miR-22/KAT6B axis is a chemotherapeutic determiner via regulation of PI3k-Akt-NF-kB pathway in tongue squamous cell carcinoma

Human tongue cancer cell lines (CAL27 and SCC9), HCT 116 were obtained from and maintained as recommended by the American Type Culture Collection (ATCC, Manassas, VA, USA). HCT 116 p53^−/− was saved by our laboratory. Above cell lines were maintained in RPMI-1640 medium (Hyclone, Logan, UT, USA), supplemented with 10% fetal bovine serum (Hyclone) at 37 °C in a humidified atmosphere containing 5% CO₂. The human embryonic kidney cell line (HEK-293 T) was cultured in DMEM (Gibco) with 10% fetal bovine serum. CDDP was purchased from Sigma-Aldrich (Sigma-Aldrich, US).

All tongue cancer tissue specimens were collected via surgical resection from patients diagnosed between March 2007 and March 2013 at the Affiliated Tumor Hospital of Guangzhou Medical University (Guangzhou, Guangdong, China). Written informed consent was obtained from all study participants. The study protocol was approved by the Ethics Committee of Guangzhou Medical University. All eligible patients received three cycles of weekly CDDP-based regimen. Patients proceeded to surgery within 4 weeks of the last dose of chemotherapy. PCR was defined as the absence of invasive tumor cells in the final surgical tongue and ambient lymph node samples. Tumor tissue was obtained by a core biopsy prior to treatment and immediately stored at − 80 °C. Total RNA was extracted with Trizol reagent (Invitrogen, Carlsbad, CA, USA) following the manufacturer’s instructions. Overall survival was computed from the day of surgery to the day of death or of last follow-up.

miRNAs from cultured cells were isolated and purified with the miRNA isolation system (Exiqon). cDNA was generated with the miScript II RT Kit (QIAGEN, Hilden, Germany), and quantitative real-time PCR (qRT-PCR) was performed by using the miScript SYBR Green PCR Kit (QIAGEN) following the manufacturer’s instructions. The miRNA sequence-specific RT-PCR primers and the endogenous control RNU6 were purchased from QIAGEN. The relative quantitative expression was calculated by normalizing the results with RNU6. The total RNA was extracted according to the Trizol protocol, and cDNAs from the mRNAs were synthesized with the first-strand synthesis system (Fermentas Life Science). The primers for C17orf91 and the selected mRNAs are shown in Table 1. Real-time PCR was carried out according to the standard protocol using an ABI 7500fast with SYBR Green detection (Fermentas SYBR green supermix). GAPDH was used as an internal control, and the qRT-PCR was repeated three times.

The miR-22 expression plasmid pSilencer2.1-U6-miR-22 and an empty control vector were gifts from Dr. Xiuwu Zhang (Duke University Medical School, Durham, USA). The cells were transfected with these plasmids using Lipofectamine 2000 (Invitrogen) in accordance with the manufacturer’s instructions in 24-well plates. After 24 h, the cells were selected with 4 μg/ml puromycin for 2 weeks, and the individual stable clones were analyzed by qRT-PCR. Tongue cancer cells were transfected with pRNAT-U6.1/sh-KAT6B plasmids (Genscript, Nanjing, China) as described above, selected with 800 μg/ml G418 and validated by western blot. The anti-miR-22 oligonucleotides (QIAGEN) and the pBabe-IκBα or pEGFP-N1-WTp53 plasmids were transfected in 10-mm dishes with Lipofectamine 2000. The cells were harvested 72 h later, and the following experiments were performed.

The Cell Titer 96® AQueous One Solution Cell Proliferation Assay kit (Promega, Madison, WI, USA) was used to determine the sensitivity of the cells to CDDP. In brief, the cells were seeded in 96-well plates at a density of 4 × 103 cells/well (0.20 ml/well) for 24 h before use. The culture medium was replaced with fresh medium containing CDDP with different concentrations for 72 h. Then, MTS (0.02 ml/well) was added. After 2 h of further incubation, the absorbance at 490 nm of each well was recorded on the Biotex ELX800. The growth rate was calculated as the ratio of the absorbance of the experimental well to that of the control well. The IC50 (the concentration of drug that results in 50% of control value) was also calculated.

The cells (1 × 10⁶) were digested with a trypsin solution, then harvested and washed twice with cold PBS. The washed cells were rinsed twice with PBS, and re-suspended in a propidium iodine solution comprising containing 40 μg/mL propidium iodine and 100 lg/ mL RNaseA (Sigma-Aldrich) in PBS without calcium and magnesium, then incubated at 37 °C for 30 min in the dark. The stained cells were passed through a nylon mesh sieve to remove cell clumps, and then analyzed with FACScan flow cytometer and the CELL QUEST analysis software (Becton Dickinson, San Jose, CA, USA). The flow cytometry analysis was repeated three times.

The total proteins were extracted from the corresponding cells, loaded and separated on 10% SDS-PAGE and then transferred to PVDF membranes (Millipore, Billerica, MA, USA). The primary antibodies used for the analysis included mouse anti-KAT6B monoclonal Ab (Abnova, Taiwan, China); rabbit anti-phospho-PI3K p85 (Tyr458) polyclonal Ab, mouse anti-phospho-Akt (Thr308) monoclonal Ab, rabbit anti-PTEN monoclonal Ab and mouse anti-κBα (Ser32/36) monoclonal Ab (Cell Signaling Technology, Danvers, MA, USA); mouse anti-tubulin monoclonal Ab (Santa Cruz, Dallas, Texas, USA).

Two single strands of the wild type 3’UTR with the miR-22 binding site and two single strands of the mutant type with seven bases deleted in the miR-22 binding site (as a mutant control) of KAT6B were synthesized with restriction sites for SpeI and HindIII located at both ends of the oligonucleotides for further cloning. The single-strand DNA sequences were as follows: wild type 3’UTR of KAT6B (sense: 5’-CTAGTCAGATTTCTTTGGGGAAAAAAGGCAGCTTTCTGTTTTATAAATGCAGACTTCTGA-3′; antisense: 5’-AGCTTCAGAAGTCTGCATTTATAAAACAGAAAGCTGCCTTTTTTCCCCAAAGAAATCTGA-3′) and the mutated type 3’UTR of KAT6B (sense: 5’-CTAGTCAGATTTCTTTGGGGAAAAAA-------TTCTGTTTTATAAATGCAGACTTCTGA-3′; antisense: 5’-AGCTTCAGAAGTCTGCATTTATAAAACAGAA-------TTTTTTCCCCAAAGAAATCTGA-3′). The corresponding sense and antisense strands were annealed and subsequently cloned into the pMir-Report plasmid downstream of the firefly luciferase reporter gene. The cells were seeded in 96 well-plates and co-transfected with the pMir-Report luciferase vector, the pRL-TK Renilla luciferase vector and the miR-22 expression vector. After 48 h, the luciferase activities were determined using a Dual-Luciferase Reporter Assay System (Promega) where the Renilla luciferase activity was used as an internal control and the firefly luciferase activity was calculated as the mean ± SD after being normalized relative to the Renilla luciferase activity. To identify the promoter of miR-22, a 2-kb region upstream of the miR-22 precursor starting site was cloned into the pGL4-reporter vector upstream of the luciferase gene and the reporter vector with site B binding site deletion as mutant was also constructed, and the reporter assay was performed as described above.

The cells were seeded in 96-well plates and co-transfected with the NF-κB reporter plasmid or the corresponding control, the pRL-TK Renilla luciferase plasmid and/or the pBabe-IκBα plasmid. Then, the luciferase activities were determined using the Dual-Luciferase Reporter Assay System (Promega).

Caspase 3 activity was determined with a caspase 3 activity kit (Beyotime, China) through the cleavage of a colorless substrate specific for caspase 3 [Ac-DEVD-p-nitroaniline (pNA)], releasing the chromophore pNA. The assays were carried out according to the manufacturer’s instructions39.

BALB/c-nude mice (female, 3–4 weeks of age, 18-20 g) were purchased from the Center of Experimental Animal of Guangdong province. All experimental procedures were approved by the Institutional Animal Care and Use Committee of Guangzhou Medical University. The BALB/c nude mice were randomly divided into 2 groups (each group contain 10 mice). CAL27 stably expressing miR-22 or the control cells (5 × 10⁶/site) were inoculated subcutaneously in the dorsal flanks of nude mice and developed into solid tumors in 7–10 days after injection. When the maximum tumor diameter exceeds 6um, each group were respectively treatment with 2 mg/kg cisplatin or equal volumes saline by intraperitoneally injection every other day for 4 weeks.

In addition, an in vivo loss-of-function study by intravenously administering miR-22 siRNA to xenograft tumor mouse model was performed. CAL27 cells (5 × 10⁶/site) were subcutaneously injected of into immunodeficient mice. All mice inoculated with CAL27 cells developed primary tumors, which were readily visualized and the maximum tumor diameter exceeds 6um in 8–10 days. We treated these mice intravenously with in vivo -jetPEI-formulated mir-22 inhibitor for 4 weeks. Separate groups of animals bearing CAL27 xenograft tumors were treated with the same doses of in vivo -jetPEI-formulated tRNA/MSA as controls (tRNA/mir-22 or control tRNA scaffold were conducted as described recently [25‐27]). Meanwhile, 2 mg/kg cisplatin or equal volumes saline was intraperitoneally injected every other day for 4 weeks. Tumor sizes were measured with caliper every 3 days and calculated by the formula V = (L × W²)/2. Thirty days after tumor implantation, the mice were sacrificed; the subcutaneous tumors were removed and weighed. Tumors were fixed in formalin and embedded in paraffin using the routine method. Serial 6.0 μm sections were cut and subjected to H&E stained with Mayer’s hematoxylin solution.

The ChIP assay was performed using the EZ-CHIPTM chromatin immunoprecipitation kit (Millipore). In brief, chromatin proteins were crosslinked to the DNA by addition of formaldehyde to the culture medium to a final concentration of 1%. After 10 min of incubation at room temperature, the cells were washed and scraped off in ice-cold PBS containing a Protease Inhibitor Cocktail II. The cells were pelleted and then resuspended in Lysis Buffer containing Protease Inhibitor Cocktail II. The resulting lysate was subjected to sonication to reduce the size of DNA to 200–1000 bp. The sample was centrifuged to remove the cell debris and diluted 10-fold in ChIP dilution buffer containing Protease Inhibitor Cocktail II. An aliquot of 5 μl of the supernatant was removed as the “Input” and saved at 4 °C. Then, 5 μg of the anti-RNA Polymerase antibody (as positive control), the normal rabbit IgG (as a negative control) (provided by kit) or the anti-p53 antibody were added to the chromatin solution and incubated overnight at 4 °C with rotation. After antibody incubation, the protein G agarose was added and incubated at 4 °C with rotation for 2 h. The protein/DNA complexes were washed with Wash Buffers four times and eluted with ChIP Elution Buffer, and the cross-links were reversed to free the DNA by the addition of 5 M NaCl and incubation at 65 °C for 4 h. The DNA was purified according to the manufacturer’s instructions, and 50 μl of the DNA from each treatment was obtained. A volume of 0.2 μl of DNA from each group was used as a template for PCR. The primers for the miR-22 promoter with p53 binding sites were as follows: Sense: 5’-GCGGTGCCGGGGCCTTAT-3′, Antisense: 5’-GGTGCAGAGGTGACCTTCTCTC-3′ (for site A, product: 148 bp); sense: 5’-CAGGGGAAGGAAGATACACAAAGT-3′, antisense: 5’-GTTCCAACGCATGAATCAGCAG-3′, (for site B, product: 173 bp); sense: 5’ ACTGAGTGTCAGCACGGACCC-3′, Antisense: 5’-TTCCAAGTTCCAAGCCCCG-3′(for site C, product: 149 bp). The primers for the positive and negative controls or “no DNA” were specific for the human GAPDH gene: Sense, 5’-TACTAGCGGTTTTACGGGCG-3′, Antisense, 5’-TCGAACAGGAGGAGCAGAGAGCGA-3′, (product: 166 bp). The PCR conditions were as follows: 1 cycle of 95 °C for 5 min; 32 cycles of 95 °C for 20 s, 59 °C for 30 s, and 72 °C 30 s; and 1 cycle of 72 °C for 10 min. The PCR samples were resolved by electrophoresis in a 2% agarose gel and stained with ethidium bromide.

IHC staining was performed according to the protocol previously described [28]. In brief, the tissue sections were deparaffinized with dimethylbenzene and then rehydrated via a graded alcohol series. Endogenous peroxidase activity was blocked with 0.3% hydrogen peroxide for 15 min. The slides were boiled in tris(hydroxymethyl) aminomethane-EDTA buffer (pH 8.0) in a microwave for 30 min to retrieve antigen. Nonspecific antigens were blocked with 10% normal goat serum for 20 min. Then, the slides were incubated with first antibody (anti-KAT6B monoclonal Ab (Abnova, Taiwan, China, 1:200 dilution); mouse anti-phospho-Akt (Thr308) monoclonal Ab, 1:200 dilution, Abnova) overnight at 4 °C in a moist chamber. The controls were treated by replacing the primary antibody with normal goat serum. On the second day, the slides were sequentially incubated with biotinylated rabbit anti-mouse antibody, streptavidin-peroxidase conjugate and 3′-3’diaminobenzidine. The immunoreactivity scoring criteria are as described previously [28]. The staining result for each section was the average scores decided by two pathologists.

To validate the association between miR-22 and response to CDDP based chemotherapy regimen, an independent set of 56 patients (28 pCR patients and 28 non-pCR patients) was analyzed by using quantitative RT-PCR. The clinicopathological features of this validation set are listed in Table 2. Consistently, we found the expression level of miR-22 was significantly higher in tumors from pCR patients than that in non-pCR patients (p = 0.015) (Fig. 1a), confirming that the higher expression of miR-22 is associated with a better response to CDDP based chemotherapy regimen. Furthermore, analyzing the disease-free survival (DFS) of these 56 tongue cancer patients after stratification by the level of miR-22 revealed that patients with higher miR-22 expression level have significantly longer DFS (P < 0.005) (Fig. 1b). To evaluate this predictive value, we used the receiver operating characteristics (ROC) curve to analyze the sensitivity and specificity of miR-22. The area under the curve (AUC) was 0.827 (confidence interval = 0.733–0.926; P < 0.0001), which indicated high accuracy of predictive value. The sensitivity and specificity were 88.3 and 65.6%, respectively (Fig. 1c). These findings suggest that increased miR-22 expression may be responsible for high sensitivity to CDDP based chemotherapy and predict pCR in tongue cancer patients.

To investigate whether miR-22 has a direct function in the response to chemotherapy in TSCC cells, we used a gain- or loss-of-function approach in SCC9 and CAL27 cells. As shown in Fig. 2a, stable miR-22 overexpressing cells were established by extraneous transfection with pSilencer2.1-U6-miR-22 plasmid. MiR-22 overexpression significantly enhanced the sensitivity of SCC9 and CAL27 cells to cDDP (Fig. 2b). Consistently, modulating miR-22 level by transfecting miR-22 inhibitor enhanced TSCC cells survival upon cDDP treatment (Fig. 2b). These results collectively support that increasing expression of miR-22 might enhance drug sensitivity to cDDP in tongue cancer cells. Furthermore, Overexpression of miR-22 in SCC9 and CAL27 cells markedly increased the cleavage of Caspase 3 and PARP1 in response to cDDP treatment, suggesting an enhanced apoptosis induced miR-22 upon drug treatment (Fig. 2c and d). We next measured the effect of miR-22 on apoptosis using propidium iodide and Annexin V apoptosis assay. The data indicated miR-22 transfection resulted in a significant increase in the percentages of cells undergoing apoptosis; whereas miR-22 inhibition noticeably prevented apoptosis of CAL27 and SCC9 cells (Fig. 2e). In addition, we validated whether miR-22 could play a role in tumor sensitivity to cDDP in vivo. MiR-22 overexpressing CAL27 cells and related control cells were transplanted into nude mice subcutaneously and developed into solid tumors in 7–10 days, and then 2 mg/kg cisplatin or equal volume saline were intraperitoneally injected every other day for 4 weeks. The tumor growth curves and tumor inhibited rate histogram were mapped. As shown in Fig. 2f and g, the tumor volume of miR-22-overexpressing xenografts decreased to a greater extent than that of control xenografts upon cDDP treatment, indicating a chemo-sensitizing effect of miR-22. Meanwhile, we performed an in vivo loss-of-function study by intravenously administering miR-22 siRNA to xenograft tumor mouse model through subcutaneous injection of CAL27 cells into immunodeficient mice, and evaluated the effectiveness of miR-22 inhibitor to sensitivity to cDDP in vivo. We treated these mice intravenously with in vivo -jetPEI-formulated tRNA/mir-22 accompanied with 2 mg/kg CDDP or equal volume saline for 4 weeks. Compared with the same doses vehicle, tRNA/mir-22 treatment showed a remarkable suppression of the chemosensitivity of xenografts tumors (P < 0.01, Fig. 2h, i). These results indicate that administration of biological miR-22 is effective to enhance tongue cancer chemotherapy sensitivity in vivo.

To elucidate the underlying mechanisms promoting chemosensitivity in human tongue cancer cells by miR-22, we used the public database TargetScan (http://​www.​targetscan.​org) to predict potential targets of miR-22, and KAT6B (MYST4) with a critically conserved binding site was selected for further identification (Fig. 3a). Notably, the endogenous overexpression of miR-22 in CAL27 and SCC9 cells was accompanied with decreased expression of KAT6B mRNA and protein (Fig. 3b). In contrast, depression miR-22 by transfecting miR-22 inhibitor increased the expression of KAT6B. To assess whether KAT6B is a direct target of miR-22, a luciferase reporter vector with a putative KAT6B 3’UTR target site for miR-22 downstream of the luciferase gene (pMir-KAT6B-Wt) and a mutant version with a deletion of 7 bp in the seed region was constructed (pMir-KAT6B-Mut). As shown in Fig. 3c, d, the luciferase reporter assay performed in CAL27 and HEK293T cells showed that miR-22 reduced the luciferase activity of the vector with the wild-type KAT6B 3’UTR, but the mutant version abrogated the repressive ability of miR-22. These results strongly demonstrated the specificity of miR-22 targeting KAT6B. To investigate the function of KAT6B, CAL27 cells were transfected with pRNAT-U6.1/sh-KAT6B to knockdown KAT6B, and sh-1# markedly decreased KAT6B protein expression (Fig. 3e). Therefore, sh-1# was selected for further study. A negative relationship between KAT6B and the response to CDDP was also observed (Fig. 3f). In addition, sh-1# promoted the effect of 10 mg/L CDDP on caspase3 activation in CAL27 cells. As shown, caspase3 cleavage was detected (Fig. 3g) and caspase3 activity increased (Fig. 3h).

Given the regulatory relationship between miR-22 and KAT6B identified in our cell culture studies, we focused whether the miR-22/KAT6B might associate with the prognosis of tongue cancer patients. The expression levels of miR-22 and KAT6B were evaluated in clinical samples of tongue cancer patients by Q-PCR. We found that the expression of KAT6B in pCR patient tumors was significantly lower than that in non-pCR patients (P < 0.05, Fig. 4a). Moreover, expression of KAT6B presented a negative correlation with miR-22 level in these 56 patients (r = − 0.626, P < 0.0001) (Fig. 4b). Additionally, we found high KAT6B level was associated with poor DFS in these 56 patients (Fig. 4c). Collectively, these data indicate that the chemo-sensitizing effect of miR-22 in tongue cancer may depend on KAT6B downregulation.

To explore the signal transduction pathway responsible for the up-regulation of chemosensitivity promoter-miR-22 in CAL27 cells, the activity of the predominant oncogenic pathway, PI3K/Akt/NF-κB, was examined. The results shown that both miR-22 overexpression and KAT6B knockdown inhibited PI3K/Akt/NF-κB activation, respectively (Fig. 5a, b), with no difference in PTEN expression (data not shown). Whereas, miR-22 inhibition further promoted PI3K/Akt/NF-κB activity (Fig. 5c). These data implied miR-22 enhanced chemosensitivity to cDDP by inhibiting PI3K/Akt/NF-κB activity in tongue cancer cells. To observe the exact role of PI3K/Akt/NF-κB and miR-22 in drug resistance, a series of treatments was applied in CAL27 cells and the drug sensitivity was measured. Treatment with 10 mg/L CDDP, 40 μM LY294004, 10 μM Bay11–7082 and IκBα expression significantly enhanced the effect of CDDP. The effect of PI3K/Akt/NF-κB inhibition combined with miR-22 overexpression was much stronger than either alone (Fig. 5d, e). To address how miR-22 or KAT6B regulated PI3K/Akt/NF-κB activity, we sought to examine the expression of some cytokines in miR-22 overexpression or KAT6B knockdown in CAL27 cells. As shown in Fig. 5f and g, S100A8, PDGF and VEGF were significantly downregulated in forementioned cell lines. Interestingly, these secretory cytokines had been proved to activate PI3K/Akt / NF-κB pathway in previous researches [29‐31]. To confirm whether miR-22/KAT6B potentially activate PI3K/Akt/NF-κB signal pathway through these secretory cytokines, indicated concentration cytokines S100A8, PDGF and VEGF were mixed as a cocktail and added into cell culture medium, then the NF-κB activity was evaluated. The data indicated that mixed cytokines attenuated inhibition of NF-kB activity in miR-22-overexpressing or shKAT6b tongue cancer cells (Fig. 5h). These results suggested that S100A8, PDGF and VEGF may be the targets or the downstream of KAT6B, which potentially activate PI3K/Akt/NF-κB signal pathway. In addition, to further verify that miR-22/KAT6B functions in the chemotherapy resistance through the AKT-NF-κB pathway, we selected the heterotopic tumor tissue obtained from the previous animal experiment, which were fixed by formaldehyde, paraffin embedded, continuous sliced and tested for the immunization of KAT6B and AKT signaling molecules. The results indicated that miR-22 does inhibit the expression of KAT6B and the phosphorylation of AKT. On the contrary, reduced miR-22 can activate the AKT signaling pathway, these data were shown in Fig. 5i.

Due to the drug resistance acquired through exposure to treatment stress, to investigate the expression pattern of miR-22 on stress, the kinetics of miR-22 expression were examined when treating CAL27 cells with different doses of CDDP. There was no significant cell death observed in CAL27 cells treated with 0.5 mg/L CDDP, and the cell survival promoting pathway was triggered. As shown in Fig. 6a, the p-PI3K, p-Akt and NF-κB targets Bcl-2 and Survivin increased and the IκBα protein level decreased in a time-dependent manner, without significant a change in miR-22 and C17orf91 expression at an early state (Fig. 6a). The miR-22 primary transcript is located in the 5′-untranslated region of C17orf91 and shares the same promoter/enhancer [32]. Therefore, we also measured the change in expression of C17orf91 under the conditions described above. In contrast, upon exposure to lethal-dose 10 mg/L CDDP, the p-PI3K, p-Akt and NF-κB targets Bcl-2 and Survivin decreased. Whereas, IκBα expression were increased (Fig. 6b). These data indicated different intensities of stress trigger different dominant signaling pathways to determine different cell fates. Due to overexpression of miR-22 in CAL27 and SCC9 cells markedly increased the cleavage of Caspase 3 and PARP1 in response to CDDP treatment, suggesting an enhanced apoptosis upon drug treatment (Fig. 2c and d). The tumor suppressor p53 activates a key function that directs cells to undergo apoptosis or senescence in response to cellular stress [33, 34]. Therefore, we asked whether p53 orchestrated chemosensitivity to CDDP with miR-22/KAT6B in tongue cancer cells. Interestingly, we found that the expression of p53 is significantly increased upon higher doses but not low doses CDDP treatment, accompanied by increased expression of miR-22 and C17orf91 (Fig. 6a, b). To further address whether unregulated miR-22 depends on p53 activation in CAL27 cells, we performed the transient transfection of CAL27 cells with pEGFP-N1-WTP5 and the subsequent p53 protein increase was accompanied by an increase in miR-22 and C17orf91 expression (Fig. 6c, d). In addition, to validate the direct association of p53 with the miR-22 promoter, we performed a ChIP assay in CAL27 cells for all the putative p53binding sites within 2-kb region. The ChIP results revealed that p53 most significantly bound to site B within the miR-22 promoter (Fig. 6e). The luciferase expression driven by the miR-22 promoter with site B was significantly higher than site mutant in HEK-293 T cells (Fig. 6f). To further confirm whether p53 can enhance miR-22 activity, we examined HCT116 WT and HCT116 p53^−/− cells, and treated these cells with CDDP. Expression of miR-22 and C17orf91were significantly activated in HCT116 WT cells, but not in p53^−/− cells, upon CDDP treatment (Fig. 6g and h). These results suggested the pattern of miR-22 expression is depended on the intensity of the stresses in the presence of p53 activation.