Proteomic profiling identifies dysregulated pathways in small cell lung cancer and novel therapeutic targets including PARP1

Abstract

Small cell lung cancer (SCLC) is an aggressive malignancy distinct from non-small cell lung cancer (NSCLC) in its metastatic potential and treatment response. Using an integrative proteomic and transcriptomic analysis, we investigated molecular differences contributing to the distinct clinical behavior of SCLCs and NSCLCs. SCLCs showed lower levels of several receptor tyrosine kinases and decreased activation of phosphoinositide 3-kinase (PI3K) and Ras/mitogen-activated protein (MAP)/extracellular signal-regulated kinase (ERK) kinase (MEK) pathways but significantly increased levels of E2F1-regulated factors including enhancer of zeste homolog 2 (EZH2), thymidylate synthase, apoptosis mediators, and DNA repair proteins. In addition, PARP1, a DNA repair protein and E2F1 co-activator, was highly expressed at the mRNA and protein levels in SCLCs. SCLC growth was inhibited by PARP1 and EZH2 knockdown. Furthermore, SCLC was significantly more sensitive to PARP inhibitors than were NSCLCs, and PARP inhibition downregulated key components of the DNA repair machinery and enhanced the efficacy of chemotherapy.

Significance: SCLC is a highly lethal cancer with a 5-year survival rate of less than 10%. To date, no molecularly targeted agents have prolonged survival in patients with SCLCs. As a step toward identifying new targets, we systematically profiled SCLCs with a focus on therapeutically relevant signaling pathways. Our data reveal fundamental differences in the patterns of pathway activation in SCLCs and NSCLCs and identify several potential therapeutic targets for SCLCs, including PARP1 and EZH2. On the basis of these results, clinical studies evaluating PARP and EZH2 inhibition, together with chemotherapy or other agents, warrant further investigation.

Figure 1. Key differences in protein expression and pathway activation between SCLC and NSCLC

(A) For each cell line, protein lysates were collected and analyzed by RPPA after growth in 10% serum, 0% serum, and serum stimulated conditions (0% serum for 24h, then 10% serum for 30min prior to harvest) to account for possible effects of medium on protein expression. Unsupervised hierarchical clustering separated SCLC cells (pink) from NSCLC cells (green) on the basis of their distinct expression of 193 total and phospho-proteins. Clustering was independent of growth conditions, with lysates from the same cell line (but different media conditions) clustering together as nearest neighbors. NSCLC cell lines with neuroendocrine features—H1155 (large cell (LC)) and H1770 (neuroendocrine (NE)) (blue)—clustered with SCLC cell lines based on similar protein expression patterns. (B) First principal component analysis using all RPPA proteins also separated NSCLC cell lines from SCLC cell lines. (C) Protein markers most differentially expressed between SCLC and NSCLC based on a FDR <1% and ≥1.5 -fold difference in mean expression. Cell lines are clustered by hierarchical clustering and results from all media conditions are shown. NSCLC cell lines with neuroendocrine features (LC/NE, blue) clustered with SCLC (orange) based on similar protein expression. (D) Proteins expressed at higher levels in SCLC or NSCLC are mapped to their respective signaling pathways.

Figure 2. mRNA expression of PARP1 in SCLC cell lines and solid tumors

(A) mRNA expression in SCLC (green) and NSCLC cell lines (pink) for genes corresponding to the total proteins dysregulated in SCLC. (B) Potentially druggable targets identified by RPPA that were also more highly expressed at the mRNA level in SCLC included PARP1, EZH2,, BCL2, PRKDC (DNA PKcs), and PCNA. (C) PARP1 mRNA expression was higher in SCLC cell lines than in other solid tumor cell lines. (D) mRNA expression of potential drug targets were higher in SCLC tumors than in NSCLC tumors or normal lung.

Figure 3. PARP1 protein expression in lung tumors

(A) Total PARP expression was higher in neuroendocrine tumors (SCLC, LCNEC, atypical carcinoid, and typical carcinoid) than in lung squamous cell carcinoma and adenocarcinoma. ^‡p=0.0002 (SCLC versus squamous tumors), ^†p=0.001 (LCNEC versus squamous), ^*p<0.0001 (SCLC or LCNEC compared to adenocarcinoma). (B) Representative PARP1 IHC staining for each tumor type.

Figure 4. SCLC and LCNEC are sensitive to PARP inhibition in vitro

(A) Cell were treated with 0.1, 1, and 10 μM AZD2281 for 24 hrs, cell extracts collected, and poly ADP ribose (PAR) levels evaluated by ELISA to assess PARP1 activity. (B) IC₅₀ values for lung cancer cell lines treated with AZD2281 for 5 days. (C) Lung and breast cancer cells were treated with increasing concentrations of AZD2281 or AG014699 for 14 days. (D) PARP1 and EZH2 knockdown by siRNA effect SCLC proliferation.

Figure 5. RAD51 foci formation (A) and modulation of proteins levels after PARP inhibitor treatment (B)

(A) Protein is localized at DNA DSBs region in response to stalled or collapsed DNA replication forks in SCLC (H69 & H82) but not in NSCLC cell (A549). Kinetics of RAD51 focus formation in NSCLC A549, SCLC H69, and SCLC H82. The percentage of cells with more than 5 nuclear foci was calculated. In each experiment, 100 nuclei were counted per data point. Error bars indicate standard error compared to unirradiated samples (*p<0.05). (B-C) Protein lysate was collected from three SCLC cell lines (H69, H82, H841) in duplicate at multiple timepoints (0-14d) after treatment with the PARP inhibitors AZD2281 and AGO14699. A time-dependent decrease was observed in multiple DNA repair proteins (B) and in other E2F1 targets such as thymidylate synthase (TS) and EZH2 (C). Note that TS follows a similar pattern to the other DNA repair proteins, while EZH2 is suppressed at 24h but recovers to baseline levels by 14d.