PARP Inhibitor Activity Correlates with SLFN11 Expression and Demonstrates Synergy with Temozolomide in Small Cell Lung Cancer

Abstract

Purpose: PARP inhibitors (PARPi) are a novel class of small molecule therapeutics for small cell lung cancer (SCLC). Identification of predictors of response would advance our understanding, and guide clinical application, of this therapeutic strategy.

Experimental design: Efficacy of PARP inhibitors olaparib, rucaparib, and veliparib, as well as etoposide and cisplatin in SCLC cell lines, and gene expression correlates, was analyzed using public datasets. HRD genomic scar scores were calculated from Affymetrix SNP 6.0 arrays. In vitro talazoparib efficacy was measured by cell viability assays. For functional studies, CRISPR/Cas9 and shRNA were used for genomic editing and transcript knockdown, respectively. Protein levels were assessed by immunoblotting and immunohistochemistry (IHC). Quantitative synergy of talazoparib and temozolomide was determined in vitro In vivo efficacy of talazoparib, temozolomide, and the combination was assessed in patient-derived xenograft (PDX) models.

Results: We identified SLFN11, but not HRD genomic scars, as a consistent correlate of response to all three PARPi assessed, with loss of SLFN11 conferring resistance to PARPi. We confirmed these findings in vivo across multiple PDX and defined IHC staining for SLFN11 as a predictor of talazoparib response. As temozolomide has activity in SCLC, we investigated combination therapy with talazoparib and found marked synergy in vitro and efficacy in vivo, which did not solely depend on SLFN11 or MGMT status.

Conclusions: SLFN11 is a relevant predictive biomarker of sensitivity to PARP inhibitor monotherapy in SCLC and we identify combinatorial therapy with TMZ as a particularly promising therapeutic strategy that warrants further clinical investigation. Clin Cancer Res; 23(2); 523-35. ©2016 AACR.

Figure 1. PARP inhibitor sensitivity in SCLC correlates with SLFN11 gene expression and varies based on PARP trapping potency

(A) Scatterplot of area under the curve IC50 drug sensitivity data from the Genomics of Drug Sensitivity in Cancer (GDSC) available datasets of three different PARP inhibitors. R and p values from Spearman’s rank correlation tests are displayed. (B) Volcano plot of the mean −ln IC₅₀ (μM) of all three PARP inhibitors identified SLFN11 as a highly significant gene by log odds with the large gene expression variation according to drug sensitivity (Supplemental Table 2). (C) SCLC cell lines with greater SLFN11 gene expression are more sensitive to PARP inhibitors and standard cytotoxic chemotherapy. (D) SCLC cell lines are more sensitive to PARP inhibitors with greater PARP trapping potency. Veliparib, the least potent PARP trapper, and rucaparib exhibit greater IC₅₀ in SCLC as compared to all other tumor cell lines (*p = 0.03 for both drugs by Wilcoxon-Mann-Whitney test). Differences between the IC₅₀ of SCLC cell lines compared to all other tumor cell lines were not statistically significant by t-tests for olaparib.

Figure 2. SCLC cell line sensitivity to talazoparib depends on SLFN11

(A) Talazoparib sensitivity in SCLC cell lines was stereospecific and correlated with SLFN11 gene and protein expression. A near infrared Western blot against the labeled proteins is displayed. (B) Linear regression of SLFN11 protein levels normalized to actin as detected by near infrared Western blotting and talazoparib sensitivity demonstrated a strong correlative trend (p = 0.06) R and p values by Spearman’s rank correlation test.

Figure 3. Loss of SLFN11 confers resistance to PARP inhibition in SCLC

(A) Left: shRNA sequences against SLFN11 and Renila luciferase (REN) as a control were cloned into the LT3GEPIR doxycycline inducible vector. DMS114 cells stably expressing these constructs were subjected to vehicle or doxycycline at 1 μg/mL or 2 μg/mL concentrations for 48h. Western blotting by near infrared is displayed. Right: Cells were exposed to 1 μg/mL of doxycycline for 48h prior to plating and were then treated with a range of talazoparib doses 24h after plating. After 5d of drug exposure, a resazurin conversion assay was performed. Data represent the mean ± SD of 3 replicates. (B) Left: sgRNA sequences against SLFN11 were cloned into a lentiCRISPR v2 backbone and transduced into NCI-H526 cells to generate stable SLFN11-knockout(KO) lines. Western blot by chemoluminescence is shown. Right: A resazurin conversion assay of the labeled stable cell lines were performed after 5d of exposure to a range of talazoparib doses. Data represent the mean ± SD of 3 replicates.

Figure 4. SLFN11 protein expression correlates with talazoparib efficacy in patient-derived xenografts and SLFN11 mRNA is expressed bimodally in primary patient samples

(A) Representative images of immunohistochemical staining against SLFN11 for all tested PDXs are shown. The H-score and modified H-score (H_mod) for SLFN11 nuclear staining for each PDX model is displayed. Scale bar, 50 μm. (B) Percent change in tumor volume at end of study for each individual animal and displayed in order of SLFN11 H-score. End of study difference in tumor size between vehicle and treatment groups were significant for JHU-LX22, JHU-LX110, and SCRX-Lu149 (p = 0.0286, p = 0.0286, p = 0.0079, respectively) P-values by the Wilcoxon-Mann-Whitney test. (C) Percent tumor growth inhibition for each PDX model. Mean ± SD shown. The delta method was used to compute the variance used for SD calculations. (D) SLFN11 gene expression of primary SCLC samples plotted with publically available datasets from The Cancer Genome Atlas (TCGA) of other histologies are displayed here. The inset displays a bimodal distribution of SLFN11 gene expression (blue dashed line). Median ± SD shown.

Figure 5. Talazoparib is synergistic with temozolomide in vitro

(A) Left: Drug synergy matrix of talazoparib and TMZ in the NCI-H146 cell line is shown. A resazurin conversion assay was performed after 5d of drug treatment. Right: Excess over highest single agent (HSA) response is displayed showing the range of excess response in the 1 : 70 to 1 : 700,000 ratio range of talazoaprib : TMZ. (B) Near infrared Western blot against MGMT in the indicated cell lines. (C) Median effect plot of a MGMT low (NCI-H146) and high (NCI-H82) cell lines. A resazurin conversion assay was performed after 5d of drug treatment. Fa, fraction affected and Fu, fraction unaffected. (D) Combination indices of MGMT high and MGMT low labeled cell lines are displayed. Two-fold serial dilutions of the combination drug are displayed. The 0.002 : 20 (μM) dose of talazoparib : TMZ is highlighted in purple.

Figure 6. Talazoparib and temozolomide exhibit marked combinatorial efficacy in patient-derived xenografts

(A) Western blot against MGMT by near infrared imaging in PDX models. (B) Tumor volume growth curves of 7 PDX models treated with the labeled single drug or combination. n = 4–5 per arm. Mean tumor volume ± SD shown.