Aurora-A down-regulates IkappaBα via Akt activation and interacts with insulin-like growth factor-1 induced phosphatidylinositol 3-kinase pathway for cancer cell survival

We used the immunohistochemical analysis to investigate Aur-A expression in primary tumor tissues. Results showed that only a few (7/30, 23.3%) matched adjacent normal tissues displayed Aur-A positive staining (Fig. 1a). However, Aur-A was significantly elevated in majority (36/55, 65.5%) of pathologically confirmed tumor specimens (Fig. 1b). Aur-A was uniformly cytoplasmic positive staining, uncoupled with its normal mitosis-related expression pattern.

We further analyzed the relationship between Aur-A expression and clinical characteristics (Table 1). Aur-A was more frequently expressed in high-grade tumors (stage III and IV, 77.8%) compared with low-grade (stage I and II, 42.1%) tumors (p = 0.008). Moreover, we observed preferential expression of Aur-A in tumor with positively (87.5%) versus negatively (56.4%) lymph node metastasized samples (p = 0.028). No significant correlation was found between Aur-A expression and other clinical characteristics including age, gender and differentiation status. Thus, the potential association between tumor overexpression of Aur-A and clinic stage or lymph node metastasis raises the possibility of specific inhibition of Aurora kinase in treatment of tongue cancer cells.

To evaluate the inhibition of Aurora kinase in TSCC cells, we used a small molecule inhibitor VX-680. Figure 2a showed that the percentage of abnormal spindle as was markedly increased in VX-680 treated mitotic cells (22.39 ± 0.98%) compared to the control mitotic cells (4.21 ± 0.91%). The abnormal spindle characterized as mono-polarity consistent with a known Aur-A inhibition phenotype [19]. Phosphorylation inhibition of histone H3 at Ser10, an in vivo substrate of Aur-B was significantly reduced in Tca8113 cells treated with VX-680 at 1 nM (12.00 ± 3.06) or 5 nM (5.80 ± 0.08), compared to the control cells (30.20 ± 8.62, Fig. 2b).

Cell survival rates were reduced by VX-680 in a dose-dependent manner as assessed by MTT assay with IC₅₀ of 6.45 ± 1.14 nM (Fig. 2c). Annexin V assay revealed that VX-680 induced apoptosis even at 1 nM as showed in Annexin V and PI staining positive (Fig. 2d). Western blot assay showed that VX-680 reduced the expression of anti-apoptotic protein Bcl-2 and increased the level of both cleaved PARP and cleaved caspase-3 in a dose-dependent manner (Fig. 2e). Caspase-3 inhibitor however reversed Bcl-2 reduction and PARP cleavage in response to VX-680 (data not shown).

Using a serum-free system, we examined cell apoptosis by Western blot and flow cytometry assay. IGF-1 increased the phosphorylation of Akt at Ser473 and its downstream target GSK3 at Ser 21/9. Expression of IκBα was however decreased by IGF-1 treatment (Fig. 3a), which also prevented VX-680 (5 nM)-induced apoptosis (Fig. 3b). Interestingly, VX-680 and an irreversible PI3K inhibitor wortmannin in combination displayed a dramatic effect in inhibiting Akt and GSK3 activity, elevating IκBα expression (Fig. 3a), and increasing cell apoptosis, compared with either VX-680 (about 7.90 ± 2.54-fold) or wortmannin (about 18.49 ± 2.88-fold) alone (Fig. 3b). We calculated the cooperative coefficient of VX-680 and wortmannin was 6.09 ± 0.35, suggesting wortmannin synergized VX-680 mediated apoptosis by inhibiting PI3K. Meanwhile, elevated levels of cleaved PARP and cleaved caspase-3 and reduction of Bcl-2 expression were observed in cells treated with VX-680 and/or wortmannin (Fig. 3a). These data together indicated that there was an intracellular cross-talk between Aurora kinases and PI3K pathway in regulating cancer cell survival. We conducted Western blot with another squamous carcinoma KB cells and observed similar results (Additional file 1).

We have showed that overexpression of Aur-A was positively correlated with lymph node metastasis (Table 1), and cell migration was closely associated with potential of tumor invasiveness and metastasis. We showed that VX-680 potently induced a dose-dependent inhibition in the migration of Tca8113 cells (Additional file 2). Similar inhibition of cell motility was also induced by Akt/protein kinase B signaling inhibitor-2 (API-2) at dose of 1 μM.

We then conducted the transwell migration assay in serum-free condition. Compared with the control cells, IGF-1 significantly enhanced migration of Tca8113 cells (about 3.5-fold), while either VX-680 or wortmannin alone at low dose could partially reduce the cell mobility induced by IGF-1 (Fig. 4). Moreover, the combination of VX-680 and wortmannin efficiently abrogated IGF-1 induced cell migration in a synergic manner. Meanwhile we performed MTT assay to detect the cell viability in the same system. These results showed that the suppression of migration by VX-680 and/or wortmannin were not due to inducing apoptosis in Tca8113 cells (data not shown). Thus, these data indicated the interaction between Aurora kinases and PI3K pathway also played a key role in cancer cell migration.

Based on above findings, we hypothesized that Aur-A and PI3K pathway might interact at Akt. The level of pAkt was decreased in cells treated with increasing concentration of VX-680 (Fig. 5a). We further overexpressed a constitutively active form of Akt (Myr-Akt1) in Tca8113 cells (Fig. 5b). MTT assay showed that the survival rate of Myr-Akt1 transfected cells was (46.43 ± 7.95% and 38.11 ± 6.16%), obviously higher than that of empty vector pUSE transfected cells (31.5 ± 1.67% and 18.93 ± 2.90%) when treated with VX-680 at 5 nM and 10 nM respectively (Fig. 5c). We performed Aur-A RNAi in vector or Myr-Akt1 transfected cells and observed similar results (Additional file 3). Together, these data suggested that Akt was a potential downstream target of Aurora kinases in enhancing cancer cell survival.

A recent study reported that Aur-A regulated NF-κB via phosphrylation of IκBα [14]. We further studied whether Aur-A regulated IκBα and its downstream targets via Akt pathway. Decreased pAkt and elevated IκBα were detected when cells were transfected with siRNA toward either Akt (Additional file 4) or Aur-A (Fig. 6a), compared with cells transfected with their scramble control respectively. Inhibition of Aur-A chemically also up-regulated IκBα level (Fig. 6b). Conversely, overexpression of Aur-A increased Akt activity and decreased IκBα level compared with the vector control (Fig. 6a). We then analyzed the expression of Bcl-xL, which is known as a NF-κB target gene closely associated with cell proliferation and apoptosis. Bcl-xL was down-regulated in Aur-A and Akt depleted cells (Fig. 6a). Immunofluorescence staining of NF-κB p65 showed that Aur-A overexpression was significantly associated with p65 nuclear translocation whereas p65 was mainly expressed in the cytoplasm in cells transfected with empty vector pCS2+ (Fig. 6c). We further showed that inhibition of PI3K with wortmannin did not prevent either an increase of pAkt and Bcl-xL or a decrease in IκBα caused by Aur-A overexpression (Fig. 6d). Interestingly, in cells incubated with Akt inhibitor API-2 or siRNA against Akt, overexpression of Aur-A however failed to reduce IκBα or raise Bcl-xL expression in comparison to the vector control (Fig. 6e and 6f). This suggested that Akt, but not PI3K, was involved in the down-regulation of IκBα by Aur-A. These results revealed that Aur-A, via its downstream target Akt, down-regulated IκBα, which then led to NF-κB nuclear translocation and subsequently activating NF-κB target gene Bcl-xL in enhancing cancer cell survival (Fig. 6g).

Fifty-five patients who performed radical surgery were original clinically diagnosed and pathologically confirmed of TSCC between 1987 and 1992. Pertinent patient clinical reports were obtained with prior patient consent and the approval of the institutional Clinical Ethics Review Board. All of the 55 specimens and additional 30 normal adjacent tissues were collected and fixed in formalin and embedded in paraffin in the diagnostic histopathology laboratory at the Second Affiliated Hospital of Sun Yat-sen University. Patient clinic pathological features were shown in Table 1. Tumors were staged according to UICC classification (1997): stage I (4 cases), II (15 cases), III (23 cases) and IV (13 cases) or histopathology classification: stage I (7 cases) stage II (26 cases) and stage III (22 cases).

VX-680 was purchased from Kava Technology, San Diego, CA., API-2 was from Calbiochem, IGF-1 from Biosource, tumor necrosis factor α (TNF-α) and wortmannin from Cell Signaling. Human tongue squamous cancer cell line Tca8113 was kindly provided by Xiao-feng Zhu (Cancer Center, Sun Yat-sen University), human oral floor cancer cell line KB was obtained from ATCC.

Aur-A immunohistostaining using an anti-Aur-A antibody (Upstate) on tongue cancer tissues was performed as previously described [8]. Moderate or strong cytoplasm staining, considered as positive reaction, was assessed semi-quantitatively by at least two independent pathologists. Specimen was determined as positive staining for Aur-A when >30% cells showed visible brown granules in the cytoplasm.

Cultured cells grown on coverslips treated with DMSO or VX-680, or transiently transfected with plasmid expressing Aur-A or empty vector pCS2+. Immunofluorescence staining of cells was performed as described [34] and analyzed with an Olympus BX51 microscope. For immunofluorescence staining of NF-κB p65, cells were treated with 50 ng/ml of TNF-α for 10 min prior to fixing as a positive control.

Tca8113 cells were incubated in 96-well plate and maintained at different doses of VX-680 for 48 h. Myr-Akt or pUSE transfected Tca8113 cells were maintained at different doses of VX-680 for 24 h. Cell survival was assessed as described previously [35].

Cells were incubated in serum-free media with indicated drugs for 12 h and subjected to flow cytometry analysis as previously described [34].

Cells were treated with DMSO or VX-680 for 48 h prior to collecting and resuspending in binding buffer. Annexin V-FITC and propidium iodide (Annexin V-FITC Apoptosis Detection Kit, Merck) were added to each sample according to the manufacturer's protocol. 4, 6-diamidino-2-phenylindole (DAPI 1 μg/ml) was used to visualize nuclei. 20~25 μl of cell suspension was transfered onto glass microscope slides respectively, and viewed immediately using a fluorescence microscope (Olympus BX51).

Western blot assay was performed as described previously [8]. Antibodies used were mouse anti-GAPDH (Ambion), rabbit anti-Bcl-2, rabbit anti-cleaved caspase-3, mouse anti-cleaved PARP (Asp175), rabbit anti-phosphorylated Akt (pAkt, Ser473), mouse anti-phospho-GSK3α/β (Ser 21/9, Cell Signaling), mouse anti-IκBα (BD), rabbit anti-GSK3β, goat anti-Akt1, rabbit anti-Bcl-xL (Santa Cruz Biotechnology) and rabbit anti-Aur-A (Upstate).

Myr-Akt1 and pUSE plasmids were generously provided by Xiao-feng Zhu (Cancer Center, Sun Yat-sen University). Transfections were conducted according to manufacturers' recommendations (Invitrogen). Tca8113 cell clones stably transfected with plasmid were selected in 400 μg/ml G418.

Transient transfection of Aur-A and its vector control pCS2+ or cotransfection of Aur-A or pCS2+ with siRNA against Akt1 or its control were conducted according to manufacturers' recommendations (Invitrogen). Lysates were prepared 48 h after transfection. Cells were treated with API (10 μM) or wortmannin (1 μM) for 24 h prior to collecting for Western blot.

siRNA for downregulating Aur-A or Akt1 expression was done by the transfection of RNA oligonucleotides with lipofectamine 2000. The sequence for siRNA against Aur-A was AUGCCCUGUCUUACUGUCA and siRNA against Akt1 was AAGGAGGGUUGGCUGCACAAA. Lysates were prepared 36 h after transfection.

Transwell assay was performed as described previously [8]. Briefly, cells were incubated in serum or serum-free media containing desired drugs for 16 h. The migrated cells in five fields were counted, and the average of each chamber was determined.

Statistical analysis was performed using SPSS version 13.0 (SPSS Inc., Chicago, IL, USA). The χ² test and Student's t-test was used to make a statistical comparison between groups. P < 0.05 was considered statistically significant. We performed each study at least three times under identical conditions.