GSK1059615 kills head and neck squamous cell carcinoma cells possibly via activating mitochondrial programmed necrosis pathway

This study tested the anti-head and neck squamous cell carcinoma (HNSCC) cell activity by GSK1059615, a novel PI3K and mTOR dual inhibitor. GSK1059615 inhibited survival and proliferation of established (SCC-9, SQ20B and A253 lines) and primary human HNSCC cells. GSK1059615 blocked PI3K-AKT-mTOR activation in HNSCC cells. Intriguingly, GSK1059615 treatment in HNSCC cells failed to provoke apoptosis, but induced programmed necrosis. The latter was tested by mitochondria depolarization, ANT-1-cyclophilin-D mitochondrial association and lactate dehydrogenase (LDH) release. Reversely, mPTP blockers (sanglifehrin A, cyclosporin A and bongkrekic acid) or cyclophilin-D shRNA dramatically alleviated GSK1059615-induced SCC-9 cell death. Further studies demonstrated that GSK1059615 i.p. injection suppressed SCC-9 tumor growth in nude mice, which was compromised with co-administration with cyclosporin A. Thus, targeting PI3K-AKT-mTOR pathway by GSK1059615 possibly provokes programmed necrosis pathway to kill HNSCC cells.


INTRODUCTION
Head and neck squamous cell carcinoma (HNSCC) is a large heterogeneous family of carcinomas, including carcinomas of face, nasopharynx, oral cavity, and larynx [1][2][3]. The overall survival is still poor for those with high-degree or malignant tumors [1][2][3][4]. One major signaling that is often dysregulated in HNSCC is phosphatidylinositol 3-kinase-AKT-mammalian target of rapamycin (PI3K-AKT-mTOR) cascade [5][6][7]. Recent research efforts have developed a novel PI3K and mTOR dual inhibitor, named GSK1059615 [8]. As compared to other PI3K-AKT-mTOR specific inhibitors, the advantage of this compound is significant, as it simantanuously blocks PI3K and mTOR [8]. Therefore, it has the potential to completely shut down the PI3K-AKT-mTOR cascade [8]. The potential effect of GSK1059615 on HNSCC cells is evaluated in this preclinical study.
The activity of GSK1059615 on other HNSCC cells was also tested. In both A253 and SQ20B HNSCC cells [17], GSK1059615 (3 μM, 72h) largely decreased cell survival (MTT OD, Figure 1D). On the other hand, the very same GSK1059615 treatment failed to inhibit the survival of two oral epithelial cell lines ("Oepi1/2") ( Figure 1D), implying that GSK1059615 could be cytotoxic only to cancer cells. In order to test the effect of GSK1059615 in primary cancer cells, a total of four lines of primary ("patient-derived") oral cavity carcinoma (OCC) cells were established (named "OCC1-4"), which were also treated with GSK1059615 (3 μM, 72h). MTT assay results in Figure 1E showed that GSK1059615 was cytotoxic to all the primary cancer cells. Remarkably, we found that GSK1059615 was more potent that other known AKT inhibitors (i.e. LY294002, Wortmannin and perifosine) in killing SCC-9 cells ( Figure 1F). Together, these results demonstrate that GSK1059615 is cytotoxic to established and primary human HNSCC cells.

GSK1059615 inhibits human HNSCC cell proliferation
Cytotoxicity in HNSCC cells could be due to proliferation inhibition. Next, proliferation of GSK1059615-treated HNSCC cells was tested by the BrdU ELISA assay and [H 3 ] thymidine incorporation assay [18]. Results from both assays demonstrated clearly  that GSK1059615 dose-dependently inhibited SCC-9 cell proliferation ( Figure 2A and 2B), as the BrdU ELISA OD ( Figure 2A) and [H 3 ] thymidine incorporation ( Figure  2B) were both decreased following GSK1059615 (1-30 μM) treatment. Expression of proliferation-associated proteins, including cyclin D1 and cyclin B1, was also significantly downregulated following GSK1059615 (1-10 μM) treatment ( Figure 2C). Notably, to test cell proliferation, cells were incubated with GSK1059615 for only 24h, when no significant cytotoxicity was yet noticed ( Figure 1A). BrdU ELISA assay was also performed to test proliferation of other HNSCC cells with GSK1059615 treatment. Results in Figure 2D showed clearly that GSK1059615 (3 μM) was anti-proliferative in two other HNSCC cell lines: SQ20B and A253. Yet, the same GSK1059615 treatment failed to inhibit proliferation of oral epithelial cells ("Oepi1/2") ( Figure 2D). In the primary OCC cells (all four lines, "OCC1-4"), treatment with GSK1059615 (3 μM, 24h) also inhibited cell proliferation, which was again indicated by BrdU ELISA OD reduction ( Figure 2E). Collectively, these results imply that GSK1059615 inhibits human HNSCC cell proliferation.

GSK1059615 fails to provoke apoptosis in HNSCC cells
We also tested the potential effect of GSK1059615 on cell apoptosis. Three different assays were performed, including the Annexin V FACS assay, Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) staining assay and Histone DNA apoptosis ELISA assay. To our surprise, the results of these assays showed that GSK1059615 (3 μM, the cytotoxic dose) failed to induce significant apoptosis in SCC-9 cells ( Figure 4A-4C). Notably, to test cell apoptosis, cells were treated with GSK1059615 for three different time points (24h/48h/72h), yet no apoptosis activation was noticed at any time point ( Figure 4A-4C). On the other hand, gemcitabine, served as a positive control, provoked profound apoptosis in SCC-9 cells (Figure 4A-4C). Further studies showed that gemcitabine, but not GSK1059615, induced obvious caspase-3/caspase-9 cleavage in SCC-9 cells ( Figure 4D). Together, GSK1059615, although is cytotoxic, but fails to induce HNSCC cell apoptosis.
The above results suggest that GSK1059615 provokes programmed necrosis to kill SCC-9 cells. To further this hypothesis, shRNA method was applied to stably knockdown cyclophilin-D, the key component of mPTP. A total of three non-overlapping cyclophilin-D shRNAs (from Dr. Xu [29]) were applied: cyclophilin-D shRNA-1/2/3. All three of them efficiently downregulated cyclophilin-D in SCC-9 cells ( Figure 5E). Remarkably, GSK1059615 (3 μM)-induced cytotoxicity was largely alleviated in the cyclophilin-D-silenced cells ( Figure 5F and 5G). Thus, in line with the above pharmacological evidences, the genetic evidences here further confirmed that programmed necrosis pathway mediates GSK1059615-induced cytotoxicity against SCC-9 cells.

GSK1059615 inhibits SCC-9 tumor growth in nude mice, and its activity is compromised with cyclosporin A co-administration
At last, we studied the anti-tumor activity of GSK1059615 in vivo, via a SCC-9 xenograft nude mice model. As demonstrated, i.p. daily administration of GSK1059615 at 30 mg/kg significantly inhibited SCC-9 tumor growth in the nude mice ( Figure 6A). Estimated tumor growth, expressed as mm 3 per day [30,31], was also dramatically inhibited with GSK1059615 administration ( Figure 6B). The weight of GSK1059615-treated tumors was also dramatically lighter than those of vehicle control tumors ( Figure 6C). Significantly, as shown in Figure  6A-6C, co-administration of cyclosporin A (5 mg/kg, i.v., daily) [32], the cyclophilin-D inhibitor, largely attenuated GSK1059615-induced anti-SCC-9 tumor activity. Thus, mPTP and programmed necrosis pathway may also be required for GSK1059615-induced anti-tumor activity in vivo. Notably, cyclosporin A alone failed to inhibit SCC-9 tumor growth in the mice ( Figure 6A-6C).Importantly, the mice body weight was not significantly different between the groups ( Figure 6D), suggesting that these mice were well-tolerated to the tested regimens. Together, GSK1059615 inhibits SCC-9 tumor growth in nude mice, and its anti-tumor activity in vivo is compromised with co-administration of cyclosporin A.
In this study, we found that GSK1059615 similarly provoked the programmed necrosis pathway in HNSCC cells, which was evidenced by mitochondria depolarization, ANT-1-cyclophilin-D mitochondrial association and LDH release. Reversely, mPTP blockers or shRNA knockdown cyclophilin-D dramatically alleviated GSK1059615-induced killing of HNSCC cells. In the SCC-9 xenograft tumor nude mice model, coadministration of cyclosporin A significantly attenuated GSK1059615-induced anti-tumor activity. Therefore, targeting PI3K-AKT-mTOR pathway by GSK1059615 possibly provokes programmed necrosis to kill HNSCC cells.

Primary culture human cancer cells and epithelial cells
Four written-informed consent patients with oral cavity carcinoma (OCC) were enrolled, who received no therapy prior to surgery. OCC tissues and the surrounding normal epithelial tissues were obtained at the time of surgery, and were separated very carefully. As described [43], the tissue specimens were washed and incubated with 0.1% collagenase I for digestion. Afterwards, cells were filtered via a 70-μm nylon cell strainer. Primary cells were cultured in complete DMEM/F12 medium with bFGF and EGF [43]. A total of four lines of primary OCC cells ("OCC1-4") and two lines of oral cavity epithelial cells ("Oepi1-2") were established. Studies requiring human tissues were reviewed and approved by the Institutional Ethics Committee and Internal Review Committee, and were conducted according to Declaration of Helsinki.

MTT assay
The routine MTT (Sigma) assay was applied to test cell survival with manufactory's recommendation [44].

Clonogenicity assay
HNSCC cells (1*10 4 per 10-cm dish) were trypsinized and suspended in 0.5% agarose-containing complete medium. Ten days following indicated treatment, the survival colonies were stained and counted manually.

LDH assay
Cell death was tested via medium release of LDH, using a routine two-step LDH assay kit (Takara, Tokyo, Japan) [45]. LDH in the conditional medium was normalized to the total LDH, reflecting cell necrosis percentage [45].

In vitro proliferation assay
Cell proliferation BrdU ELISA assay and [H 3 ] thymidine incorporation assay were described in detail in other study [18]. The values were always normalized to the untreated control group. www.impactjournals.com/oncotarget

ELISA assay of cell apoptosis
The Histone DNA apoptosis ELISA (Roche, Shanghai, China) assay was utilized to quantify cell apoptosis according to the manufacturer's instructions. Detailed protocol can be viewed in other studies [46,47]. ELISA OD at 450 nm was utilized to quantify cell apoptosis.

TUNEL assay
The TUNEL In Situ Cell Death Detection Kit (Roche, Shanghai, China) was utilized to stain nuclei of apoptotic cells. TUNEL ratio of 200 cells per treatment was recorded.

Annexin V FACS assay
Cells with treatment were incubated immediately with Annexin V-FITC (5 μg/mL, Invitrogen, Shanghai, China) and the Binding Buffer (Invitrogen), which were subjected to flow cytometry assay of Annexin V.

shRNA knockdown of cyclophilin-D
The three verified lentiviral cyclophilin-D shRNAs (with non-overlapping sequences) and the nonsense scramble shRNA were gifts from Dr. Xu [29]. The lentiviral shRNA was added directly to cells for 12h. Cells were then subjected to puromycin (5 μg/mL, Sigma) selection for additional 4-5 passages. cyclophilin-D expression in the stable cells was detected by Western blot assay.

Mitochondrial immunoprecipitation (mito-IP) assay
As described [25], Mitochondria/Cytosol Fractionation Kit (BioVision, Shanghai, China) was utilized to acquire mitochondrial proteins. Six-hundred μg of mitochondrial lysates per treatment were pre-cleared [25]. The lysate supernatant was then rotated overnight with 0.05 μg of designated antibody. Next, the protein A/G PLUS-agarose was added to capture the complex.
Pellets were washed, and resuspended in lysis buffer. The immuno-complex was then assayed via Western blot.

Tumor xenograft assay and IHC staining
Five millions SCC-9 cells per mouse were inoculated s.c. into the female nude mice (5-7 week age, 18-20 grams in weight). Within three weeks, established xenograft tumors were established with volumes around 0.1 cm 3 . The mice were then randomized into four groups as mentioned in the text (n=10 per group). Tumor size was measured once every week for a total of 5 weeks, using the modified ellipsoid formula: (π/6) × AB 2 , A represents the longest, and B represents the shortest perpendicular axis of an tumor mass [46]. The animal procedure was approved by the IACUC of authors institutions.

Statistical analysis
Results were compared by one-way analysis of variance (ANOVA) followed by Turkey's test. Values of p < 0.01 were considered as statistically significant.

Author contributions
All authors carried out all the experiments, participated in the design of the project conceived of the study, and participated in its design and coordination and helped to draft the manuscript.