The pro-tumorigenic activity of p38γ overexpression in nasopharyngeal carcinoma

It is urgent to identify and validate biomarkers for early diagnosis and efficient treatment of nasopharyngeal carcinoma (NPC). Recent studies have proposed p38 gamma (p38γ) as a cyclin-dependent kinase (CDK)-like kinase that phosphorylates retinoblastoma (Rb) to promote cyclins expression and tumorigenesis. Here the Gene Expression Profiling Interactive Analysis (GEPIA) database and results from the local NPC tissues demonstrate that p38γ is significantly upregulated in NPC tissues, correlating with poor overall survival. Furthermore, p38γ mRNA and protein expression is elevated in established NPC cell lines (CNE-1 HONE-1 and CNE-2) and primary human NPC cells, but low expression detected in human nasal epithelial cells. In established and primary NPC cells, p38γ depletion, using the shRNA strategy or the CRISPR/Cas9 gene-editing method, largely inhibited cell growth, proliferation and migration, and induced significant apoptosis activation. Contrarily, ectopic p38γ overexpression exerted opposite activity and promoted NPC cell proliferation and migration. Retinoblastoma (Rb) phosphorylation and cyclin E1/A expression were decreased in NPC cells with p38γ silencing or knockout, but increased after p38γ overexpression. Moreover, mitochondrial subcellular p38γ localization was detected in NPC cells. Significantly, p38γ depletion disrupted mitochondrial functions, causing mitochondrial depolarization, reactive oxygen species production, oxidative injury and ATP depletion in NPC cells. In vivo, intratumoral injection of adeno-associated virus-packed p38γ shRNA potently inhibited primary human NPC xenograft growth in nude mice. In p38γ shRNA virus-injected NPC xenograft tissues, p38γ expression, Rb phosphorylation, cyclin E1/A expression and ATP levels were dramatically decreased. Taken together, we conclude that p38γ overexpression is required for NPC cell growth, acting as a promising therapeutic target of NPC.

The current clinical treatments for NPC include surgery, platinum-based chemotherapy and radiotherapy [4][5][6]. The latter includes 3D conformal radiation therapy, intensity-modulated radiation therapy, particle beam therapy and brachytherapy [4][5][6]. It is estimated that 70-90% of NPC patients could respond well to radiotherapy (or in combination with chemotherapies) [1][2][3]. Yet, for the recurrent, metastatic and other advanced NPC patients, the progression-free survival (PFS) and overall survival (OS) are not satisfactory [1]. Treatments for these patients are largely limited to palliative systemic therapies [1,4,5]. It is therefore extremely important to identify and validate biomarkers for early diagnosis and efficient therapy of NPC [6].

Cell culture
The immortalized NPC cell lines, including CNE-1, HONE-1 and CNE-2, were purchased from the Cell Bank of Shanghai Institute of Biological Science, CAS (Shanghai, China). Cells were cultivated in RPMI-1640/DMEM supplemented with 8-10% FBS and antibiotics, at 37°C in a humidified 5% CO 2 incubator. The primary human NPC cells that were derived from one primary NPC patient, namely pNPC-1, as well as the primary human nasal epithelial cells (HNEpC) that were derived from two donors (pHNEpC-1 and pHNEpC-2), were provided by Dr. Chen at Jiangsu University [21]. The protocols of using human cells were approved by the Ethic Committee of Soochow University, in according with the principles of Declaration of Helsinki.

Human tissues
NPC tumor tissues and the matched adjacent nasopharynx epithelial tissues were from a set of fifteen (15) primary NPC patients. The patients were administrated at authors' institutions, provided written-informed consents and received no prior therapies before surgeries. Tissues were stored in liquid nitrogen. The protocols were approved by the Ethics Committee of Soochow University, in according to the principles of Declaration of Helsinki.

p38γ silencing or overexpression
The two non-overlapping shRNAs targeting p38γ (p38γ-shRNA-s1 and p38γ-shRNA-s2, from Dr. Shi [22]) and a negative control with the scrambled non-sense sequence (shC) were constructed into GV248 lentiviral vectors provided by Shanghai Genechem Co. (Shanghai, China). The full-length p38γ cDNA sequence (also from Dr. Shi [22]) was subcloned into the GV248 lentiviral vector (Genechem Co.). The lentivirus was produced by transfecting HEK293T cells with the plasmids using a lentivirus packaging mix (Genechem Co.). The virus was enriched and filtered. Cells were then infected with the virus and selected with puromycin for 96 h. Expression of p38γ in the stable cells was verified by qRT-PCR and Western blotting assays. For in vivo studies the p38γ-shRNA-s1 sequence was inserted into the adeno-associated virus (aav) construct, aav9 (Genechem, Shanghai, China).

p38γ knockout (KO)
A CRISPR/Cas9 PX458-GFP construct containing p38γ small guide RNA (sgRNA) was provided by Dr. Shi [22]. NPC cells were placed into six-well plates and transfected with Cas9-expressing construct (Genechem, Shanghai, China). The stable Cas9-expressing cells were established after puromycin selection. Cells were then transfected with the CRISPR/Cas9-p38γ-KO construct. The transfected cells were distributed into 96-well plates for 96 h, subject to p38γ KO screening using qPCR and Western blotting assays. At last, the single stable monoclonal p38γ knockout (p38γ-KO) NPC cells were established.

Gene detection
Detailed protocols for Western blotting, quantitative reverse transcriptionpolymerase chain reaction (qRT-PCR), co-immunoprecipitation (Co-IP) were described previously [23][24][25]. The primers of this study were provided by Dr. Shi at Soochow University [22]. The isolation of mitochondria through the high-speed centrifugation was through the Pierce kit (Pierce Biotechnology, Rockford, IL) according to the protocols attached. The uncropped blotting images were presented in Supplementary Fig. S1.

Reactive oxygen species (ROS) detection
Production of ROS was measured by a CellROX probe [28]. Cells with the designated genetic modifications were seeded into 96-well plates and were stained with CellROX Deep Red (10 µM) for 30 min at 37°C. Cells were then washed twice. CellROX fluorescence was measured at 625 nm under a Fluoroskan Ascent FL microplate reader. The fluorescence was also photographed under an Olympus fluorescence microscope (Olympus, Tokyo, Japan).

Measuring mitochondrial membrane potential
Cells with the designated genetic modifications were seeded into 24well plates. Afterward, cells were incubated in total darkness with 10 µM JC-1 (Invitrogen) for 45 min at room temperature [29]. JC-1 green monomer intensity was measured under a Fluoroskan Ascent FL microplate reader. JC-1 green monomers and red dimers (J-aggregates) were photographed as well under an Olympus fluorescence microscope (Olympus, Tokyo, Japan).

Caspase activity
The activities of caspase-3 and caspase-7 were examined by an Apo-ONE Homogeneous caspase 3/7 assay kit (Promega Corporation, Madison, WI) according to the manufacturer's protocols.
Histone DNA ELISA and ssDNA ELISA Cells with the designated treatments were plated into the 96-well plates at a density of 3 × 10 3 cells/well. After culturing for applied time periods, the Histone-bound DNA contents and single-strand DNA (ssDNA) contents were analyzed by the corresponding ELISA kits (Roche, Shanghai, China). The ELISA OD at 450 nm in each well was recorded.

ATP contents
NPC cells with the applied genetic treatments were placed into 12-well plates at 1 × 10 5 cells per well and cultured for applied time periods. An ATP assay kit (Biyuntian, Wuxi, China) was utilized to quantify ATP contents according to the attached protocols.

Rb mutation
As described previously [30], the non-phosphorylated mutant human Rb ("Rb-mut") was generated by changing 15 Ser/Thr sites to Ala, with Ser567 left unaltered, and a HA tag placed on the N-terminus. The Rb-mut construct was generated by Genechem (Shanghai, China) and was transfected to transduced to p38γ-overexpressed CNE-1 cells (OE-p38γ-L1) through Lipofectamine 3000 (24 h per round for three rounds).

Xenograft assay
The nude mice were half male half female, 4-5 week old and 18.2-19.2 g in weight. Mice were provided by Shanghai SLAC Laboratory Animal Center (Shanghai, China). Mice were housed under the Guide for the Care and Use of Laboratory Animals (NIH). The primary human NPC cells, pNPC-1, were subcutaneously (s.c.) injected to the flanks of the nude mice. Five million cells in 100 µL of DMEM/Matrigel (no serum) were inoculated. The volume of each xenograft was close to 100 mm 3 within three weeks (labeled at "Day-0"). The xenograft-bearing mice were then randomly assigned into two groups (ten mice per group), receiving intratumoral injection of designated adeno-associated virus (aav)-packed shRNA. The tumor volume was determined using the described formula [31]. The protocols were

Statistical analyses
Data were with normal distribution and were presented as mean ± standard deviation (SD). Statistical analyses were carried out using the Student's t test (Excel 2007) for comparisons between two groups, or oneway ANOVA plus a Scheffe' and Tukey Test (SPSS 23.0) for comparisons between multiple groups. P < 0.05 was considered to indicate a significant difference. In vitro experiments were repeated five times.

RESULTS p38γ is overexpressed in NPC
The Gene Expression Profiling Interactive Analysis (GEPIA) database was first consulted to test p38γ mRNA transcripts in NPC. Results showed that p38γtranscripts in NPC tumor tissues ("Tumor", n = 518) were significantly higher than those in the normal nasopharynx epithelial tissues ("Normal", n = 44) (Fig. 1A). Kaplan-Meier survival analyses, Fig. 1B, found that high p38γ expression in NPC patients correlated with poor survival.
To verify the bioinformatics results, we tested p38γ expression in local NPC tissues. A total of fifteen (n = 15) primary NPC patients were enrolled and fresh tissue specimens were obtained. The qRT-PCR assay results, Fig. 1C, showed that p38γ mRNA levels in NPC tumor tissues ("T") were significantly higher than those in the matched adjacent nasopharynx epithelial tissues ("N"). Testing p38γ protein expression, using Western blotting assays, further confirmed p38γ protein upregulation in NPC tumor tissues of four representative patients ("Patient #1/#2/#3/#4") ( Fig. 1D). The blotting data of all human tissues were combined. As shown, the p38γ protein expression in NPC tumor tissues was significantly higher than that in the normal nasopharynx epithelial tissues (Fig. 1E).
Next, the p38γ expression in different NPC cells was tested. As shown, p38γ mRNA ( Fig. 1F) and protein (Fig. 1G) expression levels were upregulated in immortalized NPC cell lines (CNE-1, HONE-1 and CNE-2) as well as in the primary human NPC cells (" pNPC-1"). Conversely, the low expression was detected in the primary human nasal epithelial cells (HNEpC) that were derived from two primary donors, pHNEpC-1 and pHNEpC-2 (Fig. 1F, G). These results show that p38γ is overexpressed in NPC. . Expression of p38γ mRNA (C) and protein (D and E) in local NPC tumor tissues ("T", n = 15) and the matched adjacent nasopharynx epithelial tissues ("N", n = 15) was shown, with results quantified. Expression of p38γ mRNA (F) and protein (G) in the listed NPC cells and the primary human nasal epithelial cells (pHNEpC-1 and pHNEpC-2) was tested as well. *P < 0.05 vs. "Normal"/"N" tissues or pHNEpC-1 cells.
To study the potential function of p38γ in non-cancerous epithelial cells, the primary human nasal epithelial cells (HNEpC)derived from two primary donors, pHNEpC-1 and pHNEpC-2, were cultured and infected with the p38γ-shRNA-s1 lentivirus. In the stable cells, depleted p38γ mRNA expression was detected (Fig. 2M). Significantly, shRNA-induced silencing of p38γ did not significantly inhibit the viability (CCK-8 OD, Fig. 2N) and proliferation (by measuring the EdU-positive nuclei ratio, Fig. 2O) in the nasal epithelial cells (Fig. 2O). These results supported a cancer cell-specific effect of p38γ shRNA.
The numbers of migrated and invaded CNE-1 cells were enhanced by ectopic p38γ overexpression (Fig. 5F, G).
To the primary human NPC cells (pNPC-1) and other immortalized cells (HONE-1 and CNE-2), the p38γ-expressing lentivirus was added, and stable cells established after puromycin selection: OE-p38γ. In the OE-p38γ NPC cells, the p38γ mRNA levels were significantly increased (Fig. 5H), but p38α mRNA levels were unchanged (Fig. 5I). In line with the results of CNE-1 cells, ectopic overexpression of p38γ in the NPC cells augmented cell proliferation (increased EdU-positive nuclei ratio, Fig. 5J) and the number of the migrated cells (Fig. 5K).
Besides locating in cell nuclei and cytosol, results from the Compartments Database (https://compartments.jensenlab.org) implied that p38γ can also locate at the mitochondrion (Fig. 6J).
We therefore analyzed whether p38γ depletion could affect mitochondrial functions in OS cells. The fluorescence confocal microscope results found that the p38γ protein (in green fluorescence) indeed localized to the mitochondria (labeled with MitoTracker Orange, in orange fluorescence) in CNE-1 cells (Fig. 6K). Furthermore, an examination of mitochondrial lysates isolated from CNE-1 cells demonstrated that p38γ was indeed enriched in the mitochondrial fraction (Fig. 6L), as indicated by VDAC1 (voltage-dependent anion-selective channel 1), the mitochondrial marker protein (Fig. 6L). Lamin-B1 is the nuclear marker protein and α-Tubulin is the cytosol marker protein (Fig. 6L). Significantly, the p38γ inhibitor PFD failed to affect p38γ mitochondrial enrichment in CNE-1 cells (Fig. 6L).
Significantly, p38γ silencing or KO disrupted mitochondrial functions, causing mitochondrial membrane potential reduction and mitochondrial depolarization. The latter was evidenced by JC-1 green monomer accumulation (Fig. 6M). Furthermore, the CellROX intensity was significantly increased in p38γ-shRNA cells and p38γ-KO CNE-1 cells, indicating ROS production and oxidative injury (Fig. 6N). In addition, the ssDNA contents were increased in Fig. 5 The cancer-promoting activity by p38γ overexpression in NPC cells. The immortalized NPC cell lines (CNE-1, HONE-1 and CNE-2) (A-K) or the primary human NPC cells (pNPC-1, H-K) were infected with the lentivirus encoding the p38γ-expressing construct ("OE-p38γ") or the lentivirus with the empty vector ("Vec"), with stable cells established after puromycin selection; Expression of listed genes was tested by qRT-PCR (A, B, H and I) and Western blotting (C) assays. Cells were further cultured for designated periods, cell viability (by recording CCK-8 OD, D) and cell proliferation (by recording the EdU-positive nuclei ratio, E and J) as well as cell migration ("Transwell" assays, F and K) and invasion ("Matrigel Transwell" assays, G) were tested by the indicated assays, with results quantified. *P < 0.05 vs. "Vec" cells.
depletion disrupted mitochondrial functions and depleted ATP in NPC cells.

p38γ silencing inhibits NPC xenograft tumor growth in nude mice
To examine the potential effect of p38γ on NPC cell growth in vivo, pNPC-1 primary cells (at 6 × 10 6 cells per mouse) were s.c. injected to the flanks of the nude mice. NPC xenograft tumors were established within three weeks after cell injection ("Day-0"). The xenograft-bearing mice were then randomly assigned into two groups. The treatment group ten mice (n = 10) received intratumoral injection of aav-packed p38γ-shRNA ["p38γ-shRNA-s1 (aav)"]. The control group contained ten mice as well and subject to intratumoral injection of aav-packed scramble control shRNA ["shC (aav)"]. The intratumoral injection of the virus was performed daily for 12 consecutive days. The tumor growth curve results, in Fig. 7A, demonstrated that p38γ-shRNA-s1 (aav) injection robustly inhibited pNPC-1 xenograft growth in nude mice (Fig. 7A). The estimated daily tumor growth was calculated by the described formulation:(tumor volume at Day-42 subtracting tumor volume at Day-0)/42 [21], and results further showed that injection of p38γ-shRNA-s1 (aav) largely inhibited pNPC-1 xenograft growth in mice (Fig. 7B). At Day-42, pNPC-1 xenografts of the two group mice were isolated and weighted individually. Results in Fig. 7C demonstrated that pNPC-1 xenografts with p38γ-shRNA-s1 (aav) injection were significantly lighter than those with shC (aav) injection. The mice body weights, as shown in Fig. 7D, were not significantly different between the two groups.
At Day-6 and Day-12, four hours after the virus injection, one tumor of each group was isolated. Fresh tumor tissues were obtained from the four total tumors. As shown p38γ protein (Fig. 7E) and p38γ mRNA (Fig. 7F) were indeed silenced in the p38γ-shRNA-s1 (aav)-injected pNPC-1 xenograft tissues. p38α protein expression was unchanged (Fig. 7E). In line with the in vitro findings, we found that Rb phosphorylation (Fig. 7G), cyclin E1 and cyclin A mRNA expression (Fig. 7F) and ATP levels (Fig. 7F) were dramatically decreased in p38γ-shRNA-expressing pNPC-1 xenograft tissues.

DISCUSSION
Recent studies have shown that p38γ can exert pro-tumorigenic activity in different human cancers. Wang et al. have shown that p38γ is overexpressed in pancreatic cancer, regulating KRAS oncogene signaling and aerobic glycolysis to promote tumorigenesis and cancer progression through PFKFB3-GLUT2 signaling [12]. Chen et al. have shown that p38γ overexpression-promoted renal cell carcinoma (RCC) cell growth, proliferation and migration, and it is a promising therapeutic target of RCC [14]. Su et al. found that p38γ expression is significantly elevated in colorectal cancer (CRC). p38γ silencing or knockout inhibited CRC cell growth, proliferation, and migration, and induced apoptosis activation [13]. In addition, Shi et al. reported that p38γ overexpression-promoted Rb phosphorylation and cyclin E1/cyclin A expression as well as osteosarcoma cell growth [22].
Here we provided strong evidence to support that p38γ is an important cancer-promoting gene and therapeutic target of NPC. GEPIA database and results from the local NPC tissues demonstrate that p38γ is significantly upregulated in NPC tissues, correlating with poor overall survival. Furthermore, p38γ mRNA and protein expression is significantly elevated in established and primary human NPC cells, whereas low expression detected in nasal epithelial cells. In NPC cells, p38γ shRNA or CRISPR/Cas9induced p38γ KO potently inhibited cell growth, proliferation, migration and invasion, and induced significant apoptosis activation. Contrarily, ectopic overexpression of p38γ exerted opposite activity and promoted NPC cell proliferation and migration. Importantly, p38γ shRNA failed to affect viability and proliferation in nasal epithelial cells. In vivo, intratumoral injection of p38γ shRNA aav potently inhibited primary human NPC xenograft growth in nude mice. Therefore, targeting p38γ could be a novel strategy to inhibit NPC.
Studies have shown that CDK-Rb-cyclin is dysregulated in NPC. Roniciclib (BAY1000394), a potent pan-CDK inhibitor, displayed promising anti-cancer activity in preclinical NPC models, either alone or in combination with cisplatin [32]. Niu et al. found that C-myc silencing inhibited NPC cell proliferation by inhibiting CDK-Rb-cyclin pathway [33]. Wu et al. reported that microRNA-188 Fig. 7 p38γ silencing inhibits NPC xenograft tumor growth in nude mice. The pNPC-1 xenograft -bearing nude mice were subject to intratumoral injection of aav-packed p38γ-shRNA ["p38γ-shRNA-s1 (aav)"] or aav-packed scramble control shRNA ["shC (aav)"]. Virus injection was performed daily for 12 consecutive days. The tumor growth curve results (A) and mice body weights (D) were recorded every six days. The estimated daily tumor growth, in mm 3 per day, was calculated by the described formula (B). At Day-42, pNPC-1 xenografts were isolated and weighted individually (C). At Day-6 and Day-12, four hours after the virus injection, one tumor of each group was isolated. Expression of listed proteins (E and G), mRNAs (F) and ATP contents (F) in fresh tumor tissue lysates were shown (E). *P < 0.05 vs. "shC (aav)" group.
Recent studies have proposed p38γ as a non-classical CDK-like kinase, phosphorylating and inhibiting the tumor suppressor protein Rb to promote the expression of cyclin A and cyclin E1. This will lead to cell cycle progression and cancer cell growth [14,20]. In line with these findings, we found that Rb phosphorylation as well as cyclin E1/A expression were significantly decreased in p38γ-silenced or p38γ-KO NPC cells, but increased after p38γ overexpression. Moreover, Rb phosphorylation and cyclin E1/A expression were decreased in pNPC-1 xenograft tumors with p38γ shRNA aav injection. The nonphosphorylated Rb-mut blocked Rb phosphorylation and decreased cyclin E1/A expression in OE-p38γ-L1 CNE-1 cells. The p38γ specific inhibitor PFD inhibited Rb phosphorylation and cyclins expression in NPC cells. Thus, p38γ-driven NPC progression is, at least in part, due to regulating Rb-cyclin cascade.
Increased mitochondrial function is essential for the progression of NPC [35][36][37]. Several key mitochondrial components were upregulated and/or hyper-activated in NPC, associated with tumorigenesis and cancer progression [35][36][37]. Conversely, mitochondria damage or dysregulation can induce death of NPC cells [35][36][37]. One important finding of this study is that p38γ localizes in mitochondria in NPC cells. p38γ was important for mitochondrial functions. On the contrary, p38γ shRNA or KO disrupted mitochondrial functions, causing mitochondrial depolarization, ROS production, oxidative injury and ATP depletion in NPC cells. In vivo, ATP depletion was detected in p38γ shRNA aavinjected pNPC-1 xenograft tissues. Therefore, maintaining mitochondrial function could be another mechanism of p38γ-driven NPC progression.

CONCLUSION
The current clinical treatments for the advanced NPC is still challenging and it is therefore extremely important to uncover the novel therapeutic targets for NPC. The results of this study suggest that p38γ is a key oncogenic gene and an important therapeutic target of NPC.

DATA AVAILABILITY
All data are available upon request.