Introduction

Acute myeloid leukemia (AML) is characterized by rapid growth of abnormal white blood cells which start in the bone marrow and interfere with the generation of normal blood cells. It can sometimes spread to important organs in the body such as the brain, spinal cord, lymph nodes, liver and spleen. CCAAT/enhancer-binding protein-alpha (C/EBPα) is a key transcription factor in the control of lineage-specific gene expression and cell proliferation in hematopoiesis and is involved in the differentiation of certain blood cells (Rahman et al. 2016; Nerlov 2004; Yoshida et al. 2012). It has been evident that inhibition of C/EBPα expression and disruption of its activity stops the differentiation of myeloid progenitors (Lin et al. 2011). In several AML subtypes, C/EBPα expression is downregulated or mutated, most prominently, in the M2 subtype where C/EBPα mutations are observed in a population of patients, resulting in a blockade of monocytic or granulocytic differentiation (Nerlov 2004). Therefore, the role of C/EBPα during granulocyte differentiation and in serving as a tumor suppressor is critically important, indicating that therapies with an increase of expression and activation of C/EBPα may counteract the disruption of differentiation in AML.

An increase of glycolysis, the breakdown of glucose into pyruvate to generate ATP, is a hallmark of cancer cells and is known as the Warburg effect (Warburg et al. 1927). Several studies have demonstrated moderate to high levels of glycolytic metabolism in AML (Herst et al. 2011; Liu et al. 2013). Moreover, highly glycolytic AML cells are more resistant to apoptosis induced by anti-AML therapy, such as retinoic acid and arsenic trioxide, suggesting potential resistance to induction chemotherapy (Herst et al. 2011; Song et al. 2016). These findings indicate that targeting glycolysis is a viable strategy for modulating chemoresistance in AML (Xu et al. 2005; Song et al. 2016).

Natural products are promising resources for cancer drug discovery (Abdel-Hamid et al. 2017; Tung et al. 2016). Glycosides of spirostan and furostan type steroids, found in a wide variety of plants, have been demonstrated to display anticancer activities against many cancers including cancers of the prostate, breast, lung and liver. Dioscin, a natural spirostan steroid glycoside widely present in the family of Dioscoreaceae and Liliaceae, has been extensively studied for its anticancer effects in a variety of cancers through caspase-dependent multiple signaling pathways including upregulation of p53 and Bax, downregulation of Bcl-2, activation of estrogen receptor-β, induction of oxidative stress, mitochondrial stress, p38 mitogen-activated protein kinase (MAPK) and c-Jun N-terminal kinase (JNK) activation, and inhibition of phosphatidyl inositol 3-kinase (PI3K)/Akt and nuclear factor kappa B (NF-κB) activity (Cai et al. 2002; Wang et al. 2014; Zhao et al. 2016a, b; Zhang et al. 2016a, b; Song et al. 2017; Tao et al. 2017). Besides, dioscin has been reported to induce demethylation of DAPK-1 and RASSF-1α genes via the antioxidant capacity to stimulate apoptosis in bladder cancer cells (Zhou et al. 2017). Furthermore, it inhibits TGF-β1-induced epithelial-to-mesenchymal transition, migration and invasion of A549 lung cancer cells (Lim et al. 2017), and inhibits VEGFR2-mediated angiogenesis through the suppression of downstream kinases including Src, FAK, Akt and Erk1/2 (Tong et al. 2014).

In this study, the integrated bioinformatics approach and functional examination have been used in the aid of apoptosis-related protein array, transcription factor (TF)-DNA array, and proteomics studies to delineate dioscin-induced anti-AML effects as well as the mechanisms that are responsible for anti-proliferative, apoptotic and differentiation activities.

Materials and methods

Materials

Iscove’s Modified Dulbecco’s Medium (IMDM), RPMI 1640 medium, fetal bovine serum (FBS), penicillin, streptomycin, and all other tissue culture reagents were obtained from GIBCO/BRL Life Technologies (Grand Island, NY). The following antibodies were used: anti-mouse and anti-rabbit IgGs, TRAIL, PARP-1, C/EBP-α, caspase-9, -8 and -3, Bid, Bak, GAPDH (Santa Cruz Biotechnology Inc., Santa Cruz, CA), Fas, FasL and caveolin-1 (Transduction Lab, Lexington, KY). CD11b-FITC, 4,6-diamidino-2-phenylindole dihydrochloride (DAPI), JC-1, leupeptin, dithiothreitol, phenylmethylsulfonylfluoride (PMSF), propidium iodide (PI), all-trans retinoic acid (ATRA) and all of the other chemical reagents were obtained from Sigma-Aldrich (St. Louis, MO).

Cell lines and cell culture

HL-60 (promyelocytic leukemia), CCRF-CEM (T cell acute lymphoblastic leukemia), and Jurkat (T cell acute lymphoblastic leukemia) were from American Type Culture Collection (Rockville, MD). CCRF-CEM and Jurkat cells were cultured in RPMI-1640 medium and HL-60 cells were cultured in Iscove’s Modified Dulbecco’s Medium (IMDM). All cells were cultured with 10% fetal bovine serum (FBS) (v/v) and penicillin (100 U/ml)/streptomycin (100 μg/ml). Cultures were maintained in a humidified incubator at 37 °C in 5% CO2/95% air.

Mitochondrial MTT reduction activity assay

The cells at a density of 3 × 105 cells/ml were cultured at a 24-well plate. After the compound treatment at the mentioned concentrations and times, the mitochondrial MTT reduction activity was assessed. MTT was dissolved in phosphate-buffered saline (PBS) at a concentration of 5 mg/ml and filtered. From the stock solution, 50 μl per 500 μl of medium was added to each well and plates were gently shaken and incubated at 37 °C for 2 h. After the loading of MTT, the medium was replaced with 1 ml of 100% DMSO and was left for 5 to 10 min at room temperature for color development. The 24-well plate was read by enzyme-linked immunosorbent assay reader (570 nm) to get the absorbance density values.

Flow cytometric detection of apoptosis

After the treatment, the cells were harvested, washed twice with ice-cold PBS, fixed with 70% ethanol at 4 °C for 30 min, and washed with ice-cold PBS. The cells were then resuspended with 0.3 ml of PI solution containing Triton X-100 (0.1% v/v), RNase (100 mg/ml) and PI (80 mg/ml) in the dark. Cells were analyzed with FACScan and CellQuest software (Becton Dickinson, Mountain View, CA). The level of apoptotic cells containing sub-G1 DNA content was determined as a percentage of the total number of cells. In another assay, the cells were stained with FITC-Annexin V/PI using an apoptosis detection kit (BD Pharmingen) according to the manufacturer’s protocol. After the incubation at room temperature for 15 min in the dark, the apoptosis was detected and quantified using flow cytometric analysis.

DNA fragmentation assay

After the treatment, the cells were collected in a buffer containing 20 mM Tris pH 7.0 and 250 mM sucrose on ice. Total DNA was extracted by Genomic DNA kits (Geneaid, Taiwan). DNA was subsequently subjected to electrophoresis on 2% agarose gels containing SYBR1 green I (1:250 dilution of stock in TE buffer) (Molecular Probes, Eugene, OR), and visualized under UV light.

Confocal microscopic examination with DAPI staining

After the treatment, the cells were fixed with 100% methanol at room temperature for 5 min and were incubated in the solution containing 1 μg/ml DAPI for nuclear staining or the indicated antibody for the detection of specific protein. The cells were analyzed using a confocal laser microscopic system (Leica TCS SP2).

Microscopic observation of cell morphology

After the treatment, the cells were collected by centrifugation and resuspended in 200 μl of PreserveCyt solution. The suspension was passed through a Thinprep processing machine, and the cells were collected. The slides were fixed in 95% alcohol and then stained with Wright-Giemsa for 5 min at room temperature. Stained cells from each treatment group were examined under a fluorescence microscope (Olympus, USA).

Human apoptosis antibody array

To investigate the pathways by which dioscin induces apoptosis, we performed a determination of apoptosis-related proteins using an antibody array (human apoptosis antibody array kit, Raybiotech, Norcross, GA, USA) according to the manufacturer’s instructions. After the treatment with dioscin, the cells were collected and 300 μg of protein extract from each sample was incubated with the antibody array membrane for 4 h. The membrane was quantified using a Biospectrum AC ChemiHR 40 system (UVP, Upland, CA, USA) and the membrane image file was analyzed using UVP analysis software.

Measurement of mitochondrial membrane potential (ΔΨm)

JC-1 was used to determineΔΨm. Cells were treated in the absence or presence of the indicated agent. Thirty minutes before the termination of incubation, the cells were incubated with JC-1 (final concentration of 2 μM) at 37 °C for 30 min. The cells were finally harvested and the accumulation of JC-1 was determined using flow cytometric analysis.

Differentiation effect assay

Differentiation effects of dioscin and ATRA were assessed using established morphologic changes using Wright’s staining and immunophenotyping with CD11b expression. Expression of the CD11b (Becton–Dickinson, NJ) immunophenotypic marker was monitored. After treatment, the cells were harvested and incubated with the desired antisera for 30 min at 4 °C, then washed twice, and fixed with staining buffer (Becton–Dickinson, NJ) before analysis. For immunofluorescence analysis, following the appropriate incubation step, the fluoresceinated CD11b antibody was incubated at 4 °C for 30 min, followed by washing and fixing. At least, 104 cells were analyzed for FACScan and CellQuest software (BD, CA).

Transcription factor-DNA interaction array analysis

Nuclear extracts from cells were prepared with nuclear extract kit (Panomics, Inc., Redwood City, CA) according to the user manual. Protein-DNA interactions were analyzed using a set of biotin-labeled DNA binding oligonucleotides. Nuclear extracts (15 μg) were pre-incubated with biotin-labeled DNA probe to allow formation of protein/DNA complexes. The protein/DNA complexes were separated from the free probes using a spin column separation system (Panomics). The bound probes were extracted and hybridized to an array membrane spotted with 56 different consensus-binding sequences according to the protocol.

Two-dimensional gel electrophoresis (2-DGE)

2-DGE was performed using Amersham Biosciences IPGphor IEF electrophoresis units (Uppsala, Sweden). Protein samples (200 μg) extracted from the cells of untreated or 2 μM dioscin treatment for 24 h were mixed up with equal volume of rehydration buffer (8 M Urea, 20 mM DTT, 0.5% Triton X-100 and 1% IPG buffer). The rehydration step was carried out with precast 13 cm IPG strips for more than 12 h at low voltage of 30 V. IEF was run following a step-wise voltage increase procedure: 500 V and 1000 V for 3 h each and 2000 V for about 16 h. After IEF, the strips were subjected in equilibration buffers (6 M Urea, 50% Glycerol, 2% SDS and 0.5 M Tris-HCl pH 6.8) for 20 min. The strips were then transferred onto the second-dimensional SDS-PAGE that was run on 1.5 mm thick of 15% polyacrylamide gels.

Silver staining

The gels were fixed with 40% ethanol and 10% glacial acetic acid for 30 min, and then incubated in a sensitizing buffer containing 30% ethanol, 4.1% sodium acetate and 0.2% sodium thiosulfate for 30 min. After washing three times in water for 5 min each, the gels were stained in 0.1% silver nitrate solution containing 0.02% formaldehyde for 40 min. Development was performed for 15 min in a solution consisting of 2.5% sodium carbonate and 0.01% formaldehyde. Ethylenediaminetetraacetic acid solution (1.46%) was used to stop the development and the stained gels were then washed three times in water for 5 min each.

In gel digestion, mass spectrometry analysis, and protein identification

After silver staining, individual protein bands were excised from the gel and digested overnight at 37 °C with trypsin (1 mg/ml; Promega Corporation, Madison, WI), then the tryptic peptide mixture was eluted for 30 min at 4 °C from the gel with 100 ml of 0.1% trifluoroacetic acid (TFA), followed by 100 ml of 0.1% TFA/60% acetonitrile, and the combined extracts lyophilized. The tryptic peptides were resuspended in 0.1% TFA and analyzed by LC-MS/MS using an LTQ-Orbitrap Velos hybrid mass spectrometer (Thermofisher Scientific, Waltham, MA). Peptide separations were performed on-line with MS by nanoflow liquid chromatography (Dionex Ultimate 3000, Dionex Corporation, CA). Samples were injected in a 10-ml volume of starting mobile phase (2% acetonitrile in 0.1% formic acid) at a flow rate of 750 nl/min onto a Nanoacquity LC system with a 75 um × 15 cm C18 column. Samples were loaded for 15 min before switching the sample loop out of the flow path and decreasing the flow to 300 nl/min. The peptides were then separated by elution with a linear gradient of 2–90% acetonitrile in 0.1% formic acid at 300 nl/min in 90 min and injected onto a Thermo Scientific Velos Orbitrap electrospray mass spectrometer running at a resolution of 60,000 for precursors. Data-dependent MS/MS analysis was carried out using Xcalibur MS acquisition software (Xcalibur 2.1, Thermo Fisher Scientific, Waltham, MA). Each scan cycle consisted of a full scan MS acquired in profile mode at 60 K resolution by the Orbitrap analyzer over the mass range 350–1600 m/z, followed by data-dependent MS/MS scans of the 20 most intense peaks. MS/MS spectra were acquired in higher energy collisional dissociation mode with 35% normalized collision energy. Dynamic exclusion was set for 90 s with a 20 ppm window and monoisotopic precursor selection was enabled. Proteins were identified by Mascot search engine (version 2.2.2, Matrix Science) against Swiss-Prot version 43 protein database. The individual score for the MS/MS spectrum of each peptide was more than 20.

Ingenuity pathway analysis (IPA)

A list of dioscin-modulating proteins was uploaded to the Ingenuity Pathway Analysis (IPA) software to investigate the biological networks associated with these proteins (http://www.ingenuity.com). This analysis uses computational algorithms to identify networks consisting of focus proteins (proteins that were present in our list) and their interactions with other proteins (“non-focused”) in the knowledge base. Scores were calculated for each network according to the fit of the network to the set of focus proteins and used to rank networks in the Ingenuity analysis. IPA uses the proteins from the highest-scoring network to extract a connectivity pathway that relates candidate proteins to each other based on their interactions. These candidates were shown to be significantly associated with the Function, Disease, and Canonical Pathways. In addition, we also searched for other proteins involved in this network using the build tools of IPA. Significance of the biofunctions and the canonical pathways were tested by the Fisher exact test p value.

Western blotting

After the treatment, cells were harvested, centrifuged, and lysed in 0.1 ml of lysis buffer containing 10 mM Tris–HCl (pH 7.4), 150 mM NaCl, 1 mM ethylene glycol tetraacetic acid, 1% Triton X-100, 1 mM PMSF, 10 mg/ml leupeptin, 10 mg/ml aprotinin, 50 mM NaF and 100 mM sodium orthovanadate. In certain experiment, the nuclear extraction kit (Panomic, Redwood City, CA) was used to separate nuclear and cytosolic fraction. Total protein was quantified, mixed with sample buffer and boiled at 95 °C for 5 min. Equal amount of protein (30 μg) was separated by electrophoresis in a 12% SDS-PAGE, transferred to PVDF membranes and detected with specific antibodies. The immunoreactive proteins after incubation with appropriately labeled secondary antibody were detected with an enhanced chemiluminescence detection kit (Amersham, Buckinghamshire, UK).

Data analysis

Experimental data are presented as the mean ± standard error of the mean (SEM) of three to five independent experiments. Normality and homogeneity of variance assumptions were checked. The statistical analysis was performed using one way analysis of variance (ANOVA) for multiple samples sets. P values less than 0.05 were considered statistically significant. Statistical Package for the Social Sciences version 16.0 (SPSS Inc., Chicago, IL, USA) and GraphPad Prism version 3.0 (GraphPad Software Inc., La Jolla, CA, USA) software were used.

Results

Dioscin displays anti-proliferative and apoptotic activities

HL-60, CCRF-CEM, and Jurkat cells were used to determine the anti-proliferative and apoptotic activity of dioscin. The data showed that dioscin induced concentration-dependent inhibition of cell survival with IC50 values at low micromolar concentrations (Fig. 1a). Flow cytometric analysis of DNA staining showed that dioscin induced both concentration- and time-dependent increase of hypodiploid (sub-G1) DNA content in HL-60 (Fig. 1b and c). Dioscin-induced apoptosis was validated using DNA fragmentation assay (Fig. 1d), revealing that dioscin effectively induced programmed cell death in leukemic cells.

Fig. 1
figure 1

Identification of apoptotic effect. a Graded concentrations of dioscin were added to the cells for 24 or 48 h. The cytotoxic effect was determined by MTT assay. b HL-60 cells were treated in the absence or presence of dioscin for 24 h. After the treatment, the cells were fixed and stained with propidium iodide to analyze DNA content by FACScan flow cytometer. c Quantitative analysis of sub-G1 (apoptosis) population using flow cytometric analysis of propidium iodide staining. d HL-60 cells were treated with 2 μM dioscin for 2, 4, and 6 h. Total DNA was extracted and subjected to electrophoresis for the detection of DNA fragmentation. Data are expressed as mean ± SEM of three to five independent experiments. *P < 0.05, **P < 0.01, and ***P < 0.001 compared with the control

Full size image

Dioscin induces the activation of TNF-receptor family and extrinsic apoptosis pathway

A differentiation block and an accumulation of immature myeloid cells characterize AML. Induction of apoptosis and differentiation by dioscin has been studied on the focus of anti-AML using AML cell line HL60, a good model for studying growth control and differentiation (Burmistrova et al. 2015; Masuda et al. 2015; Shen et al. 2011). After exposure of HL-60 cells to dioscin for 24 h, the cells were lysed and 43 apoptosis-related proteins were examined using a human apoptosis antibody array kit. As a result, several pro-apoptotic proteins including extrinsic apoptosis pathway related death ligands (e.g., tumor necrosis factor α (TNF-α) and Fas ligand (FasL)) and receptors (e.g., death receptor 6 (DR6), TNF-related apoptosis-inducing ligand (TRAIL) receptor, TNF-R1 and TNF-R2) were upregulated (Fig. 2a). Of note, the protein expressions of both cIAP-2 and CD40 also were increased (Fig. 2a). Most TNF receptors require specific adaptor protein such as Fas-Associated protein with Death Domain (FADD), TNFR-associated death domain (TRADD), receptor-interacting protein 1 (Rip1) and TNF receptor-associated factor 2 (TRAF2) for stimulating downstream signaling, cIAP-2 and CD40 are able to interact with the adaptor protein, serving as mediators of the signal transduction (Baker and Reddy 1998; Varfolomeev et al. 2008). The data indicate that dioscin induces the activation of TNF-receptor superfamily. Next, the IPA software was used to determine whether the dioscin-modulating proteins could be grouped into different functional classes. The top ten canonical signaling pathways activated by dioscin were listed and the death receptor signaling was identified as the most relevant signaling pathway (Fig. 2b).

Fig. 2
figure 2

Protein array analysis and the ingenuity pathway analysis. HL-60 cells were incubated in the absence or presence of dioscin (2 μM) for 24 h. The cells were lysed and equal amount of protein from each sample was used for the assay. a Quantitative analysis in the arrays showed differences in the apoptotic marker proteins. All the protein levels have been normalized in the positive dots. b The ingenuity pathway analysis (IPA) software was used to determine the top ten canonical signaling pathways activated by dioscin

Full size image

Dioscin induces apoptosis through extrinsic and intrinsic apoptotic pathways

Several death ligands (e.g., FasL and TRAIL), death receptors (e.g., Fas) and the signals involved in extrinsic apoptotic pathways (caspase-8 and Bid) were studied. The data demonstrated that dioscin induced a significant increase of these signal proteins in a time-dependent manner (Fig. 3a). Because Bid/tBid (truncated Bid) was a key protein required in the crosstalk between extrinsic and intrinsic pathways, the activation of intrinsic pathway was examined. As a result, dioscin induced the loss of mitochondrial membrane potential (Fig. 3b) and activation of caspase-9, a key initiator caspase in intrinsic pathway (Fig. 3a), suggesting the activation of this pathway. The activation of caspases, including caspase-8, -9 and -3, by dioscin in HL-60 cells was similar to previous report (Wang et al. 2014). Notably, the Bak expression was significantly upregulated by dioscin in the present study (Fig. 3a). It has been documented that Bid preferentially activates Bak to ensure cell death through mitochondrial dysfunction and to reach death ligand-mediated maximal induction of apoptosis (Sarosiek et al. 2013). The Bak activation was probably attributed to Bid stimulation.

Fig. 3
figure 3

Effect of dioscin on death receptor signaling and mitochondrial membrane potential. HL-60 cells were incubated in the absence or presence of dioscin (2 μM) for the indicated times. a The cells were harvested for the detection of protein expression using Western blotting or b the cells were incubated with JC-1 dye for the detection of mitochondrial membrane potential using flow cytometric analysis. Data are expressed as mean ± SEM of three to four independent experiments. *P < 0.05, **P < 0.01, and ***P < 0.001 compared with the control

Full size image

Identification of multiple transcription factors regulated by dioscin

In addition to induction of apoptosis, the differentiation therapy has been considered as a promising approach for the treatment of AML. The emerging evidence shows that differentiation-based approach to target AML is a potential strategy. Transcription factors play a major role in differentiation in a number of cell types, including various hematopoietic lineages and leukemogenesis (Ohlsson et al. 2016). It is important to identify and characterize the transcription factors that specifically activate important genes in the myeloid lineage. A functional analysis of transcription factors was performed that HL-60 cells were exposed to dioscin, rendering to protein/DNA array that permits profiling DNA binding activities of 54 transcription factors (TF). The signal intensities revealed in the membrane array and reflected the TFs binding activity. The purity of different cellular fractions was confirmed by probing each fraction for corresponding subcellular marker proteins. As shown in Fig. 4, the cytoplasmic marker glyceraldehyde 3-phosphate dehydrogenase (GAPDH) primarily localized in the cytoplasmic fraction, whereas the nuclear marker nucleolin mainly resided in the nuclear fraction, confirming the success of nuclear fractionation. The data showed that several TFs binding activities were increased by dioscin, including Myc/Max, MRE, p53, smad/SBE, and SRE; whereas, those of GR/PR, Brn-3 and AR were decreased. Furthermore, the IPA software generated top ten canonical signaling pathways listed in Fig. 5a, which highlighted several probable pathways including glucocorticoid receptor signaling, Toll-like receptor signaling, PXR/RXR activation, and acute myeloid leukemia signaling of target cells. Accordingly, the key signal that was cross-reacted by these pathways was explored (Fig. 5b). Increasing lines of evidence have demonstrated that glucocorticoid receptor can regulate gene expression by interacting with CCAAT/enhancer-binding protein α (C/EBPα) (Muratcioglu et al. 2015). Several toll-like receptor agonists can downregulate C/EBPα expression, leading to increased expressions of several genes responsible to a variety of neurological disorders (Ejarque-Ortiz et al. 2007). More importantly, the variant fusion proteins, promyelocytic leukemia zinc finger-retinoic acid receptor α (PLZF-RARα), have been identified to inhibit myeloid cell differentiation through associating with C/EBPα and inhibiting its activity (Girard et al. 2013). Therefore, C/EBPα has been considered as a cellular target if differentiation occurs under dioscin treatment. As a consequence, dioscin led to a profound increase of protein expression of C/EBPα (Fig. 5c).

Fig. 4
figure 4

Effect of dioscin on the activation of transcription factor. HL-60 cells were incubated in the absence or presence of dioscin (2 μM) for 1 h. The nuclear extracts were prepared and preincubated with biotin-labeled DNA probe to allow formation of protein/DNA complexes. The protein/DNA complexes were separated from the free probes using a spin column separation system. The bound probes were extracted and hybridized to an array membrane spotted with 56 different consensus-binding sequences according to the protocol

Full size image
Fig. 5
figure 5

Determination of canonical pathways from transcription factor-DNA array activated by dioscin-treated in HL-60 cells. a Top ten transcription factor canonical signaling pathways activated by dioscin demonstrated in Fig. 4. b The ingenuity pathway analysis (IPA) software was used to determine the canonical signaling pathways. c The cells were treated with or without dioscin (2 μM) for the indicate times. After the treatment, the cells were harvested for the detection of protein expression using Western blotting. The data have been normalized with the internal control of GAPDH expression and have been demonstrated relative to zero hour. Data are expressed as mean ± SEM of three independent experiments. *P < 0.05, **P < 0.01, and ***P < 0.001 compared with the zero hour control

Full size image

Dioscin induces myelomonocytic differentiation of AML cells

A block in terminal myeloid differentiation is a hallmark of myeloid leukemia cells. To investigate the potential role of dioscin in myeloid cell differentiation, especially in promyeloid stage, the expression of cell surface CD11b during HL-60 differentiation was examined. Dioscin-induced increase of CD11b protein expression was detected by immunostaining (Fig. 6a) and the differentiation toward myelomonocytic/granulocytic lineages was further confirmed based on hematoxylin and eosin staining and CD11b staining (Fig. 6b). Flow cytometric analysis of cell surface CD11b expression showed that both dioscin and ATRA significantly increased the expression levels (Fig. 6c and d).

Fig. 6
figure 6

Effect of dioscin on cell differentiation. a, b HL-60 cells were incubated in the absence or presence of dioscin (2.0 μM) for 72 h, or c, d the cells were incubated with the indicated compound for 72 h. a The immunofluorescence staining with CD11b-FITC was performed. The nuclei were stained with DAPI. Bar, 20 μl. b Wright-Giemsa staining and immunohistochemical staining of CD11b were performed. c The cells were incubated with FITC-conjugated anti-CD11b mouse antibody and analyzed by flow cytometry. d The quantitative analysis was performed. Data are expressed as mean ± SEM of three independent experiments. *P < 0.05 and **P < 0.01 compared with the control

Full size image

Identification of differentially expressed proteins in dioscin-treated HL-60 cells

To elucidate the differentially expressed proteins in dioscin-treated HL-60 cells, the in-gel digestions and LC-MS/MS were performed to identify proteins responsible for dioscin treatment. The analyses identified 43 differentially expressed proteins after an 8-h dioscin treatment; the names of these 43 proteins, their accession numbers in the Swiss-Port database, molecular mass, Mascot score, percentage of coverage, and their identities as well as quantitative data were summarized (Table 1). The assay also demonstrated an upregulation of C/EBPα expression to dioscin action (Table 1). Moreover, several canonical signaling pathways responded to dioscin were proposed, including glycolysis and gluconeogenesis pathways, after IPA network enrichment analysis (Fig. 7). Cellular metabolic state and individual metabolites have been suggested to regulate various cellular functions, including apoptosis and differentiation of functional phenotype of immune cells (Suzuki et al. 2016). The modification of glycolysis and gluconeogenesis pathways by dioscin might, therefore, reinforce its impact on both cell apoptosis and differentiation.

Table 1 Identification of differently expressed proteins between control and dioscin-treated HL-60 cells
Full size table
Fig. 7
figure 7

Determination of canonical pathways from IPA analysis of 2-D gel electrophoresis to identify cellular proteins. a The assay was perform as demonstrated in Table 1 and the top ten canonical signaling pathways activated by dioscin were obtained. b The ingenuity pathway analysis (IPA) software was used to determine the canonical signaling pathways

Full size image

Discussion

Owing to pharmacological significance, the steroidal saponins have caught the attention of phytochemists, biologists and drug discovery scientists. Recently intensive research has been focused on exploring saponins for anticancer therapies. Dioscin is one of the most common steroidal saponins, displaying convincing anticancer activity in several human cancer cells through apoptosis-inducing pathways (Wang et al. 2014; Zhao et al. 2016a; Zhang et al. 2016a, b). The present study has elucidated functionally significant interactions in dioscin-treated leukemic cells on transcription and proteome levels, assisting to understand both the role of individual interacting partners and the order of cascade protein interrelations in cellular regulatory and signaling pathways of both apoptosis and differentiation.

Apoptosis and differentiation machineries play a crucial role in hematopoietic cell homeostasis. A number of hematological diseases involve a deregulation of these machineries. The therapeutic strategy, therefore, is developed to restore the normal function of these machineries. The extrinsic apoptotic pathway is initiated by death receptor activation, leading to the formation of death receptor signaling platforms and activation of caspase-8 mediated cell killing (Lavrik 2014). Death receptors (e.g., TNFR1, DR4, DR5 and Fas) are cell surface receptors transmitting apoptotic signals initiated by individual ligand, such as TNF-α, TRAIL and FasL (CD95L). They can activate a caspase cascade and induce apoptosis very rapid upon ligand binding. Notably, it has been suggested that the sensitivity of leukemic cells to apoptosis is increased in a death receptor-dependent manner (Zheng et al. 2013). Apoptosis antibody array and Ingenuity Pathway Analysis (IPA) revealed that death receptor signaling pathway was identified as the highest-ranked molecular and cellular function upon dioscin-induced effects. Western blot analysis also validated the upregulation of death receptors and the ligands, and the downstream caspase activation. Similar studies have been reported by Kim and the colleagues (Kim et al. 2012). Importantly, dioscin induced a significant increase of tBid formation, a unique pro-apoptotic Bcl-2 family members integrating both the extrinsic and intrinsic pathways (Katz et al. 2012; Tiwari et al. 2015). Recent studies on structural and molecular levels of mitochondria and tBid protein have revealed that mitochondrial carrier homolog 2 participates in recruiting tBid to the mitochondria, where it activates apoptosis (Katz et al. 2012). Our data demonstrated a time-dependent loss of mitochondrial membrane potential and activation of caspase-9, confirming the crosstalk between both extrinsic and intrinsic apoptosis pathways to dioscin action. In addition, another pro-apoptotic Bcl-2 relative, Bak, was activated by dioscin. Bak is a key player to permeabilize mitochondrial outer membrane during the intrinsic apoptosis pathway. It converts to active form through undergoing a large conformational change to form apoptotic pores in mitochondrial outer membrane that allows the release of cytochrome c and other proteins to promote caspase activity to kill the cell. Because Bid has been suggested to preferentially activate Bak to ensure cell death through mitochondrial dysfunction (Sarosiek et al. 2013), our data suggest that dioscin efficiently induces caspase-dependent apoptosis through a crosstalk between extrinsic and intrinsic apoptosis pathways. In addition to caspase-dependent pathway in leukemia cells, dioscin also has been reported to induce cell death via apoptosis inducing factor (AIF) involved caspase-independent pathway in breast cancer cells (Kim et al. 2014).

Clinically effective differentiation therapy has been demonstrated in acute myelogenous leukemia. Recently, dioscin has been reported to induce differentiation of chondrocytes and osteoblasts through LDL related protein 5 and estrogen receptor pathway (Zhang et al. 2014; You et al. 2016). In contrast, it suppressed osteoclast differentiation through the inhibition of Akt activation (Qu et al. 2014). However, the effects of dioscin on leukemia differentiation and mechanism study are scarcely reported. C/EBPα is an important transcription factor in the regulation of lineage-specific gene expression and cell proliferation in hematopoiesis and is involved in the differentiation of certain blood cells (Nerlov 2004; Yoshida et al. 2012). It has been demonstrated that inhibition of C/EBPα expression and disruption of its activity stops the differentiation of myeloid progenitors (Lin et al. 2011). Consistent with its importance in normal myeloid differentiation, expression and activity of C/EBPα are impaired in several types of myeloid leukemia by various mechanisms, such as transcriptional silencing, translational inhibition, posttranslational modification, decrease in DNA binding, or point mutations resulting in increased production of a dominant negative form) (Tenen 2003; Radomska et al. 2015). AML is characterized by a block in differentiation along one or more hematopoietic lineage. It has been well identified that C/EBPα is mutated in many AML samples examined and most prominently in the promyeloid leukemia cell type (Nerlov 2004). Accordingly, re-introduction of C/EBPα expression and function may prevail against the block of differentiation, leading to apoptosis of leukemic cells and recovery of cells with normal function. Radomska and the colleagues have established a cell-based high throughput screening to identify chemical compounds capable of inducing C/EBPα expression and myeloid differentiation (Radomska et al. 2015). In this study, a protein/DNA array to profile the DNA binding activity of multiple TFs was used and the IPA analysis showed several probable pathways to dioscin action; importantly, C/EBPα plays a common and crucial role in these pathways (Ejarque-Ortiz et al. 2007; Girard et al. 2013; Muratcioglu et al. 2015). The data also revealed that dioscin induced a profound increase of C/EBPα expression and CD11b protein expression (a classical myeloid lineage marker). It also induced the differentiation of HL-60 cells toward myelomonocytic/granulocytic lineages.

It is noteworthy that C/EBPα also plays a role in death receptor involved apoptosis pathway. It has been documented that the infection of adenovirus vector expressing C/EBPα gene (Ad-C/EBPα) is able to induce apoptosis in various types of cells mainly through death receptor pathways (Wang et al. 2009). Yao and the colleagues have reported that the ectopic overexpression of HMG-box containing protein 1 in myeloid cells induces apoptosis and enhances differentiation through an increase of FasL levels and myeloid-specific transcription factor C/EBPα (Yao et al. 2005). Moreover, CD95/Fas death receptor pathway has been suggested to be an integral part of the apoptotic response associated with the end of the normal terminal myeloid differentiation program. Deregulated c-myc expression can activate this signaling pathway prematurely and C/EBPα has been implicated in the downregulation of c-myc expression (Hoffman et al. 2002). Taken together, these studies suggest that C/EBPα can be a crucial regulator responsible for leukemic apoptosis through death receptor-mediated apoptosis pathway.

The in-gel digestions and LC-MS/MS were performed to identify proteins responsible for dioscin treatment. The IPA network enrichment analysis showed that dioscin might affect the glycolysis and gluconeogenesis pathways in HL-60 cells. Because the concomitant activation or downregulation of glycolysis and gluconeogenesis would cause a metabolic stress, impairing the metabolic rewiring of cancer cells, targeting glycolysis/gluconeogenesis pathways for metabolic reprogramming may be a feasible anticancer strategy (Ma et al. 2013; Khan and Chakrabarti 2015). More than ten protein spots were found to be up-regulated by dioscin, such as C/EBPα and several glycolysis-related proteins including triosephosphate isomerase, cofilin-1 (interaction with triosephosphate isomerase) and phosphoglycerate kinase (Jung et al. 2002). In contrast, dioscin downregulated about twenty proteins in which pyruvate kinase muscle isozymes (PKM) were the most susceptible one. PKM2 is a limiting glycolytic enzyme catalyzing the final step in glycolysis that is crucial in tumor metabolism and growth. Several lines of evidence show that the knockdown of PKM2 inhibits cell proliferation and induces apoptosis in a variety of cancers (Hu et al. 2015); furthermore, upregulation of PKM2 has been demonstrated in prednisolone resistant leukemia (Hulleman et al. 2009), suggesting PKM2 as a therapeutic target. The in-gel digestions and LC-MS/MS determination showed that dioscin down-regulated PKM protein levels. The determination of its functional role on PKM2 needs further investigation.

Conclusion

Our integrated bioinformatics approach reveals the importance of dioscin-induced death receptor related apoptosis, energy metabolic pathways and differentiation induction (Fig. 8). Dioscin exerts its antileukemic effect through the up-regulation of both death ligands and death receptors and a crosstalk activation of mitochondrial apoptosis pathway with the collaboration of tBid and Bak formation. In addition, proteomics approach reveals an altered metabolic signature of dioscin-treated cells and the induction of differentiation of promyelocytes to granulocytes and monocytes in which the C/EBPα plays a key role.

Fig. 8
figure 8

Schematic figure for dioscin-mediated signaling pathways in HL-60 cells. Dioscin exerts its antileukemic effect through the upregulation of both death ligands and death receptors and a crosstalk activation of mitochondrial apoptosis pathway with the collaboration of tBid and Bak formation. In addition, proteomics approach reveals an altered metabolic signature of dioscin-treated cells and the induction of differentiation of promyelocytes to granulocytes and monocytes in which the C/EBPα plays a key role

Full size image