The FACT inhibitor CBL0137 Synergizes with Cisplatin in Small-Cell Lung Cancer by Increasing NOTCH1 Expression and Targeting Tumor-Initiating Cells

Lung cancer is the leading cause of cancer-related deaths in the world, with more than 1.3 million fatalities annually (1). Small-cell lung cancer (SCLC), which accounts for about 20% of all lung cancer, is an aggressive neuroendocrine tumor, characterized by rapid expansion and metastasis (2). Genomic characterization of SCLC tumors has not yet identified effective targets for therapy. Thus, standard chemotherapy remains the backbone of SCLC treatment and has changed little over the past three decades (3). The commonly used chemotherapeutic agents, including cisplatin, are highly cytotoxic and kill the majority of tumor cells initially, but the tumors recur rapidly. The cytotoxic activity of cisplatin is mediated by the formation of DNA-damaging adducts that activate several different signaling pathways, leading to apoptosis or cell-cycle arrest (3). However, intrinsic or acquired resistance to cisplatin remains a major limitation to curative therapies. Several mechanisms are believed to be responsible for resistance, including enhanced DNA damage repair (4), resulting in reduced apoptosis (5). For any potential new therapy to succeed, it is important to understand the molecular mechanisms that facilitate cell killing.

Tumor-initiating cells (TIC), important contributors to disease recurrence and metastasis, have been identified within most solid tumors and are associated with increased resistance to therapies, including cisplatin (6, 7). Therapies that kill non-TICs, but not TICs, may temporarily reduce the volume of a tumor, but relapse occurs because therapy-resistant TICs escape treatment and result in a drug-resistant recurrent tumor (7). TICs express specific markers, including CD133 and CD44, at much higher levels than the bulk tumor cell population (8). These markers are useful for the isolation and functional characterization of SCLC TICs and non-TICs as separate populations. Similarly to normal stem cell populations, TICs self-renew, differentiate, and express many of the same core transcription factors (9). SCLCs contain a much higher percentage of TICs than non-TICs, >65%–75%, compared with <15%–20% for non–small cell lung cancers (NSCLC; ref. 10). Improved clinical responses in SCLC may be achieved, therefore, by improved targeting of TICs, which are relatively insensitive to chemotherapy and lead to the growth of resistant tumors.

The experimental drug CBL0137 (11) is currently undergoing multicenter phase I clinical trials in metastatic or unresectable advanced solid neoplasms or refractory lymphomas (NCT01905228). CBL0137 (or the related drug quinacrine) has been shown to have potent anticancer activity in NSCLC (12), pancreatic cancer (13), breast cancer (14), and neuroblastoma (15). CBL0137 targets facilitates chromatin transcription (FACT), a histone chaperone that is expressed at high levels in tumors. Inhibition of FACT is toxic for most cancer cells because it is needed for the NFκB-induced expression of many genes, and activated NFκB is required for the survival of virtually all tumors (11). FACT is essential for the survival of glioblastoma (GBM) TICs (16), and also plays a critical role in cisplatin resistance by facilitating the repair of DNA damage (17). Interestingly, CBL0137 exhibits strong synergy with cisplatin in neuroblastoma by blocking the FACT-mediated repair of DNA damage (15). Therefore, targeting DNA repair is a potentially important strategy to enhance the effectiveness of cisplatin in SCLC.

The NOTCH signaling pathway regulates the self-renewal and survival of TICs (18). NOTCH1 is a transmembrane receptor that is activated upon ligand binding through a series of proteolytic cleavages. Once cleaved, the NOTCH1 intracellular domain translocates to the nucleus, where it binds to DNA and activates the transcription of target genes, including HES1 and HEY1 (19), whose increased expression in turn downregulates the expression of the transcription factor achaete-scute homolog-1 (ASCL1; ref. 20), which plays an important role in the proliferation and survival of SCLC cells (21). NOTCH1 can act as either a tumor suppressor or an oncogene. The tumorigenic or tumor-suppressive activities of NOTCH in different tumor types reflect its different roles in promoting or repressing the undifferentiated status of stem cells in the corresponding tissues (18). The oncogenic role of NOTCH has been identified in many cancers, including NSCLC (22), T-ALL (23), and GBM (24). In contrast, NOTCH1 signaling is suppressed in neuroendocrine tumor cells, including SCLC (25, 26), indicating that inducing its expression is an attractive strategy for treating these tumors.

The SP/KLF family of transcription factors consists of proteins with three highly conserved DNA-binding zinc finger domains, which recognize GC/CACCC boxes present in many GC-rich promoters (27). SP3 belongs to this family, which also includes SP1, 2, and 4, all of which bind to GC-rich NOTCH1 promoters (28). In human keratinocytes, KLF4 binds to the NOTCH1 promoter and, together with SP3, functions as a negative regulator of transcription, affecting recruitment of the Pol II preinitiation complex (28). Furthermore, knockdown of KLF4 and SP3 led to upregulation of NOTCH1 expression in HeLa cervical carcinoma and skin squamous carcinoma cells (SCC13) (28).

We have investigated a novel therapeutic strategy for SCLC by combining CBL0137 with cisplatin in patient-derived SCLC cells and xenografts. We tested the impact of CBL0137 on SCLC TICs in comparison with non-TICs, and the potential role of FACT in maintaining the stem cell phenotype of TICs. We also investigated the role of CBL0137 in increasing NOTCH1 expression, activating a core inhibitory signaling pathway in TICs. On the basis of previous findings and our current study, CBL0137 is a very potent anticancer drug. It inhibits FACT and NFκB activation in several different cancers (11), preferentially kills TICs, and targets NOTCH1 activation in SCLCs. CBL0137 synergizes with cisplatin in SCLCs, greatly increasing the sensitivity to this traditional chemotherapeutic agent.

The SCLC cell lines NCI-H82 (H82), NCI-H526 (H526), and NCI-H446 (H446) were obtained from ATCC, three years before being used in this study. The Rb/p53–mutant mouse SCLC KP1 cell line was a generous gift from Dr. Julien Sage (Stanford University, Stanford, CA), received a month before being used. The cells were maintained in culture for no longer than 2–3 months, and were routinely assayed for mycoplasma. The cells were grown in RPMI1640 medium supplemented with 5% (v/v) heat-inactivated FBS. For experimental purposes, the cells were cultured in SITA medium, consisting of RPMI1640 medium supplemented with 30 nmol/L selenium, 5 μg/mL insulin, 10 μg/mL transferrin, and 0.25% (w/v) BSA, EGF, and FGF (29). All cultures were incubated in 5% (v/v) CO₂ at 37°C. CBL0137 (lot # 10-106-88-30) was provided by Incuron, LLC. EGF and basic FGF were from PeproTech. Selenium, insulin, transferrin, and BSA were purchased from Sigma Chemicals. Antibodies against SOX2 (1:1,000) and OCT4 (1:500) were from Cell Signaling Technology. SSRP1 antibody (1:2,000) was from BioLegend, and β-actin antibody was from Sigma Aldrich. For immunofluorescence assays, antibody against CD133, anti-CD133/1 (AC133) conjugated with phycoerythrin (PE), and mouse IgG1-PE were from Miltenyi Biotec; and anti-CD44 conjugated with brilliant blue 515 (BB) was from BD Biosciences. The CyQUANT Direct Cell Proliferation Assay Kit was from Thermo Fisher Scientific. shRNAs to SP3 and scrambled shRNAs were obtained from Sigma Chemicals.

The H82 and H526 cells were authenticated. DNA extraction, short repeat profiling, and comparison with known cell line profiles from ATCC were performed by Genetica DNA Laboratories. The H446 cells from ATCC were not further authenticated.

Flow cytometry was performed using a FACSAria II Cell Sorter (BD Biosciences) to isolate TICs from H82, H526, or H446 cells. To obtain CD133^high and CD133^low cells, individual cells were labeled with PE-conjugated mAb against CD133. To isolate populations of CD44^high and CD44^low cells, the cells were labeled with BB-conjugated CD44 antibody. Dead cells were eliminated by DAPI staining (1 μg/mL, added immediately prior to sorting). The CD133 or CD44^high cells were cultured in SITA medium (29). CD133 or CD44^low cells were cultured in RPMI1640 with 5%–10% FBS. CD133 or CD44^high and CD133 or CD44^low cells in cell proliferation assays (CyQuant) were maintained in the SITA medium.

Cell survival was determined using the CyQUANT Fluorescent Assay (Thermo Fisher Scientific), according to the manufacturer's instructions. Briefly, the reagent was added directly to the culture medium in clear bottom black-wall 96-well plates. After a 2-hour incubation at 37°C, plates were centrifuged at 200 × g for 5 minutes, and fluorescence was read with a plate reader at excitation 480 nm and emission 535 nm. The combination index was assessed by CompuSyn software (ComboSyn, Inc; ref. 30).

For tumorsphere formation assays, TICs were FACS-sorted and plated at different dilutions in ultralow adherent 96-well plates, in supplemented SITA medium. Tumorspheres were counted after 2–3 weeks under a phase contrast microscope. Wells with a tumorsphere were counted as positive and the wells with none were counted as negative. The stem cell frequencies were calculated using an extreme limiting dilution algorithm (ELDA; ref. 31).

Quantitative real-time PCR was performed as described previously (32). cDNA was synthesized from total RNA, using a random hexamer and SuperScript III (Invitrogen). The expression levels of human SOX2, NANOG, OCT4, NOTCH1, HEY1, HES1, and ASCL1 mRNAs or control 18S rRNA were examined by using the EvaGreen qPCR master mix (Bullseye) in a LightCycler 480 (Roche). Gene-specific primers were: SOX2, forward CACACTGCCCCTCTCAC, and reverse TCCATGCTGTTTCTTACTCTCC; NANOG, forward GAAATACCTCAGCCTCCAGC, and reverse GCGTCACACCATTGCTATTC; OCT4, forward TCTCCCATGCATTCAAACTGAG, and reverse CCTTTGTGTTCCCAATTCCTTC; NOTCH1, forward CAATGTGGATGCCGCAGTTGTG, and reverse CAGCACCTTGGCGGTCTCGTA; HEY1, forward CTGGCTATGGACTATCGGAGT, and reverse GACCAGGCGAACGAGAAGC; HES1, forward AGGCTGGAGAGGCGGCTAAG, and reverse TGGAAGGTGACACTGCGTTGG; ASCL1, forward TCCCCCAACTACTCCAACGAC, and reverse CCCTCCCAACGCCACTG; and 18SrRNA, forward GCTTAATTTGACTCAACACGGGA, and reverse AGCTATCAATCTGTCAATCCTGTC.

Whole cell extracts were prepared by incubating cell pellets in lysis buffer containing 50 mmol/L Tris–HCl, pH 7.4, 150 mmol/L NaCl, 1 mmol/L EDTA, 1% NP-40, and a mixture of protease and phosphatase inhibitors (Roche). After incubation on ice for 20 minutes, cell debris was removed by centrifugation. Chromatin fractions were extracted using a subcellular protein fractionation kit (Thermo Fisher Scientific). Cell extracts containing equal quantities of proteins, determined by the Bradford method, were separated by SDS/PAGE [10% (v/v) acrylamide] and transferred to polyvinylidene difluoride membranes (Millipore). The membranes were incubated with primary antibody for 2 hours, followed by incubation with secondary antibody for 1 hour at room temperature, and developed by using enhanced chemiluminescence solution (Perkin-Elmer).

These assays were performed with the Agarose ChIP Kit (Thermo Fisher Scientific), according to the manufacturer's instructions. In brief, 2 × 10⁵ cells were fixed with 1% formaldehyde for 10 minutes at room temperature in supplemented SITA medium, followed by quenching with 125 mmol/L glycine for 5 minutes. The cells were lysed and the chromatin was fragmented by partial digestion with Micrococcal Nuclease. DNA/protein complexes were precipitated by overnight incubation with 4 μg anti-SP3 antibody (D-20, Santa Cruz Biotechnology), or 4 μg normal rabbit anti-IgG antibody (Santa Cruz Biotechnology), and then incubated with Protein A/G agarose beads for 2 hours. After reversal of protein-DNA cross-links, the DNA was purified and the abundance of the NOTCH1 promoter was analyzed by qPCR using NOTCH1-specific primers, forward 5′-AACGAGAAGTAGTCCCAGGC-3′ and reverse 5′-GCACTAGTGAGGCTCAGAGT-3′ as before (33). The PCRs for the NOTCH1 promoter sequence were performed using 2X Phusion Master Mix (New England Biolabs).

Whole cell or nuclear extracts were prepared with the Nuclear Extract Kit (Active Motif). The NOTCH1 probe containing 28 nucleotides (5′-CGGGGGTGGGGCCGGGCGGGGCGGGGCC-3′) and (5′-GGCCCCGCCCCGCCCGGCCCCACCCCCG-3′) was labeled with the Biotin 3′ End DNA Labeling Kit (Thermo Fisher Scientific). The DNA binding reaction and native polyacrylamide gel electrophoresis were performed following the instructions for Gelshift Chemiluminescent EMSA (Active Motif). The protein composition of complexes treated with CBL0137 (or untreated) was determined by performing a supershift assay, using antibodies to SP3 (D20, Santa Cruz Biotechnology). In different experiments 2 or 6 μg of SP3 antibody was used. The competition EMSA was done with a 100 fold molar excess of unlabeled probe.

Lentiviral transduction was performed as described earlier (32) using lentiviral plasmids encoding shRNAs against SP3 (TRCN 0000280421, # 1; TRCN 0000280370, # 2) or scrambled shRNA control.

All animal experiments were approved by the Cleveland Clinic Foundation Institutional Animal Care and Use Committee and conducted in accordance with the NIH Guide for the Care and Use of Laboratory Animals. Six-week-old NOD scid gamma (NSG) mice, obtained from Biological Research Unit, Cleveland Clinic, and B6.129S mice from Jackson Laboratories, were maintained in 12-hour light/12-hour dark cycles with free access to food and water. B6.129S mice were used as hosts in the immunocompetent syngeneic model with KP1 cells (34). For the patient-derived xenograft (PDX) study, we received a generous gift of PDX tumor (MSK-LX95), derived from a SCLC patient who previously received chemotherapy and relapsed (35), from Dr. Charles M. Rudin (Memorial Sloan Kettering Cancer Center, New York, NY). The PDX tumor fragments were passaged and maintained in NSG mice.

The mice were inoculated subcutaneously in the flanks with H82 cells (5 × 10⁵), or murine KP1 cells (5 × 10⁵), in medium containing 50% growth factor reduced, phenol red-free Matrigel (Corning). For the PDX experiment, 2-mm diameter fragments were inoculated subcutaneously in the flanks using a trocar. Tumors were measured by caliper 3 times weekly for the duration of the experiment. When tumors reached 3–4 mm diameter (∼20 mm³), the mice were divided into a control group that received both vehicles, and treatment groups that received either CBL0137 alone, cisplatin alone, or a combination of CBL0137 and cisplatin (n = 8 per group). Treatments were started on day 14 for H82 xenografts, KP1 syngeneic tumors, and for the PDX study, with vehicles alone, or CBL0137 (60 mg/kg, i.v.), or cisplatin (5 mg/kg, i.p.), or CBL0137 in combination with cisplatin. In the combination group, the mice received CBL0137 4 hours before treatment with cisplatin. Treatments were given once per week, either for 3 weeks, or until the tumors reached approximately 1,500–2,000 mm³, at which time mice were euthanized. Tumor volumes (V) were calculated using the volume for a prolate spheroid: v = 4/3 × π × a² × b, where a = minor radius, b = major radius.

Results from SCLC cell lines are represented by means ± SD. Data were analyzed using Student t test or by two-tailed ANOVA, using GraphPad Prism software. Differences between tumor volumes are represented by means ± SE, and compared pairwise using Student t test. P values of <0.05 are considered statistically significant. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant.

Inhibiting DNA repair is likely to enhance the effectiveness of cisplatin, and we hypothesized that combining cisplatin with CBL0137, to inhibit FACT, would be a logical approach. We evaluated the synergy between cisplatin and CBL0137 in the established patient-derived SCLC cell lines H82, H526, and H446, and also in the murine SCLC cell line, KP1, combining the two drugs at constant molar ratios of 1:1, using values above and below the IC₅₀s for each drug. By CyQUANT Direct assay, which measures proliferation as well as cytotoxicity (36), we show that the cells are more sensitive to the combination than to treatment with either drug alone (Fig. 1A–D, top). Synergism was determined by the Chou–Talalay method, which measures the median–drug effect and quantifies the combination indices of two drugs based on the growth inhibition curves of each drug alone, or of both in combination (30). The combination of cisplatin and CBL0137 was synergistic in all the cell lines, when combined at 1:1 ratios, as indicated by combination indexes substantially below 1.0 (Fig. 1A–D, bottom).

The in vivo effect of drug combination was then assessed in the H82 SCLC xenograft, murine SCLC syngeneic, and SCLC PDX models. In the H82 xenograft model, CBL0137 in combination with cisplatin significantly inhibited H82 tumor growth (P < 0.05), compared with CBL0137 alone, cisplatin alone, or vehicle control (Fig. 1E; Supplementary Fig. S1). Strikingly, there was no significant tumor growth until day 30 in mice treated with the combination of cisplatin and CBL0137, and the growth rate was much slower than with control or single-agent treatment. Mice treated with vehicle or single agents survived for 34–40 days, whereas mice receiving the combination survived for 51 days (Fig. 1E). In another experiment (Supplementary Fig. S1), tumor growth was monitored for 32 days of treatment. In immunocompetent mice, CBL0137 in combination with cisplatin substantially inhibited tumor growth, compared with CBL0137 alone (P < 0.05), cisplatin alone (P < 0.05), or vehicle control (P < 0.05; Fig. 1F). In 35 days, the tumors in vehicle or CBL0137 or cisplatin-treated mice reached a maximum size of approximately 1,000–2000 mm³, whereas the tumors did not grow further in mice treated with the drug combination, and the tumors were found to have regressed (Fig. 1F). Importantly, in the PDX study as well, CBL0137 in combination with cisplatin substantially inhibited tumor growth, compared with CBL0137 alone (P < 0.05), cisplatin alone (P < 0.05), or vehicle control (P < 0.05; Fig. 1G). There was no significant difference between control and single-agent treatment at the time of euthanasia (d50; P > 0.2; Fig. 1G). These results together indicate that CBL0137 in combination with cisplatin dramatically decreased SCLC tumor growth in SCLC cell lines as well as in xenograft and syngeneic tumors, revealing a novel potential combination therapy for this cancer.

TICs help to account for tumor recurrence after chemotherapy (37). Previously we showed that FACT is essential for the survival of GBM TICs, and that CBL0137 preferentially targets them (16). Because CBL0137 increases the sensitivity to cisplatin in SCLC, we thought that it might also target SCLC TICs. To test this idea, we assessed the effect of CBL0137 in TICs and non-TICs isolated from the cell lines H82, H526, and H446. The cell surface markers CD133 or CD44 are useful for the isolation and functional characterization of SCLC TICs and non-TICs as separate populations (7, 29). We isolated tumor-initiating populations by sorting the cells for CD133^high or CD133^low (for H82 and H526) and for CD44^high or CD44^low populations (for H446), where CD133 could not be detected. CD133 or CD44 high (top 8%–10%) and low (bottom 5%–10%) cells were sorted. The expression levels of CD133 and CD44 were confirmed by flow cytometry after growing the CD133^low and CD44^low cells in medium containing 10% serum and the CD133^high and CD44^high cells in serum-free SITA medium (ref. 29; Fig. 2A), using ultralow adherent tissue culture plates. CD133^high or CD44^high cells had higher levels of mRNAs and proteins corresponding to the stem cell transcription factors SOX2, NANOG, and OCT4, compared with CD133^low or CD44^low cells (Supplementary Fig. S2A–S2D), consistent with the tumor-initiating characteristics of these cell populations (6). Differentiation is another important characteristic of TICs (16, 29). We found that the expression levels of SOX2 were substantially decreased when the CD133^high or CD44^high cells were differentiated in serum-containing media (7, 16, 29) after around 3 weeks (Fig. 2B). It is to be noted that we detected the SOX2 band at approximately 50 kDa in H82 cells, instead of approximately 37–38 kDa, which could be due to posttranslational modifications of SOX2 (38). Consistent with our previous finding in GBM TICs (16), we observed a higher level of the FACT subunit SSRP1 in the TICs (Supplementary Fig. S3A), and treatment with CBL0137 depleted SSRP1 from the soluble nucleoplasmic fraction, leading to its accumulation in the insoluble chromatin fraction (Supplementary Fig. S3B), indicating chromatin trapping of FACT (11, 13) and confirming the inhibition of FACT by CBL0137 in these cells. Previously, we showed that CBL0137 decreases the expression levels of SOX2, NANOG, and OCT4 in GBM TICs (16). Consistently, here we show that CBL0137 treatment greatly decreased the expression level of SOX2, indicating that FACT is required for expression of this transcription factor, which is vital for TICs to self-renew (Supplementary Fig. S4A and S4B). The isolated, well-characterized TICs or non-TICs were then treated with CBL0137 or a DMSO control for 72 hours. As shown in Fig. 2C, although CBL0137 reduced both CD133^high or CD44^high and CD133^low or CD44^low cell populations in a dose-dependent manner, it was more potent against the CD133^high or CD44^high cells, emphasizing the preferential targeting of TICs by this drug.

Self-renewal is an important characteristic of tumor-initiating cells, defined by their ability to form tumorspheres from single cells in vitro (39). We assessed whether treatment of SCLC TICs with CBL0137 affected their ability to form tumorspheres in an extreme limiting dilution assay, which permits quantification of TIC frequencies (31). We exposed TICs to DMSO or CBL0137 for 24 hours, washed the drug out, and then plated the cells at densities of 5, 10, 20, or 50 cells per well in 96-well plates. Tumorsphere formation was evaluated after 14–21 days. Self-renewal of the TICs was markedly decreased by a single pretreatment with CBL0137 (Fig. 2D). These findings together suggest that CBL0137 exerts a potent inhibitory effect on TICs derived from SCLC cell lines, by targeting FACT and stem cell transcription factors that are required for TIC survival.

Overexpression of NOTCH1 in SCLC cells decreased cell proliferation and increased apoptosis (26, 40). Whole-genome sequencing of SCLC samples revealed inactivating mutations of NOTCH family genes in about 25% of the cases, suggesting a tumor-suppressive role of NOTCH (26). In this study, we observed that the expression of NOTCH1 mRNA was dramatically lower in SCLC TICs compared with non-TICs derived from H82 and H526 cells (Fig. 3A and B). Importantly, CBL0137 treatment significantly increased NOTCH1 mRNA expression in the TICs, in both cell lines (Fig. 3A and B). Next, we determined the effect of CBL0137 on the kinetics of NOTCH1 gene expression in the TICs, observing that treatment increased NOTCH1 mRNA expression after only 4 hours (Fig. 3C). Furthermore, treatment with CBL0137 significantly increased expression of the NOTCH1 target mRNAs HEY1 and HES1 in the TICs (Fig. 3D–G). Although CBL0137 also moderately increased the levels of mRNAs encoding HES1 and HEY1 in the non-TICs, the increases in the TICs were more dramatic (Fig. 3D–G).

NOTCH1 activation targets ASCL1, which is highly expressed in SCLC cells and regulates tumor-initiating capacity in these cells (26, 41). We observed a higher level of ASCL1 mRNA expression in TICs compared with non-TICs, and treatment with CBL0137 significantly reduced ASCL1 mRNA levels in these cells (Fig. 3H and I). We conclude that CBL0137 increases NOTCH1 expression and activates NOTCH1 signaling.

The above findings prompted us to investigate the mechanism by which CBL0137 activates NOTCH1 in SCLC TICs. We postulated that CBL0137 treatment might downregulate a negative regulator of NOTCH1, which in turn would increase NOTCH1 expression. It has been reported previously that the transcriptional repressors KLF4 and SP3 can modulate NOTCH1 expression by binding to its promoter (28). On the basis of this information, we determined the expression level of SP3 in TICs and non-TICs, finding it to be higher in the non-TICs (Fig. 4A). By performing chromatin immunoprecipitation assays, we show increased binding of SP3 to the endogenous NOTCH1 promoter in TICs, compared with non-TICs derived from H82 SCLC cells, and that this binding was eliminated in 2 hours by CBL0137, determined by PCR (Fig. 4B and C, top) as well as by qPCR (Fig. 4B and C, bottom). We further confirmed the effect of CBL0137 on the binding of SP3 to the NOTCH1 promoter by electrophoretic mobility shift assays (EMSA). The cells were treated with CBL0137 for different times, and EMSAs were performed with nuclear (Fig. 4D) or whole-cell lysates (Fig. 4E). In both experiments, CBL0137 impaired the binding of SP3 to the NOTCH1 promoter, although the levels of SP3 remained unchanged (Fig. 4F). Next, we explored whether CBL0137 treatment could prevent the binding of SP3 to the NOTCH1 promoter in vitro, using whole-cell lysates from H82 and H526 TICs for EMSAs with a NOTCH1 probe. In control lysates, SP3 bound to the NOTCH1 probe [Fig. 5A (lane 2) and B (lane 1]. However, when the lysates were treated in vitro with CBL0137 at different concentrations, SP3 did not bind to the probe. When the cell lysates were mixed with the probe first, and then treated with CBL0137 [Fig. 5A (lanes 4–6) and B (lanes 3–5)], SP3 binding was decreased dramatically, even at 200 nmol/L, the lowest concentration of CBL0137 [Fig. 5A (lane 4), and B (lane 5)], and was completely abolished at higher concentrations [1 or 5 μmol/L; Fig. 5A (lanes 5 and 6) and B (lanes 3 and 4)]. Controls with unlabeled competitor DNA [Fig. 5A (lane 1) and B (lane 6)] and with antibodies to SP3 [Fig. 5A (lane 3) and B (lane 2)] show that the indicated band is indeed SP3. In another experiment (Fig. 5C and D), the NOTCH1 probe was incubated with CBL0137 first, and then the whole-cell lysates of H82 or H526 TICs were added to the NOTCH1-CBL0137 mixtures. Results similar to those in Fig. 5A and B were obtained, where SP3 binding was decreased or abolished with CBL0137 treatment (Fig. 5C and D, lanes 4–6), compared with the control (Fig. 5C and D, lane 3). We conclude that CBL0137 treatment prevents the in vitro binding of SP3 to the NOTCH1 promoter in TICs and reverses binding when added to an SP3–probe complex. These results indicate that CBL0137 treatment abrogates the binding of a negative regulator to the NOTCH1 promoter, facilitating NOTCH1 transcription, which in turn facilitates SCLC cell death (26, 40, 42). To test whether downregulating SP3 could increase NOTCH1 expression, we transduced H82 TICs with two different shRNAs to SP3 or scrambled shRNA and determined the expression levels of NOTCH1 mRNA. Depletion of SP3 significantly increased the level of NOTCH1 in these cells, indicating that SP3 is a negative regulator of NOTCH1 expression (Fig. 5E and F).

Resistance to chemotherapy is a major obstacle to successful treatment of SCLC. There are currently no targeted approaches to treat this disease that are similar to those used successfully against NSCLC, and there have been no significant advances in the last 30 years (43). The standard-of-care platinum-based drugs have the ability to kill the bulk tumor, but fail to eliminate the TICs, resulting in tumor recurrence (7, 44). The most immediate therapeutic improvements against this cancer will depend on our ability to prevent or delay the emergence of chemoresistance. CBL0137 synergizes with the DNA-damaging drugs cisplatin and etoposide in neuroblastoma (15). In addition, CBL0137 increases the sensitivity to cisplatin in neuroblastoma because FACT, the target of CBL0137, is required for DNA repair (15). Therefore, targeting DNA repair is a promising stratagem to enhance cisplatin effectiveness, and provides a strong rationale for combining cisplatin with CBL0137, to inhibit FACT and achieve synergistic lethality.

The combination of CBL0137 with cisplatin is determined to be synergistic at their respective IC₅₀ ratios (30) in patient-derived SCLC cell lines as well as in a murine SCLC cell line. We demonstrate extensively that these therapeutic strategies are also effective in vivo, in experiments with SCLC xenografts, and a PDX model in immunocompromised mice, and syngeneic SCLC tumors in immunocompetent mice. Importantly, the PDX specimen we used is derived from the tumor of a patient who relapsed after initial response to the combination of cisplatin and etoposide, the standard-of-care therapy. Our data reveal that CBL0137 helps to overcome resistance to cisplatin. CBL0137 is currently in the final stages of multicenter phase I clinical trials in advanced or metastatic solid tumors and lymphomas (NCT01905228), and it has not yet exhibited dose-limiting toxicity. Therefore, using CBL0137 in combination with cisplatin in SCLC is a novel therapeutic strategy for this cancer that can be employed soon.

By using the cell surface markers CD133 or CD44, TICs have been identified in a variety of human cancers, including lung cancers (7, 29). TICs show elevated expression levels of genes encoding transcription factors that are associated with stemness, including SOX2, OCT4, and NANOG (16). We demonstrate that CD133^high or CD44^high SCLC cells have increased levels of SOX2, NANOG, and OCT4, compared with CD133^low or CD44^low cells, consistent with the tumor-initiating characteristics of these cells. As in previous reports (16, 29), we also show that culturing CD133^high or CD44^high SCLC cells in serum-containing medium leads to their differentiation, accompanied by the loss of stem cell markers.

Eradication of TICs, a major challenge for cancer therapy, can be achieved by using inhibitors that target TIC-specific pathways. We show that the FACT inhibitor CBL0137 is very potent as a single agent toward SCLC cells, and preferentially targets TICs, consistent with our previous findings in GBM (16). FACT is required for the expression of stem cell transcription factors that are vital for TICs to self-renew (16). Treatment with CBL0137 in SCLC TICs inhibits the function of FACT by depleting soluble SSRP1 and trapping it on chromatin, decreases the expression of self-renewal genes, and dramatically reduces the self-renewal potential of the TICs. Our findings indicate that CBL0137, by targeting FACT, may have greater efficacy against tumors high in TIC content, as revealed by SOX2 expression. TICs are critical determinants of drug resistance and relapse in SCLC (45), and therefore, further studies are warranted to correlate stem cell marker expression in this disease with chemotherapeutic-responses and survival.

We discovered an additional potential therapeutic activity of CBL0137, as an activator of NOTCH1 expression in SCLC. This drug preferentially increases the expression of NOTCH1 mRNA in SCLC TICs compared with non-TICs, and also increases the expression of the NOTCH1 targets HEY1 and HES1, while decreasing the expression ASCL1, which is inhibited by NOTCH signaling (40). However, a moderate increase in the HEY1 and HES1 in the non-TICs suggests that NOTCH-independent signaling pathways might be involved in the non-TICs. The NOTCH1 signaling pathway is silenced in many neuroendocrine malignancies, including SCLC (40). Activation of NOTCH1 signaling inhibited the growth of SCLC cells (25), and reduced the number of tumors and extended the survival of mice in a preclinical SCLC mouse model (26). We show that the expression of NOTCH1, HEY1, and HES1 mRNAs is very low in SCLC TICs with increased levels of ASCL1 mRNA, compared with non-TICs, suggesting that the tumor-suppressive role of NOTCH1 is not manifest in SCLC TICs. Our finding that treatment of SCLC TICs with CBL0137 increases NOTCH1 expression reveals a novel potential therapeutic role of CBL0137 in SCLC. We conclude that CBL0137 preferentially impairs SCLC TICs not only by targeting FACT, but also by activating NOTCH1. Recently, it has been reported that the loss of even one allele of Ascl1 dramatically decreases mouse SCLC tumor growth (46), indicating the desirability to focus on therapeutic targeting of the ASCL1 pathway in this disease. Our results showing decreased ASCL1 expression in SCLC TICs upon CBL0137 treatment indicates the utility of further work, to reveal the therapeutic indications of this drug in targeting ASCL1. Another very recent study shows that, in SCLC, NOTCH signaling can be both tumor suppressive and protumorigenic (47). However, our findings together reveal that CBL0137 preferentially kills SCLC TICs by activating the tumor-suppressive role of NOTCH1 by increasing the expression of NOTCH1 and NOTCH1 target genes, and by decreasing ASCL1 expression, in SCLC TICs.

CBL0137 treatment impairs both the endogenous and in vitro binding of SP3 to the NOTCH1 promoter, revealing the mechanism of CBL0137-induced NOTCH1 activation in SCLC TICs. SP3 can act as an activator (48) or a repressor of transcription (49). It is upregulated in cancer cells, for example, cervical carcinomas and keratinocyte-derived squamous cell carcinomas, where NOTCH1 expression is downmodulated (28). We observe that acute exposure to CBL0137 not only prevents the binding of SP3 to the NOTCH1 promoter, but also reverses the binding in SCLC TICs. Downregulation of SP3 increased NOTCH1 expression in the TICs, confirming the role of SP3 as a negative regulator of NOTCH1 (28).

Genomic profiling of SCLC is still in its infancy, delaying the development of molecularly targeted therapies. Our approach to use combination therapy with CBL0137 and cisplatin is an important means to circumvent the development of resistance to standard therapy (Fig. 6). Another promising approach is the use of CBL0137 as a TIC-targeting therapy to prevent the emergence of tumor recurrence by eradicating TICs (Fig. 6). Our novel finding of the role of CBL0137 in activating NOTCH1 also has therapeutic implications, opening an important new avenue to further explore in the clinical use of this drug.

No potential conflicts of interest were disclosed.

Conception and design: S. De, A. Dowlati, G.R. Stark

Development of methodology: S. De, D.J. Lindner, A. Dowlati

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. De, D.J. Lindner, C. Coleman, G. Wildey, A. Dowlati

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. De, D.J. Lindner, A. Dowlati, G.R. Stark

Writing, review, and/or revision of the manuscript: S. De, D.J. Lindner, C. Coleman, G. Wildey, A. Dowlati, G.R. Stark

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S. De, C. Coleman

Study supervision: S. De, G.R. Stark

This research was supported by a VeloSano Pilot grant from Cleveland Clinic (to G.R. Stark). We are extremely thankful to Drs. Andrei Gudkov and Andrei A. Purmal for providing CBL0137. We thank the Cleveland Clinic Flow Cytometry Core for excellent technical support. We would also like to thank Ms. Yvonne Parker for technical support and Dr. Katerina V. Gurova for helpful comments.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.