Annexin A2 could enhance multidrug resistance by regulating NF-κB signaling pathway in pediatric neuroblastoma

The human neuroblastoma cell lines SK-N-BE(1) and SK-N-BE(2) were obtained from Children’s Oncology Group (COG, USA). SK-N-BE(1) cell line was isolated from bone marrow of a 2-year-old patient when he was primary diagnosed without any treatment, and SK-N-BE(2) cell line was gained from the same patient when he has been received several courses of chemotherapy with cyclophosphamide, doxorubicin, as well as vincristine. These two cell lines had different drug sensitivity (Additional file 1: Figure S1) [11‐13].

SK-N-BE (1) and SK-N-BE (2) cell lines were maintained in RPMI 1640 (Gibco, USA) with 10% fetal bovine serum (Gibco, USA), 50 units/ml penicillin, 50 μg/ml streptomycin (Yeasen, China) and 1X ITS (Sodium Pyruvate 0.11 g/L, L-glutamine 1.5 g/L, NaHCO3 1.5 g/L (Yeasen, China)). These two cell lines were incubated humidified air supplemented 5% CO2 at 37 °C.

To enable the quantitative proteomics, we used specialized stable isotope labeling with amino acids in cell culture (SILAC) medium supplemented with 10% dialyzed fetal bovine serum (Invitrogen, Darmstadt, Germany) to label the cells, where deficient Arginine and Lysine were supplemented with either stable isotope encoded heavy Arginine and Lysine (Euriso-top) or normal Arginine and Lysine for the light. Labeling efficiency was confirmed before the following quantitative full proteome analysis.

After washing with pre-cold phosphate-buffered saline (PBS, Mediatech Inc., Manassas, VA), equal number of SILAC coded SK-N-BE (1) and SK-N-BE (2) cells were mixed and lyzed with chilled lysis buffer, 8 M urea in 100 mM NH₄HCO₃ with 1× protease inhibitor cocktail (Roche Diagnostic GmbH, Germany) and incubated on ice for half an hour. Cell debris were removed by centrifugation and the supernatant was collected. Protein concentration in lysates was determined by the Bradford assay accordingly. The extracted proteins were digested by trypsin according to the protocol. Briefly, after the alkylation and reduction, protein lysate was diluted four times with 100 mM NH₄HCO₃ (pH 8.0) to decrease the urea concentration to 2 M. Trypsin was then added at an enzyme-substrate ratio of 1:100 (w/w). The digestion was subsequently performed at 37 °C for 16 h.

Tryptic digest was then pre-separated into 20 fractions with a reverse phase C18 Xbridge column (Waters Corp., Milford, MA) by a preparative HPLC system (Shimadzu, Japan) at high pH condition. Each fraction was lyophilized and further analyzed in a nano-HPLC coupled Orbitrap Q-Exactive mass spectrometer (Thermo Fisher Scientific, Waltham, MA). The raw spectra data was processed by Maxquant software (v1.4.1.2) [14]. MS/MS spectra data was searched against the Uniprot human database (88,817 sequences) by Andromeda search engine [15].

In our study, there were 42 pediatric patients with neuroblastoma (international neuroblastoma staging system (INSS) stage I–IV & IVs) who were histologically diagnosed in Xinhua hospital affiliated to Shanghai Jiaotong University School of Medicine during January 2006 and December 2011. All of the patients enrolled in the study had received chemotherapy before surgery or not, and all of them underwent surgery or biopsy. Each tumor specimen was stored in liquid nitrogen for tissue microarray analysis. The experimental protocols were approved by the Ethics Committee of the Xinhua hospital affiliated to Shanghai Jiaotong University School of Medicine. We analyzed the recorded data of each patient and performed follow-up through phone calls.

These tissue specimens were separated out a small part and shaped in the special mold for microarray preparation. After fixed in the 4% paraformaldehyde over night, they were trimmed and embedded in paraffin as a planned array. Then, samples were sectioned (5 μm) and attached to poly-L-lysine coated slides. Immunohistochemical staining was performed using a standard immunoperoxidase staining procedure (primary antibody, ANXA2, 1:100, CST, USA). Hematoxylin was used as a counterstain. The tissue sections were viewed independently by two pathologists in a blind fashion. IHC staining was graded on a specialized scale from 0 to 4: where 0 represented negative expression, 1 represented weakly positive expression (0–10% positive cells), 2 represented mildly positive expression (10–30% positive cells), 3 represented moderately positive expression (30–50% positive cells), 4 represented strongly positive expression (50–100% positive cells). The scale was determined according to the average number of positive cells in 10 random fields of one slide.

The pGMLV-SC5 RNAi and pGMLV-SB3 RNAi vector (GeneChem, Shanghai, China, Additional file 2: Figure S2) was used to package the lentivirus for knocking down the ANXA2 expression. The shRNA sequences were designed by selecting a specific target sequence for the human ANXA2 gene, as described by GeneChem (Additional file 3: Figure S3). These oligonucleotides were annealed and subcloned according to the manufacturer’s recommendations, and a non-targeting control shRNA (scrambled control) was obtained from GeneChem. Lentivirus was packaged according to the manufacturer’s protocol of GeneChem. The produced lentiviral particles named pGC-shANXA2-LV and pGC-vector-LV. Cells were infected with lentiviruses in the presence of polybrene (5 μg/ml) for 12 h, 37 °C incubation, and then the lentiviral medium was replaced by the fresh culture solution.

Doxorubicin (Sigma, St. Louis, MO, USA) and Etoposide (Sigma) stock solution (10 mM) was prepared by dissolving it in DMSO, and solution was diluted as ten concentration gradients from 0 μM to 2 μM for doxorubicin and 0 μM to 16 μM for Etoposide before experiment. SK-N-BE(2) cells were plated at a density of 1 × 10⁴ cells per well in 96-well plates. After 24 h, cells were divided two groups and infected with pGC-shANXA2-LV or pGC-vector-LV. 48 h later, cells of two groups were treated with Doxorubicin and Etoposide respectively in the above ten concentration gradients for 72 h. Cell viability was determined using a Cell Counting Kit-8 (CCK-8) reagent (Yeasen, China) according to the manufacturer’s instructions. The absorbance of the samples at a wavelength of 450 nm was measured using a microplate reader (BioTek, USA). Drug-response curve was draw according the result of CCK-8 assay.

24-well plates were infused with 4% Poly-L-Lysine (Beyotime Biotechnology, China) solution 4 °C overnight for coating. The next day, NB cells (2×10⁴ per well) were plated 24 h before infection. 48 h after lentivirus infection, treatment of Doxorubicin or Etoposide was performed. 24 h later, cell apoptosis was determined by using the Hoechst 33342 reagent (Yeasen) according to the manufacturer’s protocol. Cell apoptosis was detected under fluorescence microscope (Olympus, Tokyo, Japan) using 348 nm excitation wavelengths.

SK-N-BE (2) cells (1×10⁵ per well) were plated in the 6-well plates, 24 h later, lentivirus (pGC-shANXA2-LV and pGC-vector-LV) infection was performed. 48 h after infection, completed medium containing Etoposide was replaced, and 48 h later, cell apoptosis was detected apoptosis by using the Annexin V-FITC kit (BD, San Diego, CA, USA). According to the manufacturer’s protocol, after cells were detached from the culture plate, they were washed twice with PBS, and were resuspended in 190 μl reaction system of binding buffer. 5 μl of Annexin V and 5 μl of propidium iodide (PI) were added for each sample. Subsequently, the cells were incubated in the dark for 15 min at room temperature. A total of at least 10,000 events were collected and analyzed by flow cytometry (BD FACScanto II) and the apoptotic ratios generated automatically in the Q2 and Q4 quadrant. Due to the fact that the color of doxorubicin was red, which was an interference in the emissional spectrum of PI, we didn’t conduct flow cytometric assay for doxorubicin.

Total cell proteins were isolated and purified using the Total Protein Kit (Sigma) according to the manufacturer’s protocol. Extraction and isolation of nuclear and cytoplasmic proteins were performed according to the manufacturer’s instructions (Yeasen, China). Briefly, after treatment, cells were washed twice with PBS, scraped and collected by centrifugation at 1500×g for 5 min. Cell pellets were resuspended in 250 μl extraction buffer A and incubated for 10 min on ice. Afterwards, extraction buffer B was added, and samples were vortexed for 30 s at 4 °C. After centrifugation at 12000×g for 5 min at 4 °C, supernatants, which contained the cytosolic fractions, were removed out. The rest pellets, which contained the nuclei, were resuspended in 50 μl of nuclear extraction buffer, and nuclear proteins were extracted by shaking the samples for 30 min at 4 °C. Afterwards, samples were centrifuged at 12000×g for 5 min at 4 °C and the supernatants were removed for analyzing. The protein concentration was determined using the BCA assay method (Pierce, USA).

Equal amounts of proteins were loaded onto an 10% polyacrylamide gel for electrophoresis and then electro-transferred onto PVDF membranes (Bio-Rad, USA). Blots were blocked with 5% BSA at room temperature for 2 h, and incubated with the appropriate primary antibodies at 4 °C overnight. After washing the blots 3 times, the membranes were incubated with HRP-anti-rabbit IgG for 1 h at room temperature. Proteins were visualized by electrogenerated chemiluminescence (ECL, Pierce Biotechnology, USA) with the Bio-Rad ChemiDoc XRS imaging system. Antibodies were purchased from Cell Signaling Technology (Beverly, USA) and used to direct against ANXA2 (1:1000, #8235), NF-κB1/p50 (1:1000, #13586), Rel-A/p65 (1:1000, #8242), Cleaved-PARP (1:1000, #5625) and AIF (1:1000, #5318), Anti-GADPH (1:2000, #2118) and Histone-H3 (1:2500, #4499) were used as loading control. The results were quantitative analyzed through the software of Image J (X64, v. 2.1.4). Total RNA from cells was isolated with Trizol reagent (Invitrogen, CA, USA). Reverse transcription was performed using the Advantage RT-for-PCR Kit (Takara, Beijing, China). Expression change of three NF-κB target genes (IL-1A, IL-1B, IL-6) in ANXA2-knockdown SK-N-BE(2) cells was assessed by qPCR assay (Takara) according to the manufacturer’s recommendations. These top and bottom primer sequence was provided in Additional file 4: Figure S4.

SK-N-BE (2) cells were seeded in a 24-well chamber slide (3×10⁵ per well). One day later, cells were transfected with lentivirus, and 48 h after transfection, tumor necrosis factor-α (TNF-α, 100 ng/ml) was added for treating 1 h. Then, cells were fixed with 4% paraformaldehyde solution at room temperature for 30 min, and were treated with 0.3% Triton X-100 for 20 min, blocked in 10% bovine serum albumin for 1 h. incubated overnight at 4 °C with primary antibodies NF-κB1/p50 (1:100, #13586, CST) and ANX A2 (1:150, #H00000302-M02, Novus, USA). Cells were washed and incubated in secondary antibody (1:200, Yeasen, China), DAPI (1:5000, Yeasen) was used to label the nuclei. Immunofluorescence was photography and analyzed by the LeicaCTR6000 microscope system combined with Image J (X64, v. 2.1.4) software. The nuclear localization of p50 was quantified according to data from 5 random figures in each group.

The pGMLV-CMV-MCS-3*flag-EF1-ZsGreen-T2A-Puro vector (GeneChem, Shanghai, China, Additional file 4: Figure S4e) was used to package the lentivirus, and we used the lentivirus to established the 3*flag-ANXA2 SK-N-BE(2) cell line as well as normal control cell line. Co-IP was performed as the procedure briefly. Lyze the cells with 5 ml pre-cold IP buffer for 30 min and sonicate the lysate 3 min on ice. Then, centrifuge the lysate at 20000 g for 20 min at 4 °C. Collect the supernatant into 15 ml tube and incubate with 50uL mouse IgG beads (Sigma, A0919) at 4 degree for 3hs and then incubate with 50uL Flag M2 beads (Sigma, A2220) overnight at four degree. Wash the beads with 1 ml IP buffer for 8 times. Vortex the beads with 200uL 3*flag peptides (Sigma, F4799) 20 min at RT to elute the target proteins. Acetone to precipitate the proteins overnight (1:8 volume), and then performed Western blotting.

Six week-old male nude mice were purchased from Shanghai Slac Laboratory Animal Co. Ltd. Animals were housed in individually ventilated micro-isolator cages and allowed chow and water ad libitum. All experiments were performed after obtaining protocol approval by the Animal Care and Use Committee of Xinhua Hospital, and in compliance with the NIH animal use guidelines. Nude mice were divided into 4 groups of 5 mice each. Each group was injected subcutaneously in the flank with 1 × 10⁷ SK-N-BE(2) cells pre-transduced with either pGC-shANXA2-LV or pGC-vector-LV in 200 μL PBS. Tumor sizes were determined with calipers every week by measuring the length and width. Tumor volumes were calculated according to the following formula: volume (mm³) = (length × width2) /2 [16]. When the tumors reached palpable size (about 100mm³), there were two groups of mice were started to receive intraperitoneal injection with the doxorubicin (1 mg/kg/every 3 days). The other two groups were received PBS intraperitoneal injection as negative control, the reference for the dose and course of treatment of doxorubicin can be found in Stewart’s and Geng’s work [16, 17]. Eight weeks after Chemotherapy, mice were sacrificed and tumor xenografts were removed, weighed, and fixed in formalin for HE and IHC staining.

SK-N-BE(1) and SK-N-BE(2) are two stable cell lines derived from the same NB patient but with different drug sensitivities. Proteins that are substantially altered between these two cell lines might directly or indirectly contribute to drug sensitivity. To investigate this possibility, we employed mass spectrometry-based proteomics (Fig 1a). Pearson correlation coefficient of the biological replicates was greater than 0.86, suggesting good reliability of protein quantification (Fig. 1b). In total, 6329 proteins were identified with a false discovery rate of <1% both at the protein and peptide level. Among these, 6230 proteins were quantified (Additional file 5: Table S1). Only proteins up- or down-regulated by greater than two-fold were considered for analysis. This left 460 significantly regulated proteins, including 244 significantly up-regulated and 216 significantly down-regulated proteins (Fig. 1c and Additional file 6: Table S2).

Levels of several known chemoresistance-related proteins were significantly altered (Table 1). MRP-1, the classic chemoresistance related protein, was almost five times more abundant in resistant cells compared with the sensitive cells. Galectin-1 and proteasome maturation protein, both of which are associated with drug resistance in many malignant cancers, showed significant up-regulation in the resistant cell line. Among the down-regulated proteins, we found several candidate proteins associated with drug-resistance, including DNA damage-binding protein 2 and DNA topoisomerase 3-alpha. Based on our datasets, we suggest there are additional proteins related to drug resistance that may be targets of further study.

Annexin A2 (ANXA2), a multi-compartment protein, is involved in cadherin and S100A10 protein binding, and is known to accelerate plasminogen activation [18]. Recent studies suggest this protein may contribute to drug resistance [19, 20]. In our study, profiling data revealed that ANXA2 was upregulated by more than one order of magnitude in the chemoresistant NB cell line SK-N-BE(2) (Suppl. Table 2). This suggests a potentially important role for ANXA2 in generation of drug resistance in NB. The mechanistic details of that role remain unknown.

To begin investigating the role of ANXA2, we reviewed historical data for each patient in our study. There were 19 girls and 23 boys. Median age at diagnosis was 2.3 years. INSS stage analysis revealed that there were 9 cases with stage I, 11 cases with stage II, 13 cases with stage III, 7 cases with stage IV, and 2 cases with stage IVs. The expression level of ANXA2 in these NB samples was determined by TMA-based IHC analysis. ANXA2 was differentially expressed among those NB samples. In 31 cases (73.8%), ANXA2 protein could be detected. In 42.9% of NB patients we found strong intensity staining (IHC staining grade 3–4, Fig. 2a-c.). We found that strong ANXA2 staining correlated with advanced stage (p < 0.0001, r2 = 0.3359, Fig. 2d). We also found that NB patients with higher expression of ANXA2 received more cycles of chemotherapy (p < 0.0001, r2 = 0.3713, Fig. 2e). Finally, to assess the prognostic value of ANXA2 expression in NB patients, we generated Kaplan-Meier curves (Fig. 2f). Patients with low expression of ANXA2 had more favorable survival than those with high expression (p = 0.008). The association between ANXA2 expression and clinical characteristics are displayed in Table 2. High levels of ANXA2 in NB were markedly associated with some indicators of NB progression, such as lymphatic and bone marrow metastasis.

To investigate the effects of ANXA2 expression on drug sensitivity, we performed a knockdown experiment using shRNA. After comparing reduction efficiency of three candidate shRNA sequences, we choose sequence sh3 for packaging lentivirus (Additional file 2: Figure S2). Using a CCK-8 cell viability assay, we found that the drug response curve was left-shifted in ANXA2 knockdown NB cells, suggesting that these cells were more sensitive to chemotherapeutic drugs (Fig.3a, b). Specifically, the IC50 of doxorubicin in ANXA2 knockdown NB cells was 2.77-fold (p = 0.034) lower than that of control cells. Similarly, ANXA2 knockdown NB cells treated with etoposide returned an IC50 value 7.87-fold (p = 0.049) lower than that of control cells. On microscopic examination, we found more extensive cytotoxic effects in ANXA2 knockdown cells compared with control cells, following treatment with chemotherapeutic agents (Fig. 3c).

To investigate whether ANXA2 knockdown had any effect on cell apoptosis following chemotherapy, we performed Hoechest 33,342 staining and flow cytometric analysis. SK-N-BE(2) cells were infected with pGC-shANXA2-LV and pGC-vector-LV prior to treatment with doxorubicin (0.5 μmol/L) or etoposide (10 μmol/L), respectively, for 48 h. As shown in Fig. 4a, ANXA2 knockdown did not induce apoptosis when cells were not exposed to the drugs. However, drug-treated cells transfected with pGC-shANXA2-LV showed greater apoptosis than cells transfected with vector alone (Fig. 4a, b). Annexin V/PI Flow cytometric analysis showed significantly increased apoptosis in the ANXA2 knockdown group compared with control group following etoposide treatment (Fig. 4c, d).

To determine whether ANXA2 knockdown activates an apoptotic pathway, the pro-apoptotic proteins AIF and Cleaved-PARP were measured by Western blot. Following treatment with doxorubicin or etoposide, we demonstrated increased expression of both AIF and Cleaved-PARP in knockdown groups compared with control groups (Fig. 4e~g). These results suggest that ANXA2 may be involved in resistance to chemotherapy-induced apoptosis.

Recent studies suggested that NF-κB pathway is closely involved in the multidrug resistance of neuroblastoma [21, 22]. We investigated whether ANXA2 regulates drug sensitivity via the NF-κB signaling pathway. First, we performed western blotting to examine which subunits of NF-κB was involved in the activation of NB cells. We found that subunit p50, but not p65, was involved in NF-κB activation (Fig. 5a~d). Further, the results clearly showed the subunit p50 was downregulated in the nucleus of ANXA2 knockdown NB cells compared with control cells following treatment with TNF-α (Fig. 5e~g).

Immunofluorescence staining for subcellular localizations of p50 in resting and stimulating states was examined. In the control group, ANXA2 and p50 were distributed across the cytoplasm and nucleus in the resting state. Following treatment with TNF-α, most p50 translocated to the nucleus. In the ANXA2 knockdown group, the situation was similar in the resting state, however, in response to TNF-α, nuclear translocation of p50 was attenuated (Fig. 6a, b).

To validate the combination of ANXA2 and p50, we performed the co-immunoprecipitation. The results showed clearly that ANXA2 could combine with the p50 subunit directly (Fig. 6c). In addition, in order to consolidate the ANXA2-NF-κB correlation, we selected three well known NF-κB downstream targets (IL-1A, IL-1B, IL-6) for qPCR analysis. Our results showed the IL-1B and IL-6 was decreased significantly in the transcriptional level after ANXA2 knockdown (Additional file 4: Figure S4a~c).

To further demonstrate the effect of ANXA2 on drug sensitivity in vivo, we generated a subcutaneous xenograft neuroblastoma model. In both control and ANXA2 knockdown groups, tumors were significantly smaller after chemotherapy (Fig. 7a). Further, tumor volume of NB xenografts transduced with Lenti-ANXA2-shRNA was smaller than that of the control (p = 0.041). (Fig. 7b). This suggests a positive effect of ANXA2 knockdown on chemotherapy in vivo.

On H&E staining, tumor tissues appeared obviously altered following chemotherapy: decreased numbers of small round NB cells, more foci of hemorrhage and more areas of necrosis. In ANXA2 knockdown groups, NB cells appeared more sensitive to doxorubicin (Fig. 7c~e). Besides that, we measured expression levels of mitochondrial apoptotic pathway components by immunohistochemistry. AIF and cleaved-PARP proteins were highly expressed in chemotherapy groups, but there was no significant difference between control and knockdown with Dox after quantification (Fig. 7c, g~h).