PARP inhibitor radiosensitization enhances anti-PD-L1 immunotherapy through stabilizing chemokine mRNA in small cell lung cancer

Abstract

Immunotherapy (IO) is an effective treatment for various cancers; however, the benefits are modest for small cell lung cancer (SCLC). The poor response of SCLC to anti-PD-1/PD-L1 IO is due in part to the lack of cytotoxic T cells because of limited chemokine expression from SCLC tumors. Immunogenic radiosensitizers that enhance chemokine expression may be a promising strategy forward. Here, we show that the PARP inhibitors (PARPi), including olaparib, talazoparib and veliparib, in combination with radiotherapy (RT) enhance the immune activation and anti-tumor efficacy in SCLC cell lines, patient-derived xenograft (PDX) and syngeneic mouse models. The effect is further enhanced by continued delivery of adjuvant PARPi. The combination treatment (PARPi with RT) activates the cGAS-STING pathway and increases the mRNA levels of the T cell chemo-attractants CCL5 and CXCL10. In addition to upregulation of transcription, the combination treatment increases chemokine CXCL10 protein levels via stabilization of CXCL10 mRNA in an EIF4E2-dependent manner. The incorporation of anti-PD-L1 IO into the PARPi with RT combination therapy further improves the anti-tumor efficacy by increasing T cell infiltration and function. This study thus provides a proof of principle for the combination of PARP inhibitors, RT and anti-PD-L1 IO as a treatment strategy for SCLC.

Fig. 1. PARPi with RT synergistically promotes cGAS-STING dependent chemokine transcription.

a Schematic of experimental design with olaparib and RT treatments in vitro and in vivo (Created in BioRender. Ran, X. (2025) https://BioRender.com/e93s595). b Survival fraction of SBC5 treated with pulse olaparib (PUL) or continuous olaparib (COM) with RT. c The tumor growth curve of SBC5 xenograft treated with VEH/OLA (n = 6 mice), RT/PUL/COM (n = 7 mice). d Upper: Quantification of olive tail movement by alkaline comet assay in treated SBC5 cells (n = 3 biological replicates) for 3 and 48 h. Lower: Representative images of DNA damage detected by comet assay in SBC5 cells upon treatment for 48 h. e Upper: The quantification of γH2AX immunostaining in each group for 3 and 48 h (n = 3 biological replicates). Lower: Representative dsDNA damage detected by γH2AX nuclear foci immunostaining in SBC5 upon treatment for 48 h. f Upper: Quantification of the frequency of cGAS⁺ micronuclei per cell (n = 3 biological replicates). Lower: Representative cGAS⁺ micronuclei staining in SBC5 upon treatment for 2 days. g Upper: Quantification of the relative intensity of p-STING normalized by β-Actin. Lower: western blot of p-STING, STING, and cGAS in SBC5 and SW1271 upon treatment for 3 days from one of two independent experiments. h Heatmap of CCL5 and CXCL10 mRNA in SBC5(Y), H146(A), 82(N), H526(P), H1048(P) and KP1(A) cells after treatment for 2 days and in SBC5 xenograft and SCRX-Lu149(A) PDX for 15 days. i, j Upper: Western blot against cGAS or STING in EGFP KO control (Ctrl) or cGAS KO or STING KO cells for SBC5 and H1048. Lower: CCL5 or CXCL10 mRNA in Ctrl or cGAS/STING KO cells upon treatment for 48 h (n = 3 biological replicates). For (c), the data are presented as mean ± SD. For (d) and (e), mean counts of olive movement or γH2AX nuclear foci per cell are presented as a line. For (b), (d)–(f), (i), (j), the data are presented as mean ± SD of three replicates for each treatment. All comparisons were calculated with one-way or two-way ANOVA. (p: ****<0.0001 < *** < 0.001 < ** < 0.01 < * < 0.05). All exact p values are listed in the source data. Source data are provided as a Source Data file.

Fig. 2. Continuous olaparib combined with RT (COM) significantly downregulates EIF4E2 in SCLC.

a Heatmap showing hierarchical clustering of significantly changed genes in SBC5 cells treated with vehicle (VEH), olaparib (OLA), RT, or combination of olaparib and RT (COM) for 3 days. b Top 10 pathways significantly modulated by the COM. The p value was calculated by GESA with two-tailed permutation testing. c Fold change of EIF4 family mRNA level treated with OLA, RT, or COM treatment as normalized by the expression to VEH. d Upper: Quantification of the relative intensity of EIF4E2 protein level normalized by the intensity of β-Actin. Lower: Western blot analysis of EIF4E2 expression in H1048 (P), H446 (N), SBC5 (Y) and KP1 (A) cells treated with vehicle, OLA, RT or COM treatment from one of two independent experiments. e Left: Schematic of KP1 allografts experimental design (n = 5 mice for each treatment, Created in BioRender. Ran, X. (2025) https://BioRender.com/e93s595). Middle: Western blot against EIF4E2 and Actin. Right: Quantification of the relative intensity of EIF4E2 protein level normalized by the intensity of β-Actin (VEH vs COM p = 0.0052, OLA vs COM p = 0.001, RT vs COM p = 0.044). Mean counts are presented as a line and the comparison was calculated with one-way ANOVA. (p: 0.001 < ** < 0.01 < * < 0.05). All exact p values are listed in the source data. Source data are provided as a Source Data file.

Fig. 3. EIF4E2 destabilizes CXCL10 mRNA to decrease CXCL10 protein level, reducing T cell tumor infiltration.

a Left: EIF4E2 and CXCL10 protein level in SBC5 with VEH or COM treatment at indicated time points from one of two independent experiments. Right: Quantification of EIF4E2 and CXCL10 protein level normalized by Actin on different days relative to the protein level at 0 h. b Relative CXCL10 mRNA level in EGFP KO or EIF4E2 KO for HEK293 and SBC5 cells (n = 4 biological replicates). c CXCL10 mRNA stability in EGFP KO control (Ctrl) or EIF4E2 KO cells for HEK293 treated with Dactinomycin (DACT) for 0, 2, 4 h (n = 3 biological replicates). d Western blot and quantification of CXCL10 protein levels in Ctrl or EIF4E2 KO for HEK293 or SBC5 cell-lines from one of two independent experiments. e Relative CXCL10 mRNA level in empty vector (Empty) or EIF4E2 overexpression for HEK293 and SBC5 cells (n = 3 biological replicates). f Left: overexpression of FLAG-EIF4E2 in HEK293 EIF4E2 KO cells. Right: CXCL10 mRNA level in HEK293 EIF4E2 KO cells expressing empty vector or FLAG-EIF4E2 after 4 h’ DACT treatment (n = 3 biological replicates). g Western blot and quantification of CXCL10 protein levels in Empty or V5-EIF4E2 overexpression for HEK293 or SBC5 cell-lines from one of two independent experiments. h Left: Schematic of SBC5-Empty or V5-EIF4E2 overexpression tumor engraftment in PBMC humanized NSG mice (Created in BioRender. Ran, X. (2025) https://BioRender.com/e93s595). Middle: Representative flow cytometry data for total T cells. Right: Cumulative data for total T cell (CD3⁺CD45⁺) in tumors on day 35 (n = 12). For (a), data are presented as individual dot plot and the correlation was analyzed by two-tailed Pearson correlation test. For (b), (c), (e) and (f), data are presented as mean ± SD and statistically analyzed by two-tailed, unpaired t test. For (h), data are presented as dot plot and statistically analyzed by the two-tailed, unpaired t test. (p: ****<0.0001 < *** < 0.001 < ** < 0.01 < * < 0.05). All exact p values are listed in the source data. Source data are provided as a Source Data file.

Fig. 4. EIF4E2 destabilizes CXCL10 mRNA via AU rich element of 3'UTR.

a Top: Schematic of the psicheck2-Rluc-CXCL10 3′UTR reporter. Left: Western blot against EIF4E2 in Ctrl or EIF4E2 KO HEK293 cells. Quantification of relative Renilla luciferase mRNA level (middle) and protein level (right) in EGFP KO control (Ctrl) or EIF4E2 KO HEK293 cells transfected with psicheck2 or psicheck2-CXCL10 3′UTR plasmids (n = 3 biological replicates for mRNA detection and n = 4 biological replicates for protein detection). b Top: Schematic of the GFP-CXCL10 3′UTR reporter. Left: Western blot against EIF4E2 in LacZ KO control (Ctrl) or EIF4E2 KO for SBC5. Right: Quantification of GFP intensity in Ctrl or EIF4E2 KO for SBC5 expressing GFP-CXCL10 3′UTR (n = 4 biological replicates). c Top: Schematic of the psicheck2-Rluc-CXCL10 3′UTR WT and 3′UTR △ARE reporter. Bottom: Quantification of relative Renilla luciferase mRNA stability in Ctrl or EIF4E2 KO HEK293 cells transfected with psicheck2-Rluc-CXCL10 3′UTR WT or △ARE reporter plasmid for 24 h and then treated with Dactinomycin (DACT) for 4 h (n = 3 biological replicates). d Top: Schematic of the CXCL10-ORF + 3′UTR WT or 3′UTR △ARE plasmid and the mRNA produced (Created in BioRender. Ran, X. (2025) https://BioRender.com/v38l510). (Left) Relative level of CXCL10 genomic DNA (n = 4 biological replicates), (Middle) CXCL10 mRNA stability (n = 3 biological replicates), and (Right) western blot and quantification of CXCL10 protein levels (from one of two independent experiments) in SBC5 expressing CXCL10-ORF + 3′UTR WT or 3′UTR △ARE e Left: Schematic of SBC5-3′UTR WT or SBC5-3′UTR △ARE tumor engraftment in PBMC humanized NSG mice (Created in BioRender. Ran, X. (2025) https://BioRender.com/e93s595). Middle: Representative flow cytometry data for total T cells. Right: Cumulative data for total T cell (CD3⁺CD45⁺) in tumors (n = 6) on day 35. For (a–d), data are presented as mean ± SD. For (a) data are analyzed by two-way ANOVA. For (b–d), data are analyzed by two-tailed, unpaired t test with an exception for (e) by two-tailed, paired t test. (p: **** < 0.0001 < *** < 0.001 < ** < 0.01 < * < 0.05). All exact p values are listed in the source data. Source data are provided as a Source Data file.

Fig. 5. Olaparib combined with RT increases T cell tumor infiltration.

a Schematic of experimental design with KP1 allografts (Created in BioRender. Ran, X. (2025) https://BioRender.com/e93s595). b Tumor growth curve and Kaplan Meier analysis of KP1 allograft (n = 6 mice for each treatment) treated with olaparib (OLA), RT, or combination (COM) therapy. OLA or vehicle (VEH) was delivered by oral gavage 5 days per week and 8 Gy of RT was delivered on day 2. c Left: Representative of IHC staining for total T cells (CD3⁺) in KP1 tumors upon treatment (n = 3 mice for each treatment, scale bar 250 μm). Right: The quantification of CD3⁺ cells in KP1 tumors. d Schematic of gating strategy for immune profiling. e Left: Representative flow cytometry data for total T cells. Right: Cumulative data for total T cell (CD3⁺CD45⁺, n = 6). f–j Cumulative data for CD4⁺T cell (CD3⁺CD45⁺CD4⁺, n = 6), CD8⁺ T cell (CD3⁺CD45⁺CD8⁺, n = 6), memory CD8⁺ T cell (CD3⁺CD45⁺CD8⁺CD44^highCD62L^low, n = 6), exhausted T cell (CD3⁺CD45⁺TIM-3⁺PD-1⁺, n = 5) and exhausted CD8⁺ T cell (CD3⁺CD45⁺CD8⁺TIM-3⁺PD-1⁺, n = 5). k Cumulative data for PD-L1⁺ cells from KP1 tumors (n = 5). For (b), tumor growth curve data are presented as mean ± SD and statistically analyzed by two-way ANOVA. Kaplan-Meier data are analyzed by log-rank (Mantel-Cox) test. For (c), data are presented as floating bars (min to max) with line at median and statistically analyzed with one-way ANOVA. For (e–k), data are presented as a mean of replicates and statistically analyzed with one-way ANOVA. (p: ****<0.0001 < *** < 0.001 < ** < 0.01 < * < 0.05). All exact p values are listed in the source data. Source data are provided as a Source Data file.

Fig. 6. Incorporation of anti-PD-L1 antibody with PARPi and RT improves anti-tumor efficacy.

a Schematic of experimental design with KP1 allografts to olaparib, RT, and anti-PD-L1 antibody (ab) combination treatment (Created in BioRender. Ran, X. (2025) https://BioRender.com/e93s595). b Tumor growth curve and Kaplan Meier plot of KP1 allografts treated with VEH+IgG (n = 6), VEH+anti-PD-L1 ab (n = 6), COM (OLA + RT)+IgG (n = 6) and COM (OLA + RT)+anti-PD-L1 ab (n = 8). COM is olaparib 50 mg/kg day 1-5 per week plus RT 8 Gy, day 2. IgG or anti-PD-L1 ab 10 mg/kg was delivered by IP injection twice a week. c–e The cumulative data for immune profiling of KP1 tumor (n = 6) treated with VEH+IgG, VEH+a-PD-L1 ab, COM (OLA + RT)+IgG or COM (OLA + RT)+a-PD-L1 ab, including total T cell (CD3⁺CD45⁺), CD4⁺T cell (CD3⁺CD45⁺CD4⁺), and CD8⁺ T cell (CD3⁺CD45⁺CD8⁺). f Schematic of experimental design with KP1 allografts to TALA, RT, TALA + RT, and TALA + RT+anti-PD-L1 ab combination treatment (Created in BioRender. Ran, X. (2025) https://BioRender.com/e93s595). g Tumor growth curve and Kaplan Meier plot of KP1 allografts treated with TALA (n = 6), RT (n = 6), TALA + RT (n = 6) and TALA + RT+anti-PD-L1 ab (n = 6). TALA 0.2 mg/kg was delivered day 1–5 per week plus RT 8 Gy, day 2. Anti-PD-L1 ab 10 mg/kg was delivered by IP injection twice a week. h–j The cumulative data for immune profiling of KP1 tumor (n = 5) treated with TALA, RT, TALA + RT or TALA + RT+a-PD-L1 ab, including total T cell (CD3⁺CD45⁺), CD4⁺T cell (CD3⁺CD45⁺CD4⁺), and CD8⁺ T cell (CD3⁺CD45⁺CD8⁺). For (b) and (g), tumor growth curve data are as mean ± SD and statistically analyzed by two-way ANOVA. Kaplan-Meier data are analyzed by log-rank (Mantel-Cox) test. For (c–e) and (h–j), data are presented as mean of n = 6 tumors and statistically analyzed with one-way ANOVA. (ns, no significance; p: ****<0.0001 < *** < 0.001 < ** < 0.01 < * < 0.05). All exact p values are listed in the source data. Source data are provided as a Source Data file.

Fig. 7. Proposed model for the role of EIF4E2 in PARP inhibitor radiosensitization increasing T cell chemokine expression.

In comparison to RT treatment, PARPi olaparib combined with RT leads to increased DNA damage and subsequent chemokine CXCL10 mRNA increase via cGAS-STING pathway. The combination treatment triggers EIF4E2 downregulation to increase CXCL10 protein levels via stabilizing CXCL10 mRNA. The upregulated CXCL10 level recruits more T cells to the tumor sites and reshapes the tumor microenvironment (Created in BioRender. Ran, X. (2025) https://BioRender.com/k72k169).