CRISPR screen identifies autophagy inhibition (GNS561) as a PARP inhibitor (AZD5305) combination strategy in small cell lung cancer Authors: Tony Yu, Ranya Barayan, Lifang Song, Vidhyasagar Venkatasubramanian, Janvi Tamakuwala, Sree N Nair, Vivek Philip, Benjamin H Lok Published: 2026 ABSTRACT AbstractBackgroundSmall cell lung cancer (SCLC) is a deadly cancer with few treatment options and poor prognosis, creating a dire need for improving therapies. Poly (ADP-ribose) polymerase inhibitors (PARPi) have been tested as a treatment strategy, but patient response varies. We aimed to identify novel approaches to sensitize SCLC to PARPi through a genome-wide CRISPR dropout screen.MethodsGenome-wide CRISPR dropout screening was conducted in two SCLC cell lines using the PARPi, olaparib, as the selection pressure. Stable shRNA-mediated knockdown cell lines were validated by Western blotting and tested for olaparib sensitivity by assaying for cell viability. Synergy between PARPi and autophagy inhibition was tested by treating SCLC cell lines and analyzing cell viability using SynergyFinder+. The therapeutic strategy combining AZD5305 (PARPi) and GNS561 (novel autophagy inhibitor) was tested in cell line-derived xenograft mouse models.ResultsCRISPR screening identified the loss of mTOR negative regulators as a mechanism of PARPi sensitivity in SCLC, and knockdown of TSC1 and TSC2 sensitized SCLC cell lines to olaparib. Therapeutic strategies combining PARPi and autophagy inhibition demonstrated synergy in SCLC cell lines, and combination therapy with AZD5305 and GNS561 was effective in cell line-derived xenograft mouse models.ConclusionsAutophagy inhibition downstream of the mTOR pathway is a mechanism of PARPi sensitivity in SCLC. This suggests that a therapeutic combination of autophagy inhibition and PARPi is a promising treatment strategy in SCLC, paving the way for the adoption of novel treatments in this disease context.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s40170-026-00443-4.Keywords: Autophagy, Cancer therapy, Combination therapy, PARP inhibitors, Therapeutic resistance, Small cell lung cancer Introduction Small cell lung cancer (SCLC) is an aggressive, neuroendocrine carcinoma which accounts for about 15% of lung cancer cases [1]. Despite the rarity of SCLC, its high growth fraction, rapid doubling time, and propensity to metastasize make it deadly, with 5-year overall survival below 10% [2]. Approximately two thirds of patients present with extensive-stage SCLC (ES-SCLC) at diagnosis, defined as when the tumor has spread beyond a tolerable radiation field [1]. The rapid growth of this tumor makes it exquisitely sensitive to DNA damage, with response to first-line platinum (cisplatin or carboplatin) and etoposide chemotherapy reaching upwards of 60–80%. In addition, concurrent thoracic radiotherapy is an option in patients with limited-stage SCLC (LS-SCLC). However, relapse is nearly ubiquitous, leading to a dismal outlook with a median survival of under 10 months. Treatment of ES-SCLC patients with anti-PDL1 immunotherapy (atezolizumab and durvalumab) enhanced median survival by about 2 months, but there remains a significant unmet need in the treatment of SCLC patients [3, 4]. A landmark 2012 study identified poly (ADP-ribose) polymerase (PARP) inhibitors (PARPi) as promising agents against SCLC, as elevated PARP1 expression was transcriptomically and proteomically observed relative to other cancer types [5]. PARP1 is a critical protein in the base excision repair pathway, and inhibition of PARP1 activity results in the accumulation of single-strand DNA breaks. PARPi also traps PARP1 on DNA which, in combination with accumulated single-strand DNA damage, can cause double-strand DNA damage upon processes such as DNA replication [6]. As such, PARPi has become an established and effective therapy against cancers with deficiencies in homologous recombination (HR), a critical double-strand DNA damage repair pathway. However, SCLC tumors exhibit a range of responses independent of HR status or PARP1 expression. For instance, relapsed ES-SCLC patients treated with talazoparib in a phase I trial demonstrated variable responses [7]. Nevertheless, the mechanisms mediating PARPi sensitivity and combination therapies incorporating PARPi continue to be investigated, which may be leveraged to enhance therapeutic outcomes [8]. To identify targets that sensitize SCLC to PARPi, we conducted a CRISPR dropout screen in two SCLC cell lines using the PARPi, olaparib, as a selection pressure. Through these screens, we identified STK11 and components of the TSC (TSC1, TSC2) and GATOR1 (DEPDC5, NPRL2, NPRL3) complexes, which act as negative regulators of mTORC1 signaling and positive regulators of autophagy. mTORC1 is a major oncogene involved in pathways such as mRNA translation and nucleotide synthesis, and mTOR inhibitors have received attention across cancer types, including in SCLC [9]. However, mTORC1 also plays a role in suppressing autophagy, a cellular recycling pathway to eliminate damaged organelles and reuse metabolites, which tumor cells can leverage to overcome stress and promote therapeutic resistance [10–12]. Taken together, this suggests that autophagy inhibition can sensitize SCLC to PARPi. Here, we demonstrate that loss of the TSC complex sensitizes SCLC cell lines to PARPi and propose autophagy inhibition in combination with PARPi as a therapeutic strategy with efficacy against SCLC both in vitro and in vivo. Materials and methods Reagents Olaparib (HY-10162), AZD5305 (HY-132167), GNS561 (HY-137978A) were acquired from MedChemExpress. ULK-101 (40403-1) and chloroquine (C6628-25G) were acquired from Cedarlane Labs. Staurosporine (S1421) was acquired from Selleck Chemicals. Reagents were dissolved in dimethylsulfoxide (DMSO; BioShop Canada, DMS555.500) except for chloroquine which was dissolved in phosphate-buffered saline (PBS; Thermo Fisher Scientific, 20012050). Cell lines The SBC5 and NCI-H82 SCLC cell lines were generously provided by Dr. Kazuhiro Yasufuku (University Health Network, University of Toronto) and the HEK293T embryonic kidney cell line was generously provided by Dr. Robert Bristow (University of Manchester). The SHP77 (CRL-2195) and NCI-H1048 (CRL-5853) SCLC cell lines were acquired from the American Type Culture Collection (ATCC). SBC5, NCI-H82, and SHP77 were cultured in Roswell Park Memorial Institute (RPMI) 1640 Medium (Thermo Fisher Scientific, A1049101) supplemented with 10% fetal bovine serum (FBS; Thermo Fisher Scientific, 12483020). HEK293T was cultured in Dulbecco’s Modified Eagle Medium (DMEM; Thermo Fisher Scientific, 11995065) supplemented with 10% FBS. NCI-H1048 was cultured in DMEM/Nutrient Mixture F-12 Medium (Thermo Fisher Scientific, 11320033) supplemented with 5% FBS, 2 mM GlutaMAX (Thermo Fisher Scientific, 35050061), 10 nM hydrocortisone (Sigma-Aldrich, H0888-1G), 5 µg/mL insulin (Sigma-Aldrich, 91077C-100MG), 10 µg/mL transferrin (Sigma-Aldrich, T1147-5X100MG), 10 nM β-estradiol (Sigma-Aldrich, E2758-1G), and 30 nM sodium selenite (Sigma-Aldrich, S5261-10G). Cell starvation was conducted by culturing cells in Earle’s Balanced Salt Solution (EBSS; Thermo Fisher Scientific, 14155063). Cell lines were verified by short tandem repeat fingerprinting through The Centre for Applied Genomics (TCAG) and tested for Mycoplasma using the MycoAlert PLUS Mycoplasma Detection Kit (Lonza Bioscience, LT07-703). CRISPR dropout screen To obtain a multiplicity of infection (MOI) of about 0.3, SBC5-Cas9 and NCI-H82-Cas9 cell lines were transduced in 15 cm plates with varying concentrations of lentivirus medium and 8 µg/mL of polybrene (Sigma-Aldrich, 107689). Transduction was conducted for 24 h before medium was replaced with fresh culture medium containing puromycin (Corning, 61-385-RA) at predetermined concentrations (SBC-Cas9: 4 µg/mL for 3 days, NCI-H82-Cas9: 1 µg/mL for 2 days) to kill non-transduced cells. Puromycin untreated cells were used as a kill negative control. Following puromycin selection, cells were counted and compared to the puromycin untreated condition to determine the lentivirus medium concentration producing about 30% cell survival. For lentiviral production with the Toronto KnockOut v1 (TKOv1) library, HEK293T cells were seeded at 8 × 106 cells per 15 cm dish at 24 h before transfection. Cells were transfected with 8 µg of TKOv1 sgRNA library, 4.8 µg of psPAX2 plasmid (Addgene, 12260), 3.2 µg of pMD2.G plasmid (Addgene, 12259), and 32 µg of polyethylenimine (Sigma-Aldrich, 764965-1G) in 2.656 mL of Opti-Minimum Essential Medium (MEM) I Reduced Serum Medium (Thermo Fisher Scientific, 31985070). Transfection was conducted for 16–24 h before medium was replaced with fresh culture medium for lentivirus collection at 48 and 72 h post-transfection. Collected lentivirus medium was centrifuged at 1,600 rpm for 15 min at 4°C. Cells were seeded to achieve 300X coverage following transduction with a MOI of about 0.3. Cells were transduced with lentiviral medium, and divided into control and olaparib treatment arms. Olaparib treatment details can be found in Table S1. Cells were passaged regularly and pooled cell pellets were collected at the indicated timepoints (Fig. 1A). Genomic DNA was extracted using the QIAamp DNA Blood Maxi Kit (Qiagen, 51194) and next-generation DNA sequencing (NGS) was conducted. DrugZ was used to identify hits lost from the olaparib treatment arm which sensitize to olaparib, with a false discovery rate (FDR) < 0.05. Gene ontology (GO) analysis was conducted using the Protein Analysis THrough Evolutionary Relationships (PANTHER) system [13]. CRISPR dropout screen identifies loss of mTOR negative regulators as PARPi sensitizers in SCLC. (A) Schematic of CRISPR dropout screen in SBC5-Cas9 and NCI-H82-Cas9 using the TKOv1 CRISPR library. Cells were treated with 2.5–5 µM of olaparib to achieve > 99% kill starting from timepoint T4 for SBC5-Cas9 or T9 for NCI-H82-Cas9. Surviving cells were sequenced at timepoints T25 and T35 for SBC5-Cas9 or T28 and T39 for NCI-H82-Cas9. Created in Biorender. (B and C) Ranking of TSC1 and TSC2 for olaparib versus control by FDR adjusted p-value in the (B) SBC5-Cas9 screen or (C) NCI-H82-Cas9 screen. LacZ is included as a control. (D) Schematic of the functional roles of the gene panel. Signaling converges upon mTORC1 inhibition and autophagy activation. Created in Biorender. (E) Validation of TSC1 and TSC2 as olaparib sensitizers in SBC5 (left), SHP77 (middle), and HEK293T (right). Western blotting shows TSC1 levels in shTSC1A and shTSC1B compared to shScr control, or TSC2 levels in shTSC2 compared to shScr control. Cell viability with olaparib treatment for 4 days in these cell lines is shown with indicated IC50 values (n = 3, **: p < 0.01, ****: p < 0.0001). (F) Validation of autophagy downregulation in SBC5 (left) and SHP77 (right) cells with knockdown of TSC1 or TSC2. Cells were untreated or starved for up to 3 h in EBSS medium with or without GNS561 (GNS). Quantification of the LC3B-II/I ratio is shown compared to the untreated shScr control Small-hairpin RNA (shRNA)-mediated knockdown Plasmid cloning The shRNA constructs were chosen from The RNAi Consortium shRNA Library (Broad Institute) against TSC1 or designed against TSC2. shScramble (Sigma-Aldrich, SHC002) non-targeting shRNA was used to control for off-target effects. Sequence details can be found in Table S2. shRNA constructs were ligated to pLKO.1 plasmid (Addgene, 8453). Ligated plasmid was added to 50 µL of Stable Competent E. coli (New England Biolabs, C3040I) and incubated on ice for 30 min, then heat shocked at 42°C for 40 s before returning to ice for 1 min. Bacteria were added to 1 mL of Super Optimal broth with Catabolite repression (S.O.C.) Medium (Corning, 46-003-CR) and incubated at 37°C while shaking for 1 h, then pelleted by centrifugation at 200 rpm for 4 min and resuspended in 100 µL of S.O.C. Medium. Bacteria were plated on ampicillin agar plates and incubated at 37°C overnight before single colonies were isolated and incubated in lysogeny broth (LB) medium supplemented with ampicillin (Thermo Fisher Scientific, BP17605) at 37°C while shaking for 16 h. Plasmid was isolated using the Monarch Plasmid Miniprep Kit (New England Biolabs, T1010) and eluted in 30 µL of nuclease-free water pre-heated at 50°C. DNA sequences were validated by Sanger sequencing through TCAG. Lentivirus generation HEK293T cells were seeded at 1.5 × 106 cells per 10 cm dish at 24 h before transfection. Cells were transfected with 1 µg of shRNA plasmid construct, 0.67 µg of psPAX2 plasmid, 0.33 µg of pMD2.G plasmid, and 4 µg of polyethylenimine in 332 µL of Opti-MEM I Reduced Serum Medium. Transfection was conducted for 18 h before medium was replaced with fresh culture medium for lentivirus collection at 48 and 72 h post-transfection. Collected lentivirus medium was filtered through a 0.45 μm filter. Lentivirus transduction Cells were seeded at 3.0 × 105 cells/well in 6-well plates at 24 h before transduction. Cells were transduced with a 1:1 ratio of cell culture medium and lentivirus medium containing 8 µg/mL of polybrene. Transduction was conducted for 24 h before medium was replaced with fresh culture medium containing puromycin at predetermined concentrations (SBC5: 4 µg/mL for 3 days, NCI-H82: 1 µg/mL for 2 days, SHP77: 8 µg/mL for 3 days, HEK293T: 4 µg/mL for 3 days) to kill non-transduced cells. Following puromycin selection, medium was replaced with fresh culture medium. Immunoblotting Immunoblotting was conducted with rabbit anti-hamartin/TSC1 (4906S), rabbit anti-tuberin/TSC2 (4308S), rabbit anti-LC3B (3868S), and rabbit poly/mono-ADP ribose (E6F6A) primary antibodies from Cell Signaling Technology at 1:2,000 dilution, and normalized to mouse anti-β-actin (3700S) primary antibody from Cell Signaling Technology at 1:10,000 dilution. IRDye 800CW donkey anti-mouse IgG (926-32212) and IRDye 680RD donkey anti-rabbit IgG (926-68073) secondary antibodies were acquired from LI-COR Biosciences. Antibodies were diluted in tris-buffered saline (TBS) with 0.1% tween 20 (TBST; BioShop Canada, TWN510.500) containing 1% bovine serum albumin (BSA; BioShop Canada, ALB005.500). Cells were pelleted by centrifugation at 300 × g for 5 min and washed with PBS before lysis in RIPA lysis buffer containing Pierce Protease Inhibitor (Thermo Fisher Scientific, A32953) and Pierce Phosphatase Inhibitor (Thermo Fisher Scientific, A32957) on ice for 30 min. Protein lysates were isolated by centrifugation at 21,300 × g for 20 min at 4°C and concentrations were determined using the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific, 23227). Laemmli sample buffer was added to each sample and samples were denatured at 95°C for 10 min. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) was conducted using 12% polyacrylamide gels with Precision Plus Protein Dual Color Standards (Bio-Rad Laboratories, 1610374), then protein was transferred onto Immobilon-FL Polyvinylidene Fluoride (PVDF) Membranes (Thermo Fisher Scientific, IPFL00010). Membranes were blocked with TBST containing 5% BSA at room temperature for 1 h and incubated at 4°C overnight in primary antibody. Membranes were washed with TBST, incubated in secondary antibody at room temperature for 1 h, and washed with TBST before visualization using the Odyssey CLx system (LI-COR Biosciences). Immunofluorescence staining Immunofluorescence staining was conducted with rabbit anti-LC3B primary antibody and Alexa Fluor 488 donkey anti-rabbit IgG (Thermo Fisher Scientific, A21206) secondary antibody at 1:1,000 dilution. Antibodies were diluted in PBS containing 1% BSA. Cell lines were seeded at predetermined seeding densities in 6-well plates (0.8–1.2 × 106 cells/well) containing sterile coverslips (Biolyst Scientific, 63787-01) coated in poly-D lysine (Thermo Fisher Scientific, A3890401) at 24 h before treatment. Cells were washed with ice-cold PBS before fixation in 4% paraformaldehyde at room temperature for 15 min. Fixed cells were then washed with PBS before permeabilization in PBS containing 0.2% triton X-100 (Sigma-Aldrich, 9410-OP) at room temperature for 15 min. Permeabilized cells were blocked with PBS containing 3% BSA at room temperature for 1 h and incubated at 4°C overnight in primary antibody. Cells were washed with PBS with 0.1% tween 20 (PBST) and incubated in secondary antibody at room temperature for 1 h. Cells were then washed with PBST, incubated in DAPI solution at room temperature for 10 min, and washed with PBST before mounting on microscope slides (Fisher Scientific, 12-550-15) with ProLong Diamond Antifade Mountant (Thermo Fisher Scientific, P36965). Cells were visualized using the Axio Observer microscope (ZEISS) and analyzed using ImageJ [14]. Cell viability assay Cell lines were seeded at predetermined seeding densities in 96-well plates (1,000–1,500 cells/well) or 384-well plates (300–500 cells/well) before same day treatment with the corresponding agent. DMSO treatment was used as a kill negative control and 10 µM staurosporine treatment was used as a kill positive control. After 4–7 days of drug exposure, cell viability was determined using the CellTiter-Glo 2.0 Cell Viability Assay (Promega, G9243) and luminescence was evaluated using a microplate reader. Colony formation assays Clonogenic survival assay Cell lines were seeded at predetermined seeding densities in 6-well plates (DMSO: 125 cells/well, olaparib and chloroquine: 250 cells/well, both agents: 1,000 cells/well) before same day treatment with the corresponding agent. DMSO treatment was used as the negative control. After the formation of colonies with 50 or more cells, plates were fixed and stained using 2 mL of crystal violet solution composed of 0.5 g of crystal violet (Fisher Scientific, C581-100), 0.85 g of sodium chloride (BioShop Canada, SOD002), 5 mL of 10% buffered formalin (Fisher Scientific, 23-245684), 45 mL of water, and 50 mL of 100% ethanol for 10 min, before colonies were manually counted. Soft agar colony formation assay 1.5 mL of 0.6% agar (Invitrogen, 16500500) dissolved in a 1:1 mixture of water and 2X RPMI 1640 Medium (Wisent, 350-200-CL) supplemented with 20% FBS was used to coat each well of a 6-well plate and allowed to solidify at room temperature for 20 min. Cell lines were seeded at predetermined seeding densities in 6-well plates (DMSO: 1,000 cells/well, olaparib: 3,000 cells/well, chloroquine: 1,500 cells/well, both agents: 6,000 cells/well) in 0.4% low melting point agar (Sigma-Aldrich, A4018-25G) dissolved in a 1:1:2 mixture of water, PBS, and 2X RPMI 1640 Medium with 20% FBS containing any treatments or the DMSO negative control, and allowed to solidify at room temperature for 1 h. 200 µL of RPMI 1640 Medium with 10% FBS was added weekly to maintain the moisture of the agar. After the formation of colonies with 50 or more cells, excess medium was removed and plates were fixed and stained using 300 µL of 0.1X crystal violet solution for 3 h. Excess crystal violet solution was removed and agar was destained with water. Agar was air-dried and colonies were manually counted. Cell line-derived xenograft models Animal studies were conducted according to an approved protocol by the Animal Care Committee at the Princess Margaret Cancer Centre (Toronto, Canada) and the institutional Animal Resource Centre guidelines. NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ (NSG) mice (Cancer Stemcell Colony) were subcutaneously engrafted with 1.5 × 106 cells suspended in a 1:1 mixture of PBS and Matrigel (VWR, CACB354234) at 6–8 weeks of age. Mouse weight was measured weekly and tumor volume was monitored three times per week using length (x) and width (y) measurements from digital calipers based on the formula xy2/2. Tumor weight was approximated from tumor volume based on the density of water. Mice were randomized to treatment arms upon reaching a tumor volume of 50–100 mm3, and were treated 5 days per week with vehicle, AZD5305, GNS561, or combination therapy via oral gavage. Vehicle was a mixture of 5% DMSO, 5% tween 80 (Sigma-Aldrich, P1754-500ML), 40% polyethylene glycol 300 (Tokyo Chemical Industry, H0543), and 50% ddH2O. AZD5305 and GNS561 were dissolved at concentrations of 10 mg/kg and 40 mg/kg, respectively. Upon reaching the volumetric endpoint at a tumor volume of 1,000 mm3, mice were humanely euthanized with CO2. Statistical analysis Immunofluorescence puncta count data were compared using Dunn’s multiple comparisons test. Cell viability data were normalized to the kill negative and positive controls, and fitted to a four-parameter logistic model. The half-maximal inhibitory concentration (IC50) value of each cell line was compared to the control cell line using Dunnett’s multiple comparisons test. Synergy data were analyzed using SynergyFinder+ [15] based on the zero interaction potency (ZIP) model. The most synergistic area score (MSAS) was calculated as the average synergy score of the most synergistic three-by-three area [16]. For in vivo Kaplan-Meier analyses, events were defined as the time to reach the volumetric endpoint and analyzed using pairwise logrank tests with Bonferroni correction. Results CRISPR dropout screen identifies loss of mTOR negative regulators as PARPi sensitizers in SCLC To determine genes associated with PARPi sensitivity in SCLC, we conducted genome-wide CRISPR-Cas9 dropout screens in two SCLC cell lines (SBC5 and NCI-H82) treated with olaparib (Fig. 1A). Cells were treated with either DMSO control or olaparib and sequenced at the indicated timepoints. sgRNAs lost under olaparib selection relative to the control were defined as PARPi-sensitizing, after applying an FDR-adjusted p-value threshold of 0.05. Top hits included genes from the TSC and GATOR1 complexes which are involved in mTOR signaling and autophagy regulation, with TSC1, TSC2, STK11, NPRL2, and DEPDC5 ranking highly across multiple time points (Fig. 1B, C; S1A, B). Many of them rank higher than the BRCA2 positive control, which is not strongly PARPi sensitizing in these screens. As expected, the LacZ non-targeting control and especially the PARP1 negative control rank towards the bottom of the screen. HUWE1 is included as an additional negative control based on a recent study which identified this gene as a biomarker for PARPi sensitivity [17], and this gene also ranks near the bottom. GO analysis of top-ranked genes revealed an abundance of hits classified as regulators of mTOR signaling and autophagy (Fig. S1C, D), which converge on negative regulation of autophagy (Fig. 1D). To validate these findings, we employed shRNA mediated knockdown of TSC1 and TSC2 in the SBC5 and SHP77 cell lines, as well as HEK293T to investigate generalizability (Fig. 1E). Cell viability assays showed a significant reduction in the IC50 of shTSC1 (6.2 µM, p < 0.0001) and shTSC2 (4.3 µM, p < 0.0001) SBC5 cells treated with olaparib, compared to the non-targeting control (13.8 µM). Similarly, SHP77 shTSC1 cells had a reduced olaparib IC50 (10 µM, p < 0.0001; 20 µM, p < 0.01) compared to the control (37 µM). These data suggest that loss of mTOR negative regulators, TSC1 and TSC2, enhanced SCLC cell line sensitivity to olaparib treatment. Additionally, reductions in the IC50 of shTSC1 (1.2 µM, p < 0.0001) and shTSC2 (1.4 µM, p < 0.0001) HEK293T cells treated with olaparib compared to control cells (4.1 µM) suggests that this mechanism may extend beyond SCLC cells. We next investigated whether knockdown of TSC1 and TSC2 conferred alterations in autophagy in SBC5 and SHP77. These cells were treated with serum and amino acid starvation for 3 h in EBSS medium to induce active autophagy, then treated with 2 µM of the autophagy inhibitor, GNS561. Western blotting showed an accumulation of LC3B-II following GNS561 treatment (Fig. 1F), with a trend towards greater accumulation in TSC1 and TSC2 knockdown cell lines compared to the controls, both following starvation and GNS561 treatment. This suggests that autophagy is indeed blocked following knockdown, leading to the accumulation of LC3B-II on unprocessed autophagosomes. Autophagy inhibition synergizes with PARPi in SCLC Among its many roles, mTORC1 is known to suppress autophagy, a cellular recycling pathway which can promote therapeutic resistance [10–12]. Moreover, studies in cancer cells show upregulation of autophagy in response to PARPi to promote survival [11, 12], leading us to hypothesize that autophagy inhibition is the link between PARPi sensitization and the CRISPR screen hits we observed in SCLC cell lines. We therefore investigated whether autophagy inhibitors would synergize with PARPi. We first compared the inhibitory activity of both the established autophagy inhibitor, chloroquine (CQ), and the novel agent, GNS561. SCLC cells were treated with serum and amino acid starvation for up to 3 h in EBSS medium to induce active autophagy, then treated with autophagy inhibitor, either 5 µM of CQ or 2 µM of GNS561. Western blotting showed an accumulation of LC3B-II with both CQ and GNS561 treatment, while immunofluorescence staining showed a greater extent of LC3B puncta accumulation following GNS561 treatment compared to CQ (Fig. 2A), nonetheless demonstrating the capacity for both agents to inhibit autophagy. In SBC5 cells, a significant accumulation of LC3B puncta was observed by immunofluorescence staining with both CQ (1.01, p < 0.0001) and GNS561 (11.00, p < 0.0001) treatment compared to the 3-hour starvation control (0.08). In SHP77 cells however, a significant accumulation was observed only with GNS561 (3.31, p < 0.0001) treatment compared to the control (0.08), while there was no significant difference with CQ (0.11, p > 0.05) treatment. Indeed, monotherapy with GNS561 demonstrated greater potency across SCLC cell lines, with IC50 values ranging from 0.7 to 2.4 µM (Fig. 2C), compared to IC50 values of greater than 5 µM for CQ (Fig. S2A). Autophagy inhibition synergizes with PARPi in SCLC. (A) Cells were starved for up to 3 h in EBSS medium with or without CQ and GNS561 (GNS). CQ and GNS561 efficacy was validated by LC3B-II accumulation via Western blot (top) and LC3B puncta accumulation via immunofluorescence staining (bottom). Cell puncta were analyzed using ImageJ and representative images for each condition are shown (n = 3, ns: not significant, ****: p < 0.0001). (B) Validation of olaparib and AZD5305 (AZD) efficacy by reduction in PARylation following PARP inhibition. Cells were treated with PARPi for 72 h. (C and D) Cell viability showing sensitivity of SCLC cell lines to (C) GNS561 treatment for 7 days or (D) AZD5305 treatment for 4 days, with indicated IC50 values (n = 3). (E) Synergy between GNS561 and AZD5305 treatment for 7 days in SCLC cell lines (n = 3). A representative cell viability heatmap is shown for NCI-H82. MSAS values indicate the average synergy score in the most synergistic three-by-three regions (boxed). Synergy scores above 10 are considered synergistic In combination with GNS561, we chose AZD5305, a next-generation PARPi with enhanced PARP1 specificity [18]. Treatment of SCLC cells with 10 µM of AZD5305 produced a similar inhibition of PARylation to 10 µM of olaparib (Fig. 2B). SCLC cell lines displayed a range of sensitivities to AZD5305 as a monotherapy, from 5 nM in NCI-H1048 to > 25 µM in SHP77 (Fig. 2D), consistent with past observations that SCLC demonstrates variable responses to PARPi. Combination therapy with GNS561 and AZD5305 consistently demonstrated efficacy across SCLC cell lines with additive to mildly synergistic effects (Fig. 2E, S2E). MSAS ranged from 5.08 to 8.77 and maximum synergy scores ranged from 10.10 to 14.76 (Table 1). Table 1Synergy scores of AZD5305 in combination with GNS561 and ULK-101Cell LineCombinationOverall Synergy ScoreMost Synergistic Area ScoreMaximum Synergy ScoreNCI-H82AZD5305 + GNS5615.148.7712.38SHP773.328.1514.76NCI-H10482.845.0810.10SHP77AZD5305 + ULK-1014.839.2117.59NCI-H8215.9324.6333.53 Synergy scores of AZD5305 in combination with GNS561 and ULK-101 Interestingly, combination therapy with CQ and olaparib, an established autophagy inhibitor and PARPi, demonstrated a lack of synergy. Clonogenic survival assays of this combination in SBC5 and SHP77 cells demonstrated additivity rather than synergy (Fig. S2B). Similarly, synergy analysis of this combination showed a lack of interaction (Fig. S2C, D), with MSAS ranging from 0.12 to 0.81 (Table S3). Synergy analysis of other combinations of these agents generally displayed additivity as well (Fig. S3A, B), with MSAS ranging from 4.88 to 5.57 (Table S3), though olaparib and GNS561 was another effective combination with an MSAS reaching 15.39 in one replicate. Altogether, we demonstrated that novel inhibitors of autophagy (GNS561) and PARP1 (AZD5305) elicit synergy likely due to their greater potency compared to prior generations of these agents. We further investigated an ULK1 inhibitor, ULK-101, as a mechanistically different way to target autophagy. As a single agent, ULK-101 demonstrated varying efficacy across SCLC cell lines. In SHP77, the IC50 of 1.2 µM is comparable to GNS561, while in NCI-H82, the IC50 of > 10 µM is much higher than GNS561 (Fig. S3C). Combination therapy with ULK-101 and AZD5305 also demonstrated synergistic effects (Fig. 2F), with MSAS ranging from 9.21 to 24.63 and maximum synergy scores ranging from 17.59 to 33.53 (Table 1). This suggests that targeting autophagy in general—rather than just lysosomotropic agents—is an effective combination strategy with PARPi. Autophagy inhibition and PARPi combination is effective against SCLC xenograft models Given the synergy we observed in cell lines, we next tested whether this combination is effective in vivo. We engrafted NSG mice with either NCI-H82 or SHP77 cells, and randomized them to one of four treatment arms: vehicle, AZD5305, GNS561, and combination treatment. We monitored the tumor volume over the duration of the study, with a volumetric endpoint of 1,000 mm3. In NCI-H82 xenografts, the tumor growth curve indicates a trend towards reduced growth in the combination treatment arm with a 280 mg reduction in average tumor weight compared to the vehicle control (Fig. 3A), corresponding to a 50.3% tumor growth inhibition (Fig. S4A). This is more evident in SHP77 xenografts, where the combination treatment arm showed a more distinct separation in the tumor growth curve and a 250 mg reduction in tumor weight compared to the control (Fig. 3B), corresponding to a 69.3% tumor growth inhibition (Fig. S4B). Moreover, Kaplan-Meier analyses for both xenograft models demonstrated a significant improvement in survival for the combination treatment arm compared to the vehicle control (p < 0.05), while no significant survival benefit was observed for either of the single agent treatment arms alone (Fig. 3A, B). Autophagy inhibition and PARPi combination is effective in SCLC xenograft models. (A) Tumor growth curve (left), tumor weight (center), and Kaplan-Meier survival curve (right) of NCI-H82 xenograft mice treated with vehicle (VEH, n = 7), AZD5305 (AZD, n = 5), GNS561 (GNS, n = 8), or combination (COMBO, n = 8) therapy. Each treatment was delivered by oral gavage five days per week (ns: not significant, *: p < 0.05). (B) Tumor growth curve (left), tumor weight (center), and Kaplan-Meier survival curve (right) of SHP77 xenograft mice treated with vehicle (VEH, n = 6), AZD5305 (AZD, n = 9), GNS561 (GNS, n = 8), or combination (COMBO, n = 7) therapy. Each treatment was delivered by oral gavage five days per week (ns: not significant, *: p < 0.05). (C and D) Validation of AZD5305 efficacy by reduction in PARylation following PARP inhibition (left) or GNS561 efficacy by LC3B-II accumulation following autophagy inhibition (right) in (C) NCI-H82 and (D) SHP77 xenografts. Quantification of the LC3B-II/I ratio is shown compared to the vehicle control. COMBO 48 hr mice were treated with combination therapy for 48 hours upon reaching a tumor volume greater than 500 mm3 We monitored the weight of each animal throughout the study as a measure of therapeutic toxicity. No mice reached the humane endpoint of 20% reduction in body weight. In the NCI-H82 study, only one mouse in the combination treatment arm experienced weight loss greater than 15% of body weight at any point in the study, while one mouse from both the AZD5305 and combination treatment arms of the SHP77 study exceeded this threshold at any point in the study (Fig. S4). These mice recovered quickly after withholding treatment for 2–3 days, indicating limited toxicity. Target engagement of AZD5305 and GNS561 was probed by Western blot. In models treated with AZD5305 or combination therapy, inhibition of PARylation was observed (Fig. 3C, D). Interestingly, while accumulation of LC3B-II was observed in models treated with combination therapy for a shorter, 48 h duration, there was no accumulation of LC3B-II at volumetric endpoint in either GNS561 or combination therapy treated tumors (Fig. 3C, D). This suggests that autophagosomes may still be cleared following long-term autophagy inhibition, potentially highlighting the need to investigate the optimal timing of combination therapy. Ultimately, we demonstrated that autophagy is a targetable pathway that enhances PARPi efficacy against SCLC. Discussion SCLC remains one of the most challenging malignancies to treat. The poor prognosis associated with this patient population and the lack of effective therapies to overcome treatment resistance requires further investigation. Recently, PARPi has gained traction as a novel therapy for treating SCLC through the induction of DNA damage. However, response to PARPi therapy is variable, so methods to improve PARPi response rates are needed. We conducted CRISPR dropout screening in two SCLC cell lines (SBC5 and NCI-H82), identifying mTOR and autophagy regulatory genes—particularly TSC1 and TSC2—as significant top hits which sensitize SCLC cells to PARPi. GO analysis further classified many of these hits as negative regulators of mTOR and positive regulators of autophagy, and we corroborated the PARPi sensitivity phenotype using TSC1 and TSC2 knockdown cell lines. Combining autophagy inhibition and PARPi, we observed synergy between GNS561 and AZD5305, two novel therapeutics which are currently under clinical investigation. This was further demonstrated in two SCLC cell line-derived xenograft mouse models (NCI-H82 and SHP77), in which combination therapy provided a significant improvement in survival with limited toxicity. These data demonstrate that autophagy inhibition is a viable strategy to sensitize SCLC cells to PARPi therapy, especially with more potent and novel inhibitors. GNS561 is a novel and potent autophagy inhibitor which, similar to CQ, is lysosomotropic and inhibits autophagic flux [19]. It has also been implicated in inhibiting critical lysosomal enzymes like PPT1 and cathepsins, as well as directly activating apoptotic pathways through the disruption of lysosomal membrane integrity [19]. As GNS561 localizes most prominently to the liver in rat models, it has entered into clinical trials against cholangiocarcinoma and hepatocellular carcinoma [20]. It also localizes highly to the lung and stomach in these models, suggesting favorable kinetics in the context of SCLC [19]. AZD5305 is a next-generation PARPi which has superior PARP1 specificity compared to previous PARPi [18]. The reduced targeting of PARP2, a PARP1 homolog, has been demonstrated to reduce the toxicity of this PARPi in rat models, making it more favorable as a therapeutic agent. These factors make both agents attractive candidates for the treatment of SCLC and may explain the superior synergy observed compared to other combinations. Interestingly, long-term treatment of cell line-derived xenograft models with GNS561 resulted in reduction of LC3B-II accumulation back to baseline levels compared to short-term treatment. This may suggest that alternate mechanisms are used to clear autophagosomes, making the timing of combination therapy a critical factor in optimizing its efficacy. In addition to GNS561, we tested the ULK1 inhibitor, ULK-101, which inhibits autophagy induction [21]. In combination with AZD5305, ULK-101 also displayed synergy in SCLC cell lines. This indicates that synergy with PARPi is not limited to lysosomotropic agents such as CQ and GNS561, but rather occurs between through PARPi and autophagy inhibition in general. While also promising, the kinetics, efficacy, and toxicity of ULK-101 in animal models are unknown. Previous studies have identified the functional role of the TSC complex in restraining cell proliferation through mTORC1 suppression. It exists as a heterodimeric complex on the lysosome when active, with TSC1 acting as a scaffold to prevent the ubiquitin-mediated degradation of TSC2, allowing the latter to promote the GTPase activity of Rheb-GTP and inhibit mTORC1 [10]. Hence, the TSC complex was a surprising target for PARPi sensitization, given the many oncogenic roles of mTORC1 across cancer types. In SCLC as well, the PI3K/mTOR pathway is a known oncogenic pathway which can promote resistance to both chemotherapy and radiotherapy, and PI3K upregulation is a relatively common alteration [1, 9]. However, the mTOR pathway is also known to inhibit cell survival through autophagy, with mTOR activation sensitizing to radiotherapy and doxorubicin chemotherapy [22, 23], through mechanisms such as autophagy inhibition [24–26]. Moreover, autophagy upregulation has been described as a stress response which promotes survival following PARPi-induced DNA damage, acting as a resistance mechanism in cancers more commonly treated with PARPi such as breast and prostate cancers [11, 12]. Although autophagy inhibition has not been investigated in the context of PARPi in SCLC, autophagy has been identified as a resistance mechanism to chemotherapy in preclinical models [9]. Consistently, we observed that autophagy inhibition synergizes with PARPi both in vitro and in vivo, potentially through suppression of SCLC cancer cell survival mechanisms following PARPi treatment. While we propose a model in which autophagy allows SCLC cells to overcome PARPi-induced stress, we did not investigate the mechanism of action in relation to DNA damage response, which is a limitation of this study and worthy of future investigation. Interestingly, studies in other cancers have implicated autophagy in regulating double-strand DNA damage repair through either HR or non-homologous end joining (NHEJ), with active autophagy promoting HR-mediated repair and autophagy inhibition promoting error-prone NHEJ [12, 27]. Although the link between autophagy and DNA repair has not yet been fully elucidated, one study identified CHK1 activation downstream of autophagy [27]. Given that CHK1 is a critical cell cycle checkpoint protein which promotes DNA repair, this may explain the role of autophagy in regulating DNA damage response. Additional studies are needed to fully map this interaction and to uncover its importance in the context of SCLC. Furthermore, the proposed link between autophagy and DNA repair suggests that autophagy may be implicated in resistance to other DNA damaging therapies, such as radiotherapy. This supports the investigation of combinations like autophagy inhibition and radiotherapy, however past clinical trials studying this combination in SCLC were terminated due to poor patient accrual (NCT00969306, August 2009; NCT01575782, April 2012). In summary, we have identified a novel therapeutic combination, GNS561 and AZD5305, for SCLC therapy. We demonstrate that loss of autophagy-promoting genes, like TSC1 and TSC2, sensitizes SCLC to PARPi, and that autophagy inhibition is an effective therapeutic option for SCLC when combined with PARPi. Addition of autophagy inhibitors to PARPi treatment regimens may improve treatment efficacy against SCLC, paving the way for broader application of PARPi to combat this therapeutically challenging disease. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 2 Supplementary Material 3 Supplementary Material 4 Supplementary Material 5 Acknowledgements The authors thank Dr. Kazuhiro Yasufuku (University Health Network, University of Toronto) and Dr. Robert Bristow (University of Manchester) who generously provided cell lines used in this study. The authors thank the other members of the BHL laboratory at the Princess Margaret Cancer Centre for their support in the preparation of this work, and the Advance Optical Microscopy Facility department at UHN for providing assistance in immunofluorescence visualization work. Abbreviations American Type Culture Collection Bovine serum albumin Chloroquine Dulbecco’s modified Eagle medium Dimethylsulfoxide Earle’s balanced salt solution Extensive stage-small cell lung cancer Fetal bovine serum False discovery rate Gene ontology Homologous recombination Lysogeny broth Limited stage-small cell lung cancer Minimum essential medium Multiplicity of infection Most synergistic area score Next-generation DNA sequencing Non-homologous end joining NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ Protein analysis through evolutionary relationships Poly (ADP-ribose) polymerase Poly (ADP-ribose) polymerase inhibitor Phosphate-buffered saline Phosphate-buffered saline with 0.1% tween 20 Polyvinylidene fluoride Roswell Park Memorial Institute Small cell lung cancer Sodium dodecyl sulfate-polyacrylamide gel electrophoresis Super optimal broth with catabolite repression Tris-buffered saline Tris-buffered saline with 0.1% tween 20 The Centre for Applied Genomics Toronto knockout v1 Zero interaction potency Author contributions Conceptualization: TY, RB, VP, and BHL; methodology: TY, RB, LS; investigation: TY, RB, LS, VV, JT, SNN; validation: TY, RB, LS, VV, JT, SNN; data curation: TY, RB, LS, VV; data analysis: TY, RB; visualization: TY, RB; writing—original draft preparation: TY, RB; writing—review and editing: TY, RB, VV, JT, VP, BHL; supervision: VP, BHL. All authors read and approved the final manuscript. Funding This work was supported by the Conquer Cancer Foundation of the American Society of Clinical Oncology and International Association for the Study of Lung Cancer. Research at the BHL laboratory is funded by the Canada Foundation for Innovation; Canadian Institutes of Health Research; Clinical and Translational Science Center at Weill Cornell Medical Center, MSKCC (UL1TR00457); National Institute of Health/National Cancer Institute (U01CA253383); and Terry Fox Research Institute Program Project Grant (TFRI Project #1124). TY is supported by funding from the Department of Medical Biophysics, Canadian Institutes of Health Research, and The Strategic Training in Transdisciplinary Radiation Science for the 21st Century (STARS21). RB is supported by funding from the Institute of Medical Science. Data availability The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. Declarations Ethics approval and consent to participate Animal studies were conducted according to an approved protocol by the Animal Care Committee at the Princess Margaret Cancer Centre (Toronto, Canada) and the institutional Animal Resource Centre guidelines. Consent for publication Not applicable. Competing interests BHL reports grants from Pfizer; personal fees from Daiichi Sankyo; and grants, personal fees, and non-financial support from AstraZeneca outside the submitted work. The remaining authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Tony Yu and Ranya Barayan contributed equally to this work. References Associated Data Supplementary Materials Supplementary Material 1 Supplementary Material 2 Supplementary Material 3 Supplementary Material 4 Supplementary Material 5 Data Availability Statement The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.