Targeting sphingosine kinase 1/2 by a novel dual inhibitor SKI-349 suppresses non-small cell lung cancer cell growth

Sphingosine kinase 1 (SphK1) and sphingosine kinase (SphK2) are both important therapeutic targets of non-small cell lung cancer (NSCLC). SKI-349 is a novel, highly efficient and small molecular SphK1/2 dual inhibitor. Here in primary human NSCLC cells and immortalized cell lines, SKI-349 potently inhibited cell proliferation, cell cycle progression, migration and viability. The dual inhibitor induced mitochondrial depolarization and apoptosis activation in NSCLC cells, but it was non-cytotoxic to human lung epithelial cells. SKI-349 inhibited SphK activity and induced ceramide accumulation in primary NSCLC cells, without affecting SphK1/2 expression. SKI-349-induced NSCLC cell death was attenuated by sphingosine-1-phosphate and by the SphK activator K6PC-5, but was potentiated by the short-chain ceramide C6. Moreover, SKI-349 induced Akt-mTOR inactivation, JNK activation, and oxidative injury in primary NSCLC cells. In addition, SKI-349 decreased bromodomain-containing protein 4 (BRD4) expression and downregulated BRD4-dependent genes (Myc, cyclin D1 and Klf4) in primary NSCLC cells. At last, SKI-349 (10 mg/kg) administration inhibited NSCLC xenograft growth in nude mice. Akt-mTOR inhibition, JNK activation, oxidative injury and BRD4 downregulation were detected in SKI-349-treated NSCLC xenograft tissues. Taken together, targeting SphK1/2 by SKI-349 potently inhibits NSCLC cell growth in vitro and in vivo.


INTRODUCTION
Lung cancer is still a common global malignancy and a leading cause of cancer-related human mortalities [1,2]. There are two main subtypes of lung cancer, small cell lung cancer (SCLC) and non-small cell lung cancer (NSCLC). Among them, NSCLC accounts for over 80-85% of all lung cancer [1,2]. Metastatic, recurrent and other advanced NSCLC have very limited clinical treatment options and patients are often with extremely poor prognosis [3]. Therefore, uncovering the molecular mechanisms of NSCLC and identifying novel biomarkers could be crucial for early diagnosis and better treatment [4].

Cellular function studies
Normal distribution was assumed for all endpoints. For in vitro experiments, the estimated sample sizes were exceeded to ensure enough power and using standard deviations (SD). In addition, in vitro experiments were repeated in different NSCLC cells to enhance the power and the confidence of the result outcomes. Each in vitro cellular experiments were repeated five times. NSCLC cells or the epithelial cells were distributed into 96-well/12-well/6-well plates or tissue culturing slides at 70-80% confluence and treated with SKI-349 (at the applied concentrations) or the vehicle control. Cells were then maintained in the conditional medium for indicated time periods and were subject to different cellular functional assays. Cell counting kit-8 (CCK-8) viability assay, colony formation, propidium iodide (PI)-FACS, cell proliferation detection by measuring nuclear 5-ethynyl-2'-deoxyuridine (EdU) staining, BrdU incorporation ELISA assay, the in vitro cell migration/invasion by the "Transwell"/"Matrigel Transwell" assays, cell apoptosis detection by measuring the nuclear terminal deoxynucleotidyl transferase (TdT) dUTP nick-end labeling (TUNEL) staining and Annexin V-PI FACS, the caspase-3 and caspase-7 activity assays, trypan blue-staining assay of cell death, mitochondrial depolarization detection by measuring JC-1 green monomer intensity, reactive oxygen species (ROS) detection by the CellROX dye assay, lipid peroxidation detection by the thiobarbituric acid reactive substances (TBAR) activity assay were described in detail in our previous studies [21,22,31,32]. Assays of the SphK activity and ceramide contents were described previously as well [21]. For "Transwell" studies, cells with SKI-349 treatment were allowed to migrate for only 24 h, and no significant cytotoxicity was yet detected.

Gene and protein expression detection
The detailed protocols of Western blotting, quantitative real time-PCR (qRT-PCR), and data quantification were described in our previously studies [21,22]. The verified mRNA primers were provided by Genechem (Shanghai, China). The uncroppred Western blotting images were presented in Fig. S1.

Constitutively-active mutant Akt1
As reported [21], the recombinant constitutively-active Akt1 (caAkt1, S473D) adenoviral construct was transduced to pNSCLC-1 cells, and stable cells established by FACS sorting and selection. caAkt1 expression in the stable cells was verified by Western blotting.

BRD4 overexpression
A BRD4 expression GV248 lentiviral construct was provided by Dr. Zheng [33] and was transduced to primary human NSCLC cells. Puromycin was then added to select stable cells for another 24 h. Expression of BRD4 was verified by Western blotting.

SphK1/2 dual silencing
The SphK1 shRNA lentiviral particles and SphK2 shRNA lentiviral particles, both from Santa Cruz Biotech (Santa Cruz, CA), were together added to primary human NSCLC cells for 36 h. Puromycin was then added to select stable cells for another 24 h. Expression of SphK1/2 in the stable cells was tested by Western blotting.

Tumor xenograft studies
The nude mice (18.5-19.0 g, 5-6 week old, half male hale female) were maintained under the Animal Facility of Soochow University. pNSCLC-1 cells (6 × 10 6 cells per mouse) were subcutaneously (s.c.) injected to the flanks of the nude mice. The patient-derived xenograft (PDX) NSCLC model was established by subcutaneous injection of pNSCLC-1 primary cells (6 × 10 6 for each mouse) to the nude mice. pNSCLC-1 xenografts were formed 20 days following initial cell injection, with tumor volume close to 100 mm 3 ("Day-0"). pNSCLC-1 xenograft-bearing nude mice were then intraperitoneally injected with SKI-349 (at 10 mg/kg body weight, every other day, for seven times/14 days [18]) or the vehicle control ("Veh", [18]). The tumor volumes were calculated using the described formula [21] and recorded weekly from "Day-0" to "Day-42". For animal xenograft studies, power and sample sizes were estimated using empirical data from extensive experience in our lab for the expected SD and were supported by the current sample size calculators. All animal procedures were approved by IACUC (institutional animal care and use committee) and Ethic Committee of Soochow University (SDM2021-1103).

Statistical analysis
Data were always with normal distribution and were presented as mean ± SD. The one-way analysis of variance (ANOVA) plus Tukey's multiple comparison test (GraphPad Prism 5.01) were utilized for the comparison of multiple groups. For comparison between two groups, the Student t test (Excel 2007) was utilized. IC-50 was calculated by nonlinear regression analysis using GraphPad Prism 5.01. P-values < 0.05 were statistically significant.
Y. Xue et al.
SKI-349, at 1-10 μM, potently inhibited BrdU incorporation in pNSCLC-1 cells, suggesting its anti-proliferative activity (Fig. 1C). Moreover, the percentage of EdU-positively stained nuclei was significantly decreased in SKI-349 (1-10 μM)-treated pNSCLC-1 cells (Fig. 1D, E), further supporting the anti-proliferative activity by the dual inhibitor. The results of these titration experiments ( Fig. 1A-E) showed that 5 μM of SKI-349 induced robust anticancer activity in NSCLC cells, and this concentration was close to IC-50 (the concentration resulting in 50% CCK-8 OD reduction), this concentration was selected for the following studies.
Next we tested whether the dual inhibitor could affect cell cycle progression. Following SKI-349 (5 μM) treatment, the G1-phase pNSCLC-1 cell percentage was significantly increased, whereas the S-phase cell percentage was decreased (Fig. 1F). These results implied that the dual inhibitor induced G1-S arrest in primary NSCLC cells. Furthermore, SKI-349 (5 μM, for 24 h) potently inhibited pNSCLC-1 cell in vitro migration and invasion ( Fig.  1G-H).
Experiments were carried out to further study the potential activity of SKI-349 in other NSCLC cells. The primary NSCLC cells deriving from two other primary patients (pNSCLC-2 and pNSCLC-3, see our previous study [21]) and the immortalized NSCLC cell lines (A549 and NCI-H1944) were subject to SKI-349 treatment. The CCK-8 viability assay results showed that SKI-349, in a dosedependent manner, deceased viability in the tested primary and immortalized human NSCLC cells (Fig. 1I). The IC-50 was again close to 5 μM in the tested NSCLC cells (Fig. 1I). Moreover, SKI-349 (5 μM) inhibited cell proliferation (EdU-positive nuclei percentage reduction, Fig. 1J) and migration ("Transwell" assays, Fig. 1K). These results clearly supported the anti-NSCLC cell activity by the SphK1/2 dual inhibitor. We also tested the potential effect of SKI-349 on the non-cancerous normal epithelial cells. In BEAS-2B lung epithelial cells and primary human lung epithelial cells ("pEpi") [21], SKI-349 (5 μM) treatment failed to exert significant inhibition on cell viability (CCK-8 OD, Fig. 1L), proliferation (EdU-positive nuclei percentage, Fig. 1M) and migration (Fig. 1N). These results implied a cancer cell-specific effect by the dual inhibitor.

SKI-349 provokes apoptosis in NSCLC cells
SphK inhibition can induce significant apoptosis activation in NSCLC cells [11,13,16,17]. Next, we tested the potential effect of SKI-349 on cell apoptosis. Treatment with SKI-349 (5 μM, 48 h) robustly increased the caspae-3 activity and the caspase-7 activity in pNSCLC-1 primary cells (Fig. 2 A). Figure 2B confirmed that cleavages of caspae-3, caspase-9 and PARP [poly(ADP-ribose) polymerase] were increased in SKI-349-treated pNSCLC-1 cells. SKI-349 failed to significantly increase TRAIL expression and caspase-8 cleavage in pNSCLC-1 cells (Fig. 2B). Insertion of homo/ hetero-oligomerized Bax and Bak into the mitochondrial outer membrane led to pore formation, membrane permeabilization, and depolarization of the mitochondria, which will cause Cyto-C release and mitochondrial apoptosis cascade activation [34,35]. We found that the cytosol cytochrome c release and Bax expression were significantly increased in pNSCLC-1 cells after SKI-349 treatment (Fig. 2C). The dual inhibitor also induced significant mitochondrial depolarization, which was evidenced by the accumulation of JC-1 green monomer (Fig. 2D).
weights, as shown in Fig. 6D, were however not significantly different between the two groups.

DISCUSSION
The results of the present study supported that targeting SphK1/2 by SKI-349 could result in profound anti-NSCLC cell activity. In primary human NSCLC cells and immortalized lines, SKI-349 potently inhibited cell proliferation, cell cycle progression, migration and viability, while provoking apoptosis. It was however non-cytotoxic to lung epithelial cells. The dual inhibitor inactivated SphK and resulted in ceramide accumulation in NSCLC cells (Fig. 6N). Significantly, there was a time-dependent response following treatment of SKI-349 treatment in NSCLC cells. SKI-349 first resulted in SphK1/2 inhibition, ceramide accumulation and signaling changes (12 h ) and lipid peroxidation (by measuring TBAR activity, (F)) were tested. pNSCLC-1 and pNSCLC-2 cells were pretreated for 45 min with the antioxidant NAC (n-acetyl-L-cysteine 500 μM), the JNK inhibitor SP600125 (JNKi, 10 μM) or 0.1% DMSO, followed by SKI-349 ("SKI", 5 μM) stimulation and cells were further cultivated for 72 h; Cell death (G) and apoptosis (H) were tested. Data were presented as mean ± SD (n = 5). *P < 0.05 versus "Veh" treatment. # P < 0.05 (A-C). # P < 0.05 versus "DMSO " group (G, H). Experiments in this figure were repeated five times, and similar results were obtained. Scale bar = 100 μm (E).
xenograft growth in nude mice. Notably, SKI-349-caused anti-NSCLC cell activity was significantly more robust than the SphK1 inhibitor or plus the SphK2 specific inhibitor. Therefore, concurrent inhibition of SphK1 and SphK2 by SKI-349 resulted in robust killing of NSCLC cells. Targeting SphK1/2, using genetic methods or pharmacologic agents, has been verified as an important strategy to inhibit NSCLC [11,15,16,66,67]. While the role of SphK1 in NSCLC has been well-studied [11-14, 66, 67], recent studies have focused on SphK2 in tumorigenesis and progression of NSCLC. Wang et al., have shown that SphK2 is overexpression in NSCLC, which is correlated with disease grade, lymph node status and NSCLC stage as well as tumor size and histology type [15]. Overexpressed SphK2 could be a valuable biomarker for prognosis and promising therapeutic target for NSCLC [15]. Yang et al., reported that ABC294640, the SphK2 inhibitor, sensitized TRAIL-induced NSCLC cell apoptosis possibly through upregulating death receptor4/5 (DR4/5) [17]. Liu et al., demonstrated that small interference RNA (siRNA)-induced silencing of SphK2 inhibited NSCLC cell proliferation and chemo-sensitized NSCLC cells to gefitinib-induced apoptosis [16].
We here showed that Akt-mTOR inactivation, JNK activation and oxidative injury were detected in SKI-349-treated NSCLC cells and SKI-349-administrated pNSCLC-1 xenograft tissues. More importantly, caAkt1, the antioxidant NAC and the JNK inhibitor SP600125 all ameliorated SKI-349-induced NSCLC cell death. Therefore, alteration of these signaling cascades together could explain the superior anti-NSCLC activity by the novel SphK1/2 dual inhibitor (Fig. 6N).
BRD4 regulates epigenetic processes by associating with acetylated-histones [62][63][64]. BRD4 is also essential for the transcription elongation and expression of various oncogenes by associating with pTEFb (positive transcription elongation factor b) and by phosphorylating RNA polymerase II [63,65]. Recent studies have proposed BRD4 as an important therapeutic target of NSCLC, and it is essential for the expression of Bcl-2, c-Myc, cyclin D1 and other oncogenic genes [63,65]. Liao et al., have reported that BRD4 is overexpressed in NSCLC tissues and is correlated with histological type, lymph node metastasis, tumor stage and differentiation, and the poor prognosis [76]. Gao et al., showed that genetic silencing or pharmacological inhibition of BRD4 inhibited NSCLC cell growth possibly by downregulating eIF4Emediated transcription [77]. BRD4 inhibition or depletion enhanced TRAIL-induced NSCLC cell apoptosis by inactivating NFκB cascade [78].
BRD4 physically associated with YAP/TAZ transcription factors, increasing expression of a number of different growth-regulating genes that are essential for the progression of cancer cells [79]. Frequent ARID1A (the AT-rich interactive domain 1A [SWI-like] gene) depletion, detected in 20% of all lung cancers, induced chromatin remodeling and glycolysis, inhibiting cell death induced by the BRD4 inhibitor JQ1 [80]. He et al., have shown that BRD4 inhibition and ARID2 depletion synergistically inhibited expression of DNA repair-related genes and induced robust cytotoxicity in cancer cells [81]. One important finding of the

N.
Fig. 6 SKI-349 administration inhibits NSCLC xenograft in nude mice. The mice bearing pNSCLC-1 xenografts were subject to the applied SKI-349 administration or vehicle control treatment ("Veh"), with 10 mice per group (n = 10). The tumor volumes (A) and the mice body weights (D) were measured every 6 days ("Day-0" to "Day-42"); The estimated daily tumor growth (B) and weights of pNSCLC-1 xenografts at "Day-42" (C) were measured. At "Day-6" and "Day-12", one pNSCLC-1 tumor per group were isolated, and the relative SphK activity was tested (E). Expression of listed genes and proteins in the described tumor tissues were tested (F, G, H, J, K, L and M). The relative lipid peroxidation intensity (TBAR activity, (I)) was examined as well. The proposed signaling pathway of the study (N). Data were presented as mean ± SD. *P < 0.05 versus "Veh" treatment. "n.s." stands for non-statistical difference (P > 0.05).
present study is that SKI-349 downregulated BRD4 and inhibited BRD4-dependent genes (Myc, cyclin D1 and Klf4) in primary NSCLC cells (Fig. 6N). More importantly, SKI-349-induced NSCLC cell death was ameliorated by ectopic overexpression of BRD4. BRD4 silencing by the SphK1/2 dual inhibitor could be another reason to explain its superior anti-NSCLC cell activity. Novel and more efficient targeted therapies are urgently needed for NSCLC. We showed that targeting SphK1/2 by SKI-349 potently inhibited NSCLC cell growth in vitro and in vivo. However, the conclusion was based on the in vitro cellular studies and animal xenograft results. The efficacy and safety of the dual inhibitor against human NSCLC warrant further characterizations under clinical studies. The underlying mechanisms of SKI-349induced anti-NSCLC cell activity need more exploration as well.

DATA AVAILABILITY
All data are available upon request.