEM. Statistic test used in (B–D), (K), and (N): two-way ANOVA ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, ^∗∗∗∗p < 0.0001. Data presented in (F) are representative of one experiment. Data shown in panels (A–D), (H), (K), and (N) are representative of three independent experiments. See also Figure S3.

Figure 4

Adapalene reprograms the detrimental SASP triggered by docetaxel through AP-1 (A) Heatmap showing the differentially expressed transcription factors in PC3shTIMP1 cells treated with docetaxel, adapalene, and the Combo. Genes reported with FDR <0.05 in at least one comparison to the vehicle. (B) Focus on the heatmap showing the differentially expressed transcription factors in PC3shTIMP1 cells having an opposite expression pattern in cells treated with docetaxel compared to vehicle and in the Combo compared to docetaxel. (C) Schematic representation of (D and E). (D) Representative pictures of AP-1-mediated transcriptional activition depicted as mCherry fluorescence. (E) Quantification of mCherry positive PC3shTIMP1 cells upon docetaxel, adapalene, or Combo treatments in the presence or absence of the AP-1 inhibitor (AP-1i) T-5224. (F) Quantification of SA-β Gal staining of PC3shTIMP1 cells treated with docetaxel, adapalene, or the Combo in the presence or absence of the AP-1i. (G) Representation of the experimental design (upper panel) and fold change in proliferation of PC3shTIMP1 cells treated with the cm of prostate tumor cells treated with docetaxel or the Combo in the presence or absence of the AP-1i, normalized on t0 (lower panel). (H) Experimental design (upper panel) and wound confluence (%) of PC3shTIMP1 cells treated with the cm of cells treated with docetaxel and the Combo (lower panel). (I) Representative pictures of the wound healing assay of (H) (scale bar: 400 μm). (J) Fold change in mRNA levels of the indicated tumor-promoting secreted factors. (K) Fold change in mRNA levels of different activators of the anti-tumor immune response. Data presented in (G) and (H) are shown as mean ± SEM. Statistic test used: two-way ANOVA, ∗∗∗∗p < 0.0001. Data are representative of three independent experiments. Data presented in (E), (F), (J), and (K) are shown as mean ± SD. Statistic test used in (E): one-way ANOVA followed by Tukey’s test ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, ^∗∗∗∗p < 0.0001. Data presented in (F–H), (J), and (K): two-way ANOVA ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, ^∗∗∗∗p < 0.0001. Data represent two independent experiments. See also Figure S4.

Figure 5

Adapalene, in combination with docetaxel, enhances senescence and activates an anti-tumor immune response (A) Experimental design. (B) Tumor volume in mm³ of mice treated with the indicated drugs (n = 5 for each group) (left), and area under the curve calculated on tumor volume (right). (C) Immunofluorescence staining for NK cells (by using anti NK1.1 antibody, in red) and cleaved caspase 3 (CC3, in green) in tumor sections (scale bar: 21 μm). (D) Quantification of CC3 positive cells for field in tumors sections from (C). (E) FACS plot for NK1.1 and CD45⁺ cells gated in live cells in tumors treated with the indicated drugs. (F) Percentage of NK1.1⁺ cells in tumors treated with the indicated drugs. (G) Immunofluorescence staining for NK cells (by using anti NKp46 antibody, in green) in tumor sections (scale bar: 33.1 μm). (H) Quantification of NK cells for field in tumors sections from (G). (I) Fold change in mRNA levels of recruiting factors for NK cells and granzyme B and IFNγ in tumors treated with the indicated drugs. (J) Schematic representation of the experimental design of (K and L). (K) Percentage of tumor-infiltrating NK cells (NK1.1⁺, CD3⁻, NKp46⁺) gated in CD45⁺ cells in tumors of mice untreated or treated with Combo and the Anti-GM1 Asialo. (L) Tumor volume in mm³ of mice treated with the indicated drugs in the presence or absence of the depleting antibody for NK cells (Vehicle, Iso n = 4, Vehicle, Asialo n = 4, Combo, Iso n = 3, Combo, Asialo n = 4). Data presented in (B) and (L) are shown as mean ± SEM. Statistic test used: two way ANOVA ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, ^∗∗∗∗p < 0.0001. Data in (B) are representative of two independent experiments. Data presented in (L) are representative of one experiment. Data presented in (D), (F), (H), (I), and (K) are shown as mean ± SD. Statistic test used: one-way ANOVA followed by Tukey’s test ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, ^∗∗∗∗p < 0.0001. Data represent one experiment. See also Figure S5.

Figure 6

RAR agonist in combination with docetaxel potentiates the anti-tumor activity of human-derived NK cells (A) Experimental design. (B) Representative images showing cytotox emission (in green) of PC3 cells pre-treated with docetaxel and Combo and co-cultured with human NK cells (scale bar: 400μm). (C) Quantification of cytotox emission expressed as fold change in total integrated intensity (GCU x μm²/image) of PC3 cells pre-treated with the indicated compounds and co-cultured with human NK cells (target:effector 1:5 ratio). (D) Quantification of cytotox emission expressed as fold change in total integrated intensity (GCU x μm²/image) of LNCaP cells pre-treated with the indicated compounds and co-cultured with human NK cells (target:effector 1:5 ratio). (E) Quantification of cytotox emission expressed as fold change in total integrated intensity (GCU x μm²/image) of 22RV1 cells pre-treated with the indicated compounds and co-cultured with human NK cells (target:effector 1:5 ratio). (F) Schematic representation of the experimental layout of (G). (G) Caspase 3/7 integrated intensity emission normalized on time 0 (t0) in PC3 cells treated with the indicated compounds (left) and engineered with different shRNAs: EV (circle), shIL-33 (square), and ShIL-12 (triangle), (right). (H) RT-qPCR analysis in PC3 cells showing the expression of NK ligands on the cancer cells. (I) FACS analysis showing the percentage of positive PC3 cells expressing ULBP2. (J) FACS analysis showing the percentage of positive PC3 cells expressing MICA. (K) Schematic representation of the experimental design of (L). (L) Cytotox green emission integrated intensity normalized on time 0 (t0) in PC3 cells treated with the indicated compounds in the presence or absence of neutralizing antibody against NKG2D. (M) Representation of the experimental design of (N and O). (N) Tumor volume in mm³ of mice treated with vehicle (n = 15), vehicle+NK (n = 16), docetaxel (n = 7), docetaxel+NK (n = 7), Combo (n = 15), Combo+NK (n = 16). Mice from vehicle, vehicle NK, docetaxel, and docetaxel NK were euthanized at an earlier time point (day 44) because of tumor ulceration. (O) FACS analysis showing the percentage of hCD56⁺ cells gated in hCD45⁺, mCD45^-. Data presented in (G), (I), (J), and (O) are shown as mean ± SD. Data presented in (C–E), (L), and (N) are shown as mean ± SEM. Statistic test used in (C–E), (G), and (N): two-way ANOVA ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, ^∗∗∗∗p < 0.0001. Statistic test used in (I) and (J): one-way ANOVA followed by Tukey’s test ^∗∗∗p < 0.001. Data presented in (C–E), (G–J), (L), and (N) represent two independent experiments. Data presented in (O) represent one experiment. Statistic test used in (L): two-way ANOVA. Lines with different symbols are statistically different from each other. See also Figure S6.