Aurora-A contributes to cisplatin resistance and lymphatic metastasis in non-small cell lung cancer and predicts poor prognosis

The present study included 102 patients with NSCLC, diagnosed at the Cancer Center, Sun Yat-sen University between 2002 and 2003. The patients, 70 males (69%) and 32 females (31%), ranged in age from 38 to 75 years (mean 60 years). Histological examination was performed on formalin-fixed tissues in all cases and tumors were diagnosed and classified according to the AJCC (American Joint Committee on Cancer guidelines) classification[22].

All the patients (from stage I to stage III) underwent radical surgery of primary tumor and lymph nodes, and received adjuvant cisplatin-based chemotherapy after surgery. No chemotherapy or radiotherapy was given to patients before surgery. Written informed consent for the use of the tissues was obtained from all patients before surgery, and the study was approved by the Institute Research Ethics Committee of Sun Yat-sen University.

The human lung adenocarcinoma cell lines A549 and A549/DDP (cisplatin-resistant variant) were graciously provided by Dr. Xiaofeng Zhu, Sun Yat-sen University[23]. The lung cancer cell lines NCI-H460 (H460) and H460/DDP (cisplatin-resistant variant) were graciously provided by Dr. Liwu Fu, Sun Yat-sen University[24]. Both cell lines were cultured in RPMI-1640 supplemented with 10% fetal bovine serum at 37°C in humidified 5% CO₂ incubator. The A549/DDP and H460/DDP cells were cultured with 6 μM cisplatin (Sigma–Aldrich, St.Louis, MO) to maintain drug resistance.

Cells were transfected with an empty vector pCS2+ or pCS2 + -Aurora-A (a gift from Joan Ruderman, Harvard Medical School, Boston, MA), or interfering RNA (siRNA) and negative control using Lipofectamine 2000 as described earlier[17]. The negative control (NC) siRNA and siRNA against Aurora-A (5′-AUGCCCUGUCUUACUGUCA-3′) were synthesized by GenePharma Company (Shanghai).

Plasmid pLVX-DsRed-N1-Monomer (Clontech) expressing Aurora-A was from Dr. Feimeng Zheng, Sun Yat-sen University. The shRNA sequences against GFP (Sence-5′-GCAAGCTGACCCTGAAGTTCAT, Antisense-5′-ATGAACTTCAGGGTCAGCTTGC) and Aurora-A (Sence-5′-CACATACCAAGAGACCTACAA, Antisense-5′-TTGTAGGTCTCTTGGTATGTG) were cloned into pLKO-Tet-On (Addgene). Lentiviruse was produced in 293T cells as described earlier[25].

The MTT assay was used to assess cell viability. 5,000 cells were plated in 96-well plates and then were exposed to various concentrations of cisplatin, VX-680 (Kava Technology), MLN8237 (Selleck Chemicals) or combinations. Following treatment, 0.5 mg/ml MTT (Sigma–Aldrich) solution was added to each well, and incubated for 4 hours (h). After the incubation, culture media was discarded followed by addition of 0.15 ml DMSO and vibration for 10 minutes (min). The absorbance at 570 nm was determined. The cell inhibition ratio was calculated as a fraction of the untreated controls.

5,000 cells were seeded into six-well plates in triplicate and incubated for 8–10 days. Colonies were stained with crystal violet and counted.

Cells were lysed with the RIPA buffer on ice before being subjected to western blot analysis. The protein concentration was detected by the Bradford method. Total proteins were fractionated using SDS-PAGE and transferred onto nitrocellulose membrane. The membrane was blocked and incubated with mouse anti-GAPDH antibody (Ambion Biotechnology), rabbit anti-Aurora-A antibody (Upstate), rabbit anti-p-Aurora-A antibody (Thr288, Cell Signaling), rabbit anti-histone H3 antibody (Epitomics), and rabbit anti-p-histone H3 antibody (Ser10, SAB).

Cells were fixed with 4% paraformaldehyde at room temperature for 15 min and permeabilized in 0.25% Triton X-100 in PBS for 10 min, and incubated in 3% BSA blocking solution for 30 min. The cells were incubated with anti-Aurora-A (Upstate) and α-tubulin (Sigma) antibodies for 1 h. The immune complexes were detected with goat anti-mouse conjugated to Alexa-488 and goat anti-rabbit conjugated to Alexa-546 secondary antibody (Molecular Probes). Nucleus was stained with DAPI.

Apoptosis analysis was conducted with an Annexin V-FITC Apoptosis Detection Kit (KeyGen Biotech) according to the manufacturer’s protocol. The percentage of apoptotic cells was determined using FACS flow cytometer equipped software (BECKMAN).

Cells transfected with RNAi or not were seeded in 24-well plates and grown until 70-80% confluence. The cells were then serum starved for 24 h. A linear wound was created using a pipette tip and observed and photographed at various times as indicated in the figure legends.

Formalin-fixed, paraffin-embedded samples were cut in 4 μm sections and mounted on slides. Slides were deparaffinized, rehydrated, and treated with 3% H₂O₂ in methanol for 15 min to inhibit endogenous peroxidase. Pretreatment was done in microwave with Tris/EDTA buffer solution (pH 8.0). After transfer to a humidified chamber, the slides were blocked with 10% normal goat serum at room temperature for 30 min and incubated with mouse anti-Aurora-A monoclonal antibody (Sigma, A1231, 1:200 dilution). Detection of the primary antibody was done using the Envision DAKO EnVision Detection kit, (3, 3-diaminobenzidine; code K 5007, DAKO Cytomation, Glostrup, Denmark) according to the protocol of the manufacturer. Finally, samples were counterstained with hematoxylin, dehydrated, and mounted. The positive control sample was a normal colonic mucosa section known to express Aurora-A. The negative control was obtained by replacing the primary antibody with a normal murine IgG.

Each case was rated according to a score that added a scale of intensity of staining to the area of staining. At least 10 high-power fields were chosen randomly, and >1,000 cells were counted for each section. The intensity of staining was graded on the following scale: 0, no staining; 1, weak; 2, moderate; 3, strong. The area of staining was evaluated as follows: 0, no staining of cells in any microscopic fields; 1+, <30% of tissue stained positive; 2+, between 30% and 60% stained positive; 3+, >60% stained positive[26]. The minimum score when summed (extension + intensity) was therefore, 0, and the maximum. Two independent pathologists (W.H.Z. and J.X.), blind to follow-up data, were responsible for IHC staining evaluation. A third pathologist arbitrated when any discrepancy arose between these two pathologists.

Receiver operating characteristic (ROC) curve analysis was carried out to select Aurora-A cutoff score[27, 28]. In brief, at each immunohistochemical score, the sensitivity and specificity for the outcome under study was plotted, generating an ROC curve. The score localized closest to the point at both maximizing sensitivity and specificity, the point (0.0, 1.0) on the curve, was selected as the cutoff score above which Aurora-A expression was considered high. To facilitate ROC curve analysis, the survival features were dichotomized: survival (death VS. others (censored, alive or death from other causes)).

All patients had follow-up records for over 5 years. After the completion of therapy, patients were observed at 3 month intervals during the first 3 years and at 6 month intervals thereafter. OS was defined as the time from diagnosis to the date of death or when censored at the latest date if patients were still alive. PFS was assessed from the first day of treatment to the earliest signs of disease progression as determined by CT or MRI imaging using RECIST (Response Evaluation Criteria In Solid Tumors) criteria, or death from any cause.

The specificity for antibody targeting Aurora-A was showed in Additional file1: Figure S1 by western blot analysis. The staining of Aurora-A was nearly negative (Figure 1A and A’) in adjacent non-neoplastic lung tissues, whereas NSCLC tissues showed elevated expression of Aurora-A (Figure 1B and C). Aurora-A staining located in both nucleus and cytoplasm, predominantly in nucleus (Figure 1B’ and C’). To select a relevant immunohistochemical cutoff score to describe Aurora-A overexpression in NSCLC, receiver operating characteristic (ROC) curve analysis was carried out. As shown in Additional file1: Figure S2, the Aurora-A cutoff scores for OS and PFS were 3.4 (p = 0.004) and 3.2 (p = 0.014) respectively. We thus divided the cohort into high (score ≥ 4) and low (score < 4) populations based on the cutoff points.

Clinicopathological features were listed in Table 1 in relation to Aurora-A expression status. High expression of Aurora-A was detected in 57 out of 102 (55.9%) selected NSCLC tissues and low in other 45 (44.1%) cases. In high-Aurora-A expression group, the incidence of cases in stage III was statistically significantly higher than in the low-Aurora-A expression group (77.2% vs. 51.1%, respectively; p = 0.006). Further correlation analysis demonstrated that high Aurora-A expression was significantly associated with clinical stage (p = 0.018), lymph node metastasis (p = 0.038) and recurrence (p = 0.005). Regarding other clinicopathological variables, there was no statistically significant correlation observed in age, gender, smoking history, carcino-embryonic antigen (CEA), histology, differentiation and tumor stage.

The Kaplan–Meier analysis indicated that the overall survival (OS) of patients with high Aurora-A expression was significantly poorer than those with low Aurora-A expression (the median duration of OS 27.25 vs. 72.2 months, respectively, p < 0.001; Figure 2A). Also, elevated expression of Aurora-A predicted an inferior progression-free survival (PFS) with the median duration of PFS 15.5 vs. 57.5 months respectively (p < 0.001, Figure 2B).

Further analysis was performed between Aurora-A expression and subsets of NSCLC patients with each clinical stage. High expression of Aurora-A associated with a significant trend toward worse OS and PFS in patients of stage I (p = 0.032 for OS and p = 0.007 for PFS, Figure 3A and B), stage II (p = 0.026 for OS and p = 0.038 for PFS, Figure 3C and D), and stage III (p = 0.017 for OS and p = 0.002 for PFS, Figure 3E and F).

Multivariate analysis using the Cox proportional hazards model revealed that high Aurora-A expression was an independent and significant prognostic factor for OS (hazard ratio: 3.311; 95%CI: 1.899-5.775; p < 0.001; Table 2) and PFS (hazard ratio: 3.360; 95%CI: 2.066-5.466; p = 0.003; Additional file2: Table S1). Of other parameters, CEA level was evaluated as a positive independent prognostic factor for OS and PFS, tumor size for PFS.

Since high level of Aurora-A was correlated with poor prognosis in NSCLC patients that treated by platinum-based chemotherapy, we sought to explore the association between Aurora-A expression and cisplatin-resistance in vitro. We detected the protein expression level of Aurora-A in A549 and drug-resistant A549/DDP cells by Western blot and immunofluorescence analysis. As shown in Figure 4A (upper panel) and B, Aurora-A expression was elevated in cisplatin-resistant A549/DDP cells. A549/DDP cells with higher Aurora-A level also showed raised proliferation (Additional file1: Figure S3). In another lung cancer drug-resistant cell line H460/DDP, Aurora-A expression was elevated relative to H460 cells (Figure 4A, lower panel). Then we performed forced expression of Aurora-A in A549 and H460 cells (Figure 4C), and treated the cells with increasing concentrations of cisplatin. Forced expression of Aurora-A increased the resistance of A549 and H460 cells to cisplatin in cell viability assay. Colony formation assay indicated that overexpression of Aurora-A enhanced cell growth and resistance to cisplatin in A549 cells (Figure 4D).

We knocked down Aurora-A expression in A549/DDP and H460/DDP cells by siRNA (Figure 5A, left panel), and found that it resulted in a significantly enhanced sensitivity to cisplatin (Figure 5A, right panel). Colony formation assays showed that down-regulation of Aurora-A by Tet-inducible RNAi suppressed proliferation of A549/DDP cells severely in the presence of cisplatin (Figure 5B). We then used VX-680, a small-molecule inhibitor of the Aurora kinases, and Aurora-A kinase specific inhibitor MLN8237 to determine whether inhibition of Aurora-A activity would induce the similar effect. As expected, incubation of A549/DDP cells with increasing doses of VX-680 or MLN8237 led to decrease in Aurora-A phosphorylation at Thr288 (Figure 5C). Phosphorylation inhibition was also observed in histone H3 at Ser10 (Additional file1: Figure S4). A549/DDP and H460/DDP cells were treated with single-agent DDP, VX-680, MLN8237, or DDP in combination with VX-680 or MLN8237. The results showed that the cytotoxicity effect of DDP, VX-680 or MLN8237 was not obvious, while combination of DDP and VX-680 or combination of DDP and MLN8237 produced significant higher cytotoxicity effects (Figure 5D; Additional file1: Figure S5). We also assessed apoptotic cell death by annexin V-FITC staining. The combination treatment induced higher proportion of apoptosis, compared with single-agent treatment (Figure 5E).

Given the association between Aurora-A expression levels and metastasis in NSCLC patients, we sought to identify whether inhibition of Aurora-A could reverse migration ability of cisplatin-resistant cells. We employed wound-healing assay to evaluate the effect of repression of Aurora-A on cell migration. As shown in Figure 6A, with DDP treatment, the migration ability of A549/DDP cells was obviously higher than A549 cells, and which was effectively inhibited by lack of Aurora-A expression. Similar results were observed by using Aurora kinase inhibitor VX-680 or MLN8237 (Figure 6B).