MYC drives platinum resistant SCLC that is overcome by the dual PI3K-HDAC inhibitor fimepinostat

Mouse tumor cell lines were generated following a protocol previously reported [14]. NCI-H146, NCI-H209 and NCI-H69 cells were purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA) and confirmed by short tandem repeat (STR) profiling. SCLC cell lines were cultured in Advanced RPMI supplemented with 1% FCS and 2 mM glutamax in humidified incubators at 37 °C and 5% CO₂.

Cells were plated in clear bottom 384 well plates (Greiner bio-one) in culture media using liquid handling robotics (Beckman Coulter, Biomek NXP). Cells were treated with 355 kinase inhibitors (Selleckchem Kinase library, L2000) at a final concentration of 100 nM. Plates were sealed with gas permeable tape (Roll-Seal, Sigma-Aldrich) to stop evaporation and incubated in a humidified culture incubator at 37 °C and 5% CO₂ for 7 days. Alamar blue viability dye (ThermoFisher) was added using liquid handling robotics and fluorescence measured using a ClarioStar plate reader (BMG Labtech). Background fluorescence was calculated as the mean of wells containing media alone and subtracted from well fluorescence. The mean fluorescence in vehicle control wells was considered 100% viable and fluorescence values for all drug wells expressed as a percentage viability. Z-scores were calculated for each well and a drug considered to be a hit if the Z score was lower than -3. Experiments were performed in duplicate, and a quality control cut off correlation between replicate plates of 0.75 was used. Cell number was optimized for each cell line based on Alamar blue fluorescence after a 7-day incubation.

Rb1^fl/fl;Trp53^fl/fl and Rb1^fl/fl;Trp53^fl/fl;Myc^LSL/LSL mice have been described previously [15, 16]. NSG mice were purchased from Australian BioResources (Garvan Institute of Medical Research, Sydney, Australia). All animals were housed in a specific-pathogen-free (SPF) vivarium at the Monash Medical Centre (Clayton, Victoria, Australia). All mouse studies were performed in accordance with the ethics approval from the Hudson Institute Animal Ethics Committee. For spontaneous SCLC models, 6–8-week-old mice were anaesthetized with an intraperitoneal injection of 1.25% (v/v) Avertin (Sigma-Aldrich, Missouri, USA) and 10⁶ pfu Ad5-CGRP-Cre (Viral Vector Core Facility, University of Iowa, IA, USA) was delivered intranasally. For xenograft studies, mice were subcutaneously injected with 10⁶ cells resuspended 50% (v/v) Matrigel (Corning, NY, USA). Tumor measurements were taken with digital calipers and tumor volume in mm³ calculated as (width² x length)/2. Animals were sacrificed upon reaching ethical endpoints that include, but are not limited to, breathing difficulties, ≥ 20% body weight loss or 800mm³ tumor volume. Drug treatments in flank models began when tumors reached 150-200mm³. Drug treatment in GEMMs was commenced when tumors were visible by computed tomography. Fimepinostat, 70 mg/kg was administered daily per. os. in a 30% captisol solution (CyDex pharmaceuticals). Mice in the platinum/etoposide cohort were injected with carboplatin (60 mg/Kg in PBS) and etoposide (10 mg/kg) intraperitoneally once a week for three weeks with one day between agents. On the day of carboplatin injection mice received an intraperitoneal injection of 1 mL of PBS to minimize kidney toxicity.

Lungs were removed and rinsed in PBS containing MgCl₂ (100 mg/L) and CaCl₂ (100 mg/L) and diced. Tissue was digested in collagenase IV (2.5 mg/mL) and DNAse I (100 μg/mL) in PBS at 37 °C for 90 min with rotation. Cells were resuspended in fetal calf serum, passed sequentially through 70 μm and 40 μm filters and plated in DMEM/F12 (Gibco). Cells were allowed to attach for 90 min in a humidified incubator with 5%CO₂ and media containing epithelial cells was removed and viable lung epithelial cell cells plated for drug treatments.

Mice were anaesthetized with isoflurane and Computed Tomography (CT) images were captured using a Siemens Inveon Small Animal PET/SPECT/CT scanner at 39.95 µm resolution, 80 kV, with 500µA current. Mice were fitted with a heart rate and respiration monitor (BioVet) and scans gated to maximum expiration. Images were processed with Fiji and Analyze 12.0 (AnalyzeDirect).

Cell lysates were resolved through acrylamide gels using SDS-PAGE and transferred to PVDF-FL membranes (Millipore). Membranes were blocked in Odyssey blocking buffer (Li-Cor) and incubated in specific primary antibodies including MYC (Abcam, 32,072), pAkt (Abcam, 38,449), Acetyl-Histone H3 (Abcam, 4729) and actin (Abcam, 3280). Membranes were incubated with IRDye fluorescent-conjugated secondary antibodies (Li-Cor), and protein expression was detected using Odyssey Infrared imaging system (Li-Cor).

Lungs were inflation-fixed in 10% neutral buffered formalin for 24 h prior to paraffin embedding. Histological assessment was performed on H&E-stained sections. For immunohistochemical analysis, sections were dewaxed, re-hydrated and subjected to microwave-based antigen retrieval with 20 min boiling in citrate buffer (10 mM citrate, 0.05% Tween-20, pH 6.0) under pressure. Sections were probed with primary antibodies against MYC (Abcam, Cambridge, UK) or PCNA (Dako). Blocking and secondary antibody staining were performed using Vectastain ABC Elite kits (Vector Laboratories, Burlingame, CA, USA).

The frequency of MYC amplification increases in relapsed SCLC [11]. To determine whether MYC expression is similarly increased in SCLC cell lines after platinum resistance is acquired two independent cell lines derived from a genetically engineered mouse model driven by the loss of Rb1 and Trp53 from lung neuroendocrine cells (RP mouse model) were repeatedly treated with an LD₅₀ dose of carboplatin and cells considered resistant if an increase in LD₅₀ of at least fivefold was achieved. The LD₅₀ of naïve and resistant clones (denoted “R”) was shown to exceed this threshold: B37 4 μg/mL, B37R 24 μg/mL, EN84 2 μg/mL, EN84R 23 μg/mL. Western blot analysis for MYC showed that increased platinum resistance was co-incident with an increase in MYC expression (Fig. 1B). To determine whether MYC expression is associated with platinum treatment in vivo, SCLC was initiated in the RP mouse model by intranasal inoculation of adenoviral Cre-recombinase under control of a neuroendocrine promoter (Ad5-CGRP-Cre) [15, 16]. These mice develop SCLC with a median survival of ~ 200 days. 80 days after disease initiation, mice were randomized into two groups, one received vehicle (PBS) and the other received three cycles of carboplatin (60 mg/kg) and etoposide (10 mg/kg). After three cycles treatment ceased and mice were monitored until they reached ethical endpoint (Fig. 2A). Platinum-etoposide chemotherapy significantly increased the overall survival of RP mice (vehicle 180.5 days, Pt/Etop 319 days, p < 0.0001 Log-Rank Mantel-Cox test), but mice did relapse and ultimately succumb to SCLC (Fig. 2B). Immunohistochemical analysis of MYC expression showed very few MYC positive cells in the primary lung tumors from vehicle control mice (mean of 2.28%) but following platinum and etoposide treatment a significant increase in the proportion of MYC positive cells in tumors was observed (mean of 16.14%, p < 0.0001, students t-test). Together these data confirm that increased MYC expression is observed following platinum-based chemotherapy and that MYC expression is coincident with acquired platinum resistance in vitro and in vivo. However, these data do not directly assess whether enforced MYC expression drives platinum resistance. To address this, mouse SCLC cell lines derived from RP mice were engineered to stably overexpress MYC and expression confirmed by western blot (Fig. 1C). Three independent matched pairs of cells were treated with titrating concentrations of platinum for 7 days and in each pair of cell lines the overexpression of MYC alone resulted in a significant increase in platinum resistance (Fig. 1D). These data confirm that MYC expression is not only coincident with but is also a driver of platinum resistant SCLC in vitro. To determine whether sustained MYC expression drives platinum resistance in vivo we took advantage of a mouse model of SCLC driven by the loss of Rb1 and Trp53 combined with sustained expression of MYC exclusively in neuroendocrince cells of the lung (RPM mice) [15]. These mice rapidly develop SCLC and have a median survival of ~ 100 days. Disease was initiated in RPM mice by intranasal inoculation of Ad5-CGRP-Cre and disease onset and progression monitored by computed tomography (CT). A baseline scan was performed 25 days after Cre delivery and mice were subsequently monitored every 10 days by CT (Fig. 3A). When tumor was observed mice were randomized into vehicle and platinum-etoposide treatment groups and treated as described for the RP mouse model. Strikingly, RPM mice were completely refractory to platinum-etoposide treatment. No reduction in tumor volume was observed in RPM mice treated with platinum and etoposide (Fig. 3B). A subtle, but insignificant increase in median survival was observed (vehicle: 23 days on treatment. Platinum and etoposide: 31 days on treatment). Together these data show that elevated MYC expression is not only coincident with platinum resistance but is a potent driver of resistance both in vitro and in vivo.

Identification of drugs that are effective against platinum resistant SCLC, or that will re-sensitize patients to platinum-based therapy are an urgent unmet clinical need. To identify novel drugs, a library of 355 kinase inhibitors (Table S1) was screened for their ability to kill platinum resistant mouse-derived SCLC cell lines at a final concentration of 100 nM. Viability z-scores were calculated in response to each agent and drugs that achieved a z-score less than -3 were prioritized. We identified drugs that have previously been shown to be effective against SCLC including PLK1, Aurora kinase A and CDK inhibitors validating the screening approach. However, the most effective drug was the novel dual PI3K and HDAC inhibitor fimepinostat (Fig. 4A). Importantly, fimepinostat was the most effective drug against all mouse derived cell lines irrespective of their platinum resistance or MYC expression status (Fig. 1B) suggesting that fimepinostat may be broadly effective in the treatment of SCLC. To determine whether fimepinostat efficiently kills human SCLC cell lines we determined the LD₅₀ dose of fimepinostat in four established cell lines representative of the recently described neuroendocrine subgroups of SCLC defined by ASCL1 or NEUROD1 expression [17] (SCLC-N: NCI-H82. SCLC-A: NCI-H209, NCI-H146 and NCI-H69) and two patient derived xenograft (PDX) lines [18]. Like the observation in mouse derived SCLC cell lines, all human SCLC cell lines were sensitive to fimepinostat with LD₅₀ doses between 5.93 and 23.45 nM. This was irrespective of prior chemotherapy treatment, MYC status or SCLC subgroup. The two PDX lines tested were exquisitely sensitive to fimepinostat with LD₅₀ values of 0.00025 and 0.039 nM (Fig. 4C). Drugs like PLK1 inhibitors have previously been shown to be effective against SCLC in cell lines but have ultimately failed in clinical testing due to toxicity. Fimepinostat has undergone clinical testing and has been granted orphan drug approval for the treatment of diffuse large B-cell lymphoma [19] and therefore has a known safety profile. However, to indicate whether fimepinostat is likely to have a useful therapeutic window in SCLC the LD₅₀ dose was determined for a matched pair of platinum naïve and resistant mouse derived SCLC cell lines and primary mouse lung epithelial cells. We observe significantly less toxicity in normal lung epithelium than observed in either SCLC line (Fig. 4D) suggesting that fimepinostat could be an effective treatment for SCLC whilst preserving normal lung epithelium.

MYC is often amplified via gene doubling, tandem duplication, or chromosomal translocation. Indeed, the MYC gene is amplified in 40% of human cancers, including breast [20], ovarian [21], prostate [22], hepatocellular [23], colon [24], and lung cancer [25]. MYC has largely been considered undruggable through direct approaches which has led to alternative and indirect approaches to treat MYC expressing tumors. One such approach is the inhibition of bromodomain and extra-terminal domain (BET) family of proteins. BET proteins bind acetylated histone lysine residues, leading to recruitment of P-TEFb via its BRD4 domain to sites of active transcription of genes such as MYC. The prototypical BRD2/4 inhibitor, JQ-1 has been shown to decrease MYC expression in other tumor indications [26, 27]. However, we found that treatment of two independent, MYC expressing, platinum resistant mouse derived SCLC cell lines had no impact on MYC expression even at super-physiological doses (Fig. S1).

Fimepinostat is a dual histone deacetylase (HDAC1/HDAC2/HDAC3/HDAC10) and phosphoinositide 3-kinase (PI3Kα/PI3Kβ/PI3Kδ) inhibitor. Both pathways are the focus of intense clinical investigation and have seen agents approved by the FDA for cancer treatment [28, 29]. The PI3K pathway and HDAC activity are implicated in MYC expression. PI3K signaling leads to activation of AKT and inhibition of GSK3β, which phosphorylates MYC at Thr-58 leading to degradation of MYC protein. Hence, inhibition of PI3K prevents inhibition of GSK3β thereby promoting MYC turnover [30‐32]. HDAC inhibitors lead to acetylation of MYC at lysine 323 and decreased MYC mRNA and protein expression [33]. Therefore, to determine whether fimepinostat reduced MYC expression in SCLC, two MYC expressing, platinum resistant mouse-derived SCLC cell lines were treated with titrating concentrations of fimepinostat for 24 h. This timepoint precedes the initiation of cell death observed following fimepinostat treatment. Reduction in MYC expression was observed in response to doses as low as 10 nm in B37R2 cells and 500 nm in EN84R2 cells (Fig. 5). This is consistent with studies showing fimepinostat decreases MYC expression in diffuse large B-cell lymphoma and NUT midline carcinoma [34, 35]. Fimepinostat reduced Akt phosphorylation and increased histone H3 acetylation confirming inhibition of PI3K and HDAC respectively (Fig. 5). Importantly, these doses are lower than the drug concentration achieved in patient serum following treatment [36]. Together these data show that fimepinostat is an efficient PI3K and HDAC inhibitor and that it reduces MYC expression. Moreover, we show that fimepinostat efficiently kills SCLC cell lines in vitro and that this is not restricted by MYC expression or platinum resistance status.

Our data show single agent efficacy of fimepinostat at concentrations that are achievable in patients. Therefore, to determine whether fimepinostat can reduce tumor volume in vivo, SCLC cell lines were injected into the flanks of host mice. Mouse derived SCLC cell lines B37R and EN84R were transplanted subcutaneously into the flanks of immune competent C57BL/6 mice. The human SCLC cell line NCI-H209 and PDX lines 102 and 109 were transplanted subcutaneously into the flanks of NOD.Cg-Prkdc^scid Il2rg^tm1Wjl/SzJ (NSG) mice. Once tumor volume reached 150-200mm³ mice received 70 mg/Kg fimepinostat or vehicle by daily oral gavage for 28 days or until tumors reached 800mm³ defined as an ethical endpoint. Fimepinostat treatment significantly reduced tumor growth in vivo in all models tested which is remarkable given the rapid growth observed in the vehicle control cohorts (Fig. 6A-E). The previously untreated NCI-H209 and two PDX lines had the most dramatic response to fimepinostat (Fig. 6C-E). Together these data show the single agent efficacy of fimepinostat against tumor cell lines that are representative of the genetic diversity observed in patients. However, these models are very aggressive, are grown in the flank rather than the lung and in the case of the human derived lines are transplanted into immunocompromised mice. Our data show that MYC is a potent driver of platinum resistance (Figs. 1, 2 and 3) and that fimepinostat efficiently reduces MYC expression. Therefore, we hypothesized that whilst fimepinostat efficacy is not dependent on MYC expression it will have the dual capacity to reduce MYC expression extending the duration of response to platinum in addition to its single agent efficacy. To address this, disease was initiated in the autochthonous RPM mouse model driven by deletion of Trp53, Rb1 and gain of MYC expression in pulmonary neuroendocrine cells and in the context of an intact immune system. Disease progression was monitored by CT imaging and once tumor was detected mice were randomized into four groups who received (i) vehicle, (ii) carboplatin and etoposide, (iii) fimepinostat or (iv) carboplatin, etoposide and fimepinostat (Fig. 7A). As observed previously no appreciable decrease in tumor size was detected in the platinum / etoposide cohort. In contrast, fimepinostat alone reduced tumor volume (Fig. 7B) and fimepinostat in combination with carboplatin and etoposide led to a very significant reduction in tumor mass (Fig. 7B). Histological and immunohistochemical analysis of lungs showed tumor cell death following each of the treatment arms which was most pronounced in the fimepinostat, carboplatin and etoposide cohort (Fig. 7C). This was accompanied by a significant decrease in cell proliferation based on the percentage of PCNA positive cells within the tumor mass (Fig. 7C, D). Interestingly, in mice treated with fimepinostat, carboplatin and etoposide we observed large regions of disrupted lung tissue architecture likely due to the destruction of tumor that previously occupied the lung (Fig. 7C). No overall survival benefit was observed following carboplatin and etoposide treatment alone. In contrast, fimepinostat monotherapy provides a significant increase in overall survival (p < 0.0001, log-rank Mantel-Cox test) and in this model is superior to standard of care chemotherapy. The combination of fimepinostat, carboplatin and etoposide produce a very significant increase in overall survival, taking the median survival from the 8 days on treatment observed in the vehicle group to 54.5 days in the combination therapy group (p < 0.0001, log-rank Mantel-Cox test) (Fig. 7E).

Together these data directly show that elevated MYC expression drives platinum resistant SCLC in vitro and in vivo. We identify fimepinostat as a drug that efficiently reduces MYC expression and has single agent efficacy against SCLC in the low nM range. Finally, our data show that the combination of fimepinostat with platinum and etoposide provides a significant survival benefit in an autochthonous mouse model of platinum resistant SCLC.