Quantitative Proteomic Profiling of Tachyplesin I Targets in U251 Gliomaspheres

Tachyplesin I is a cationic peptide isolated from hemocytes of the horseshoe crab and its anti-tumor activity has been demonstrated in several tumor cells. However, there is limited information providing the global effects and mechanisms of tachyplesin I on glioblastoma multiforme (GBM). Here, by using two complementary proteomic strategies (2D-DIGE and dimethyl isotope labeling-based shotgun proteomics), we explored the effect of tachyplesin I on the proteome of gliomaspheres, a three-dimensional growth model formed by a GBM cell line U251. In total, the expression levels of 192 proteins were found to be significantly altered by tachyplesin I treatment. Gene ontology (GO) analysis revealed that many of them were cytoskeleton proteins and lysosomal acid hydrolases, and the mostly altered biological process was related to cellular metabolism, especially glycolysis. Moreover, we built protein–protein interaction network of these proteins and suggested the important role of DNA topoisomerase 2-alpha (TOP2A) in the signal-transduction cascade of tachyplesin I. In conclusion, we propose that tachyplesin I might down-regulate cathepsins in lysosomes and up-regulate TOP2A to inhibit migration and promote apoptosis in glioma, thus contribute to its anti-tumor function. Our results suggest tachyplesin I is a potential candidate for treatment of glioma.


Introduction
Gliomas, the most common group of primary brain tumors, are subcategorized into astrocytomas, oligodendrogliomas and ependymomas. According to World Health Organization (WHO), glioblastoma multiforme (GBM), the most malignant and lethal form of brain tumor in adults, is a grade IV astrocytoma with very high morbidity and mortality. The disease has a very poor prognosis with short median survival, only about 15 months, despite current multimodal treatment including maximal surgical resection if feasible, followed by a combination of radiotherapy and/or chemotherapy [1]. Therefore, it is imperative to present new and more effective therapeutic interventions to better control GBM.
In fact, the short median survival of GBM is largely ascribed to the inevitable tumor recurrence. Recent research has paid more attention to the existence of glioma stem cells (GSCs), which are a subgroup of tumor cells with properties that resemble those of neural stem cells, and are able to drive tumorigenesis and likely contribute to rapid tumor recurrence [2]. These cells were first described more than ten years ago and have been demonstrated with the capability of multi-lineage differentiation, self-renewal and extensive proliferation [3]. In addition, GSCs can endure and even thrive in stressful tumor conditions, including hypoxia, oxidative stress, inflammation, acidic stress, and low glucose [4]. Moreover, their resistance to conventional therapy and promotion of tumor angiogenesis also influence clinical practice [5,6]. Thus, GSCs provide new insight into the strategy in GBM therapy.
Three-dimensional growth model, a growth sphere formed by cancer stem cells under specific culture conditions in vitro, is a more reasonable model for tumor biology and drug screening in vitro studies [7,8]. Likewise, GSCs also have the characteristic of forming spheres and clinical data show that the rates of existence of gliomaspheres were more prominent in high grade malignant gliomas [9]. Previously, we isolated gliomaspheres from U251 glioma cell lines and tried to apply it for drug screening. We found that there were undifferentiated GSCs and differentiated cancer cells with different differentiation degrees in gliomaspheres, which were similar to the growth state of glioma in vivo [10]. Our previous data showed that gliomaspheres express stem cell biomarkers nestin and CD133, which are certain phenotypes of GSC, and tachyplesin I inhibited the viability and proliferation of gliomaspheres dose dependently, by damaging the plasma membrane and inducing differentiation of GSCs [11]. These findings indicate that tachyplesin I is a potential anti-tumor drug which may be used in GBM therapy.
tachyplesin I, a cationic peptide with 17 residues (NH 2 -K-W-C-F-R-V-C-Y-R-G-I-C-Y-I-R-R-C-R -CONH 2 ), was originally isolated from hemocytes of the horseshoe crab (Tachypleus tridentatus) [12]. It has the ability of anti-enzymatic hydrolysis due to two disulfide-stabilized β-hairpins [13]. Several studies have demonstrated that tachyplesin I can inhibit the proliferation and affect the differentiation of tumor cells, such as hepatocarcinoma, gastric adenocarcinoma and leukemia [14,15]. This peptide has also been demonstrated to activate the classic complement pathway to lyse and kill tumor cells and to alter the expression of tumor suppressor genes and oncogenes to induce cell differentiation and reverse the malignant phenotype [16,17]. Most interestingly, the negatively charged components of cancer cells, which are quite different from neutral normal cells, are more vulnerable by the positively charged cationic peptides, including tachyplesin I. The electrostatic attraction between cancer cells and cationic peptides is believed to play a major role in the selective disruption of cancer cell membranes, which avoids traditional mechanism of drug resistance [18].
Although the anti-tumor effect of tachyplesin I has been studied to some extent, the mechanism of anti-tumor activity in GBM is largely unknown. In recent years, proteomics has been shown to be a powerful approach for exploring the molecular mechanisms of anti-tumor drugs. In this study, our primary goal was to identify the changes in protein expression profile of U251 gliomaspheres under the treatment of tachyplesin I, which may help us to better understand the molecular mechanisms underlying potential anti-glioma drugs. Here, both gel-based and shotgun proteomic approaches were performed to gain a higher proteome coverage and better quantification results [19]. Proteomic analysis using two dimension difference gel electrophoresis (2D-DIGE) and stable isotope dimethyl labeling based Liquid chromatography-mass spectrometry/mass spectrometry (LC-MS/MS) revealed that 192 proteins were differentially expressed in U251 gliomaspheres in response to tachyplesin I. Biological involvement of these proteins are further discussed in detail through signaling pathways and protein-protein interaction network analysis. Furthermore, the expression of cathepsins in lysosomes and TOP2A was further validated by Western blot and PRM, due to their important involvement in the anti-tumor activity of tachyplesin I, by inhibiting migration and promote apoptosis of glioma cells, respectively.

Protein Expression Profile of tachyplesin I Treated U251 Gliomaspheres Using 2D-DIGE Analysis
The 2D-DIGE images, which were scanned at the wavelengths of 488/520, 532/580, and 633/ 670 nm, visualize the protein expression pattern in the cells ( Figure 1A). In the image analysis, 1298 protein spots were detected. Of these, 35 spots with fold change larger than ±1.5 were considered significantly altered in tachyplesin I treated U251 gliomaspheres compared with untreated control ( Figure 1B). Among the protein spots that satisfied the statistical criteria, 26 were confidently identified by MALDI-TOF/TOF analysis. Out of 26 identified proteins, 13 were up-regulated while the others were down-regulated in tachyplesin I treated U251 gliomaspheres. Up-regulated proteins were mainly involved in regulation of cell cycle and apoptosis, and cytoskeleton proteins (Table 1). Conversely, down-regulated proteins were involved in glycolysis, response to stimulus and calcium or ion binding (Table 1). Several proteins (Vimentin, Phosphate carrier protein, mitochondrial and Guanine nucleotide-binding protein G(q) subunit alpha) were identified more than once in different location of 2D-DIGE gel, suggesting diverse protein isoforms, such as the occurrence of post-translational modification. Representative images of one up-regulated protein endothelin-converting enzyme 1 (ECE1) and one down-regulated protein alpha-enolase (ENO1) in different dose groups are shown in Figure 1C. Western blot assay was performed to confirm the results obtained from 2D-DIGE experiment and the results were consistent ( Figure 1C). identified by MALDI-TOF/TOF analysis. Out of 26 identified proteins, 13 were up-regulated while the others were down-regulated in tachyplesin I treated U251 gliomaspheres. Up-regulated proteins were mainly involved in regulation of cell cycle and apoptosis, and cytoskeleton proteins (Table 1). Conversely, down-regulated proteins were involved in glycolysis, response to stimulus and calcium or ion binding (Table 1). Several proteins (Vimentin, Phosphate carrier protein, mitochondrial and Guanine nucleotide-binding protein G(q) subunit alpha) were identified more than once in different location of 2D-DIGE gel, suggesting diverse protein isoforms, such as the occurrence of post-translational modification. Representative images of one up-regulated protein endothelin-converting enzyme 1 (ECE1) and one down-regulated protein alpha-enolase (ENO1) in different dose groups are shown in Figure 1C. Western blot assay was performed to confirm the results obtained from 2D-DIGE experiment and the results were consistent ( Figure 1C).    Figure 1B; b Average ratios of spot abundance of tachyplesin I-treated samples relative to the control, represent data from three separate experiments and student's t test p values are given as a measure of confidence for the ratio of each spot measured; c 0: control group; 10: 10 µg/mL dose group; 40: 40 µg/mL dose group; d Functional categories according to Gene ontology and panther biological process annotations; e ND, not detected or p value > 0.5.

Relative Quantification Using Dimethyl Labeling Based LC-MS/MS Analysis
Peptide samples from the control, and 10 µg/mL and 40 µg/mL tachyplesin I-treated U251 gliomaspheres were labeled with dimethyl stable isotope tags. To obtain reliable quantification results, we conducted one forward and one reverse dimethyl labeling experiments. A total of 74,240 peptides from 4891 proteins were identified in the forward-labeling samples and 73,892 peptides from 4854 proteins in the reverse-labeling samples (Supplementary Materials Tables S1-S4). In both forward and reverse labeling experiment, the labeled peptides account for more than 99.8% of total identified peptides, indicating a good labeling efficiency. A total of 5737 proteins were reliably quantified in both the forward and reverse labeling experiments, of which 4008 proteins were overlapped ( Figure 2B). The protein ratios of L/H and M/H in the forward labeling experiment and protein ratios of M/L and H/L in the reverse labeling experiment indicate the relative abundance of proteins in 10 µg/mL and 40 µg/mL tachyplesin I-treated groups compared to the control. The log 2 transformed protein ratios between two different experimental groups all form a symmetric distribution curve with the peak around zero (the original ratio = 1) (Figure 2A), and proteins that were increased or decreased in the forward-labeling experiment were also increased or decreased in the reverse-labeling experiment ( Figure 2C), suggesting that there was no bias in the labeling and LC-MS experiments. Only those proteins with fold changes >2 and quantified in both forward and reverse labeling experiments were reported as differentially expressed proteins. Among 4088 proteins, the expression levels of 166 were significantly altered by tachyplesin I treatment. Among them, 55 were up-regulated ( Table 2) while 111 proteins were down-regulated (Table 2). Figure 2D shows representative mass spectrometric results for the identification and quantification of the peptide DPDAQPGGELMLGGTDSK from cathepsin D, which clearly reveals the down-regulation of this protein in both sets of experiments.

Relative Quantification Using Dimethyl Labeling Based LC-MS/MS Analysis
Peptide samples from the control, and 10 μg/mL and 40 μg/mL tachyplesin I-treated U251 gliomaspheres were labeled with dimethyl stable isotope tags. To obtain reliable quantification results, we conducted one forward and one reverse dimethyl labeling experiments. A total of 74,240 peptides from 4891 proteins were identified in the forward-labeling samples and 73,892 peptides from 4854 proteins in the reverse-labeling samples (Supplementary Materials Tables S1-S4). In both forward and reverse labeling experiment, the labeled peptides account for more than 99.8% of total identified peptides, indicating a good labeling efficiency. A total of 5737 proteins were reliably quantified in both the forward and reverse labeling experiments, of which 4008 proteins were overlapped ( Figure 2B). The protein ratios of L/H and M/H in the forward labeling experiment and protein ratios of M/L and H/L in the reverse labeling experiment indicate the relative abundance of proteins in 10 μg/mL and 40 μg/mL tachyplesin I-treated groups compared to the control. The log2 transformed protein ratios between two different experimental groups all form a symmetric distribution curve with the peak around zero (the original ratio = 1) (Figure 2A), and proteins that were increased or decreased in the forward-labeling experiment were also increased or decreased in the reverse-labeling experiment ( Figure 2C), suggesting that there was no bias in the labeling and LC-MS experiments. Only those proteins with fold changes >2 and quantified in both forward and reverse labeling experiments were reported as differentially expressed proteins. Among 4088 proteins, the expression levels of 166 were significantly altered by tachyplesin I treatment. Among them, 55 were up-regulated ( Table 2) while 111 proteins were down-regulated ( Table 2). Figure 2D shows representative mass spectrometric results for the identification and quantification of the peptide DPDAQPGGELMLGGTDSK from cathepsin D, which clearly reveals the down-regulation of this protein in both sets of experiments.

Cellular Functions of Differentially Expressed Proteins and Associated Pathways
Systematic gene ontology (GO) analysis of 192 differentially expressed proteins identified from both 2D-DIGE and dimethyl labeling proteomic approaches was performed using PANTHER and DAVID tools. Molecular function analysis revealed that the majority of the differentially expressed proteins demonstrated catalytic (42.93%), binding (26.18%) and structural molecule activities (10.99%) ( Figure 3A). The biological processes altered by tachyplesin I treatment were most involved in metabolic processes (30.13%), cellular processes (19.88%), developmental processes (8.43%), localization (8.43%) and biological regulation (8.13%) ( Figure 3B). Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways including lysosome pathway (15 proteins), glycosaminoglycan degradation pathway (6 proteins), antigen processing and presentation pathway (8 proteins), DNA replication pathway (5 proteins), type I diabetes mellitus pathway (4 proteins) and glycolysis/gluconeogenesis pathway (4 proteins) are the top pathways altered in response to tachyplesin I treatment (Table 3).

Cellular Functions of Differentially Expressed Proteins and Associated Pathways
Systematic gene ontology (GO) analysis of 192 differentially expressed proteins identified from both 2D-DIGE and dimethyl labeling proteomic approaches was performed using PANTHER and DAVID tools. Molecular function analysis revealed that the majority of the differentially expressed proteins demonstrated catalytic (42.93%), binding (26.18%) and structural molecule activities (10.99%) ( Figure 3A). The biological processes altered by tachyplesin I treatment were most involved in metabolic processes (30.13%), cellular processes (19.88%), developmental processes (8.43%), localization (8.43%) and biological regulation (8.13%) ( Figure 3B). Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways including lysosome pathway (15 proteins), glycosaminoglycan degradation pathway (6 proteins), antigen processing and presentation pathway (8 proteins), DNA replication pathway (5 proteins), type I diabetes mellitus pathway (4 proteins) and glycolysis/gluconeogenesis pathway (4 proteins) are the top pathways altered in response to tachyplesin I treatment (Table 3).

Cellular Functions of Differentially Expressed Proteins and Associated Pathways
Systematic gene ontology (GO) analysis of 192 differentially expressed proteins identified from both 2D-DIGE and dimethyl labeling proteomic approaches was performed using PANTHER and DAVID tools. Molecular function analysis revealed that the majority of the differentially expressed proteins demonstrated catalytic (42.93%), binding (26.18%) and structural molecule activities (10.99%) ( Figure 3A). The biological processes altered by tachyplesin I treatment were most involved in metabolic processes (30.13%), cellular processes (19.88%), developmental processes (8.43%), localization (8.43%) and biological regulation (8.13%) ( Figure 3B). Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways including lysosome pathway (15 proteins), glycosaminoglycan degradation pathway (6 proteins), antigen processing and presentation pathway (8 proteins), DNA replication pathway (5 proteins), type I diabetes mellitus pathway (4 proteins) and glycolysis/gluconeogenesis pathway (4 proteins) are the top pathways altered in response to tachyplesin I treatment (Table 3).

tachyplesin I Reduces Expressions of Several Lysosomal Acid Hydrolases
As shown in Figure 4A, consistent with the results of proteomic analysis, protein level of lysosomal protective protein (CTSA) was verified to be down-regulated by tachyplesin I using Western blot. Further, other family members of cathepsins, including cathepsin B (CTSB) and cathepsin D (CTSD), as well as cathepsin A (CTSA) were analyzed by PRM mass spectrometry with three technical replicates. For each protein, two unique peptides were selected and monitored for quantification. The skyline software was used to extract the peak areas (area under the curve, AUC) of six to seven strongest transition ions for each peptide (Supplementary Materials Table S5). The normalized sum AUC of all the transitions for each peptide are showed in Figure 4B, which demonstrates that two unique peptides derived from the same protein have a consistent trend, and variations among different technical replicates are small. The results of PRM analysis showed that tachyplesin I down-regulated the levels of CTSA, CTSB and CTSD, which are consistent with dimethyl labeling results.

Protein-Protein Interaction Network of Differentially Expressed Proteins
Protein-protein interaction (PPI) network was established based on the total 192 differentially expressed proteins related to tachyplesin I treatment, including 26 proteins found in 2D-DIGE analysis and 166 proteins found in dimethyl labeling-based LC-MS analysis. Among them, 180 proteins could connect into a network through direct interaction or an intermediate partner at the PPI level ( Figure 5A). Interestingly, DNA topoisomerase 2-alpha (TOP2A) seemed to be the crucial protein in the effects of tachyplesin I as it has the most numerous connections and forms the most complex link with other proteins in the signal network ( Figure 5B).

Protein-Protein Interaction Network of Differentially Expressed Proteins
Protein-protein interaction (PPI) network was established based on the total 192 differentially expressed proteins related to tachyplesin I treatment, including 26 proteins found in 2D-DIGE analysis and 166 proteins found in dimethyl labeling-based LC-MS analysis. Among them, 180 proteins could connect into a network through direct interaction or an intermediate partner at the PPI level ( Figure 5A). Interestingly, DNA topoisomerase 2-alpha (TOP2A) seemed to be the crucial protein in the effects of tachyplesin I as it has the most numerous connections and forms the most complex link with other proteins in the signal network ( Figure 5B).

Confirmation of the Involvement of TOP2A in the Effects of tachyplesin I and Correlation with Clinical Prognosis in TCGA Database
Western blot result ( Figure 5C) showed that the expression level of TOP2A was dose-dependent increased after treatment of tachyplesin I, which was consistent with the result of dimethyl labeling based LC-MS/MS analysis (Table 2). At the same time, the expression level of TOP2A was also checked by PRM analysis. As shown in Figure 5D, after tachyplesin I treatment, the expression level of TOP2A was up-regulated, which verified the data obtained from dimethyl labeling based quantification. Then, we used cBioPortal tool to analyze the relationship between the mRNA transcript level of TOP2A and clinical prognosis of GBM patients based on TCGA database to examine the effects of tachyplesin I by targeting on TOP2A. As shown in Figure 5E, patients with alterations in TOP2A at mRNA transcript level have a better prognosis compared with those without alterations in TOP2A. The analysis showed a significantly better overall and disease-free survival of patients with over-expression of TOP2A.

Discussion
More and more studies have shown that certain cationic antimicrobial peptides (AMPs), which are toxic to bacteria but not to normal mammalian cells, exhibit a broad spectrum of cytotoxic activity against cancer cells [20]. tachyplesin I, which is isolated from hemocytes of the horseshoe crab, has been identified as a member of AMPs and exhibits cytotoxic activity against cancer cells. However, it is uncertain why only some types of AMPs get kill cancer cells, while others not. Besides, whether the molecular mechanisms underlying the antitumor and antimicrobial activities are the same or not remains unclear. Through this study we aim to identify the protein targets of tachyplesin I in U251 gliomaspheres by carrying out a large-scale proteome analysis, which can help us to better understand the molecular mechanisms underlying AMPs as potential anti-glioma drugs.
In this study, gel-based 2D-DIGE and stable isotope dimethyl labeling based LC-MS/MS analysis were combined to reveal the alteration in proteome of U251 gliomaspheres treated with tachyplesin I. A total of 192 differentially expressed proteins were identified, most of which are involved in the cellular process of metabolism, especially glycolysis process, and many proteins are localized as cytoskeleton proteins and lysosomal acid hydrolases. Especially, the expression level of some proteins of interest was validated by PRM, a high-resolution method first published in 2012 and had several potential advantages over traditional approach [21]. For example, PRM spectra are highly specific as a result of all the product ions of a peptide are recorded to confirm peptide identity, while traditional MRM analysis can only monitor one transition of a precursor peptide at a time. Moreover, high-resolution of the orbitrap mass analyzer can separate co-eluted background ions, thus increasing selectivity [22].
One of the hallmarks of tumor cells is the preference of glycolysis over oxidative phosphorylation as the main source of energy. Although glycolysis yield less ATP compared to oxidative phosphorylation with the same amount of beginning materials, tumor cells overcome this disadvantage by increasing the up-take of glucose, thus facilitates a higher rate of glycolysis [23]. Studies have showed that glycolysis plays a role in the invasion activity of glioma cells and is becoming a potential drug target [24]. In this study, glycolytic/gluconeogenesis enzymes including alpha-enolase (ENO1), gamma-enolase (ENO2), triosephosphate isomerase (TPI1) and phosphoglycerate kinase 1 (PGK1) were down-regulated in response to tachyplesin I treatment, indicating that tachyplesin I may disrupt the normal energy metabolism process in gliomaspheres through reduced glycolysis, thus contributing to its anti-tumor effect.
Uncontrolled and invasive proliferation is one feature of grade IV glioma, and in order to block and restrain mitotic division, cytoskeleton has been a time-honored target in cancer chemotherapy [25]. In this study, tachyplesin I treatment on U251 gliomaspheres altered the expression of 18 cytoskeleton proteins as classified by PANTHER classification system. Among them, vimentin and ezrin, which are known to be involved in the regulation of metastasis, were down-regulated under the treatment of tachyplesin I, suggesting that cytoskeleton are influenced by tachyplesin I, thus contributes to its anti-tumor activity.
Out of 192 altered proteins, 15 are lysosomal acid hydrolases, including proteases, glycosidases, sulfatases, lipases and so on. In addition, DAVID pathway classification system revealed lysosome as the most significantly altered pathway. More and more experimental evidences suggest that tumor invasion and metastasis are associated with alterations in lysosomes and increased expression of the lysosomal proteases termed cathepsins [26]. In this study, cathepsins consist of cathepsin A, B and D were down-regulated in response to tachyplesin I treatment. Cathepsin A, also called lysosomal protective protein, is a serine carboxypeptidase implicated in autophagy. It induces tumor cell dissemination and a significant increase in cathepsin A activity in lysates of metastatic lesions of malignant tumor was observed compared to primary focus lysates [27]. Cathepsin B is a lysosomal cysteine protease of the papain family of enzymes that function as an endopeptidase and an exopeptidase [28]. Cathepsin D, an aspartic protease resides in membrane of lysosomes, is involved in autophagy and apoptosis pathways [29]. Interestingly, it has been shown that cathepsin B and D play an important role in human glioma progression and invasion [30]. The expression and enzyme activity of cathepsin B and D gradually increased in high-grade glioblastoma. Inhibition of cathepsin B or D activity attenuates extracellular matrix degradation thus reduces migration of glioma cells [31]. Our data showed that the levels of these cathepsins were significantly decreased in tachyplesin I-treated gliomaspheres compared with untreated cells. All those evidences indicate the potential of tachyplesin I as a therapeutic agent for glioma by targeting the lysosomal activity.
In further PPI analysis of differentially expressed proteins, DNA topoisomerase 2-alpha (TOP2A) was shown to be the possible critical target protein of tachyplesin I. TOP2A is a nuclear enzyme for regulation of DNA topology and replication. TOP2A was discovered to be the target of many anti-tumor drugs which had already been widely used in clinic. Previous reports have shown that DNA damage and fragmentation induced by covalent binding of TOP2A to DNA, and forced expression of TOP2A in cells triggered the apoptotic cell death [32,33]. In addition, the TOP2A level has a close relationship with the activity of these anti-tumor drugs and a high level of TOP2A is the foundation of drug susceptibility. Meanwhile, decreased level, altered phosphorylation or mutation of TOP2A could induce the loss of anti-tumor drug target and develop multiple drug resistance (MDR), which has been confirmed in atypical MDR studies with many cell lines [34,35]. Our results showed that there was an increased TOP2A level in U251 gliomaspheres treated with tachyplesin I and it suggest the possible synergistic effect with TOP2A-targeting drugs, combination of which may be more effective on targeted goals and improve chemotherapy effect.

Cell Culture and Treatment with tachyplesin I
U251 human glioma cells were obtained from the Chinese Academy of Sciences Cell Band (Shanghai, China) and cultured in RPMI-1640 medium containing 10% fetal bovine serum (FBS) and 100 units/mL penicillin/streptomycin at 37 • C in a humidified atmosphere with 5% CO 2 . The U251 cells in the logarithmic growth phase were thoroughly dissociated to prepare single-cell suspensions. Cell suspensions were washed twice in PBS and resuspended in Neurobasal-A medium with 1× B27 plus 50 ng/mL basic fibroblast growth factor (bFGF) and 50 ng/mL epidermal growth factor (EGF). After 7 days culture, clones of different morphological types were collected. The obtained cells which exhibited certain glioma stem cell phenotypes [11] were cultured as gliomaspheres and passaged every 7 days, based on sphere size. tachyplesin I was synthesized by Hanyu Bioengineering Company (Shenzhen, China) with a purity of >95%. Concentrations of tachyplesin I for cell exposure were determined by cell viability assay as described previously [11]. The second generation gliomaspheres were treated with 0, 10, 40 and 80 µg/mL of tachyplesin I for 24 h, and then cells were centrifuged and collected.

CyDye Minimal Labeling of Protein Samples and 2D-DIGE Electrophoresis
Proteins were extracted from gliomaspheres treated with 0, 10, 40 and 80 µg/mL of tachyplesin I using the lysis buffer containing 7 M urea, 2 M thiourea, 4% (w/v) CHAPS and 30 mM Tris-HCl. The concentrations of proteins were determined with the 2-D Quant kit (GE Healthcare, Piscataway, NJ, USA) according to the manufacturer's instructions. Then an equal amount (25 µg) of each protein sample was minimally labeled with Cy3 or Cy5 fluorescent dyes (GE Healthcare) according to the manufacturer's recommended protocols and the internal standard, resulting from pooling equal aliquots of all experimental samples, was labeled with Cy2.
Six differently pooled samples (Table 4), which comprised equal amounts of Cy3-and Cy5-labeled protein samples and Cy2-labeled internal standard, were then separated by first dimension of isoelectric focusing using 24 cm IPG strips (pH 3-11, nonlinear gradient, GE Healthcare), followed by second dimension separation into 12.5% SDS-PAGE gels. Gels were then scanned with different channels for Cy2-, Cy3-, and Cy5-labeled proteins, using a Typhoon Trio Variable Mode Imager (GE Healthcare). The resulting 18 maps were imported into DeCyder 2D v6.5 (GE Healthcare) for statistical analysis. Each gel was separately processed by the Differential In-gel Analysis (DIA) module for spot detection, background subtraction and in-gel normalization before processed by the Biological Variation Analysis (BVA) module for spot matching and intercomparison across the six gels. Student's t-test was used to analyze the significance of protein spots between two groups, and one way ANOVA was subsequently used to assess the biological significance among all the experimental groups. Statistically significant spots (p < 0.05) with an average ratio ≥1.5 or ≤−1.5 were chosen for protein identification.

In-Gel Digestion and Protein Identification by MALDI-TOF/TOF
For identification of spots of interest, a gel was prepared by separating 1 mg of unlabeled proteins pooled from all the samples. The gel was stained by Coomassie Brilliant Blue G-250 and destained by water to reveal the protein spots. After matching to the analytical DIGE gel, each spot of interest was manually excised from the gel and put into a 1.5 mL tube, followed by thorough decoloration with 50% acetonitrile in 25 mM ammonium bicarbonate and dehydration in 100% acetonitrile. Then each gel piece was digested overnight at 37 • C by trypsin in 25 mM ammonium bicarbonate buffer. Peptides were extracted from each gel piece, desalted, and identified by an UltrafleXtreme MALDI-TOF/TOF mass spectrometer (Bruker Daltonics, Billerica, MA, USA) according to previously described [20].

Dimethyl Labeling of Protein Samples
Cells were lysed with a lysis buffer containing 4% SDS, 100 mM Tris, pH 8.0 and 1× protease inhibitor cocktail (Roche, Indianapolis, IN, USA). Protein concentrations were determined using a Pierce BCA Protein Assay Kit (Thermo Fisher Scientific, Waltham, MS, USA). One milligram protein from each sample was reduced with 5 mM dithiothreitol, alkylated with 15 mM iodoacetamide, and precipitated by methanol and chloroform [36]. The resulting pellets were resuspended in lysis buffer containing 8 M urea, 0.1 M Tris-HCl, pH 8.5 and the concentration of urea was diluted to below 2 M before overnight digestion with trypsin (Promega, Madison, WI, USA).

High pH Fractionation of Peptides and LC-MS/MS Analysis by Obitrap
The dimethyl-labeled sample was resuspended in 1% formic acid (FA), loaded into SepPak column, and fractionated into five fractions by eluting the peptides with 3%, 6%, 9%, 15% and 80% (vol/vol) acetonitrile in 5 mM ammonium formate (pH 10.0), sequentially. After lyophilization in Speedvac, samples were resuspended in 0.1% FA and analyzed by a Q-Exactive orbitrap mass spectrometer (Thermo Fisher Scientific) coupled to an Easy-nLC 1000 (Thermo Fisher Scientific) ultrahigh pressure liquid chromatography (UHPLC). The LC separation system consisted of a trap column (100 µm i.d. × 4 cm) and an analytical column (75 µm i.d. × 20 cm) both packed with 3 µm/ 120 Å C18 resins (Dr. Maisch HPLC GmbH, Ammerbuch, Germany). The eluting buffers were 0.1% FA in H 2 O (buffer A) and 0.1% FA in 99.9% ACN (buffer B). The peptides were first loaded onto the trap column and then separated by the analytical column with 50 min gradient from 7% to 22% buffer B followed by 10 min gradient from 22% to 35% buffer B at a flow rate of 300 nL/min. MS data was acquired in data dependent acquisition (DDA) mode. Survey full scan MS spectra (m/z 350-1550) were acquired in the Orbitrap with resolution of 70,000, target automatic gain control (AGC) value of 3 × 10 6 , and maximum injection time of 100 ms. Dynamic exclusion for scanned presursors was employed for 60 s. After each MS scan, the 10 most intense precursor ions (z ≥ 2) were sequentially isolated and fragmented by higher-energy collisional dissociation (HCD) using normalized energy 27% with an AGC target of 1 × 10 5 and a maxima injection time of 50 ms at 17,500 resolution.
Raw data were searched through UniProt Homo sapiens protein database containing 70,076 sequence entries via Sequest HT algorithm with the following parameters: two missed cleavage sites by trypsin, 10 ppm mass tolerance for precursors, 0.02 Da mass tolerance for fragments, and carbamidomethylation (+57.021 Da) of cysteineas static modifications. Moreover, the following dynamic modifications were also set: oxidation of methionine (+15.995 Da), deamidation of asparagine or glutarnine (+0.984 Da), and dimethylation for light-labeled (+28.031 Da) or medium-labeled (+32.056 Da) or heavy-labeled (+36.076 Da) lysine, and N-terminus. All the identified peptides were filtered by FDR <0.01 as reliable identification. Protein Discoverer was used for relative quantification. Differentially expressed proteins were considered for ratios ≤0.5 (down-regulated) and ≥2 (up-regulated).

Bioinformatic Analysis
The function reports of the candidate proteins whose expression was altered in U251 gliomaspheres due to the effect of tachyplesin I treatment were obtained from the UniProt database (http://www.uniprot.org/) and the protein list of UniProt IDs was input into the PANTHER classification system (http://pantherdb.org/) for GO analysis according to their molecular functions and biological processes. The relevant signaling pathways highly associated with the effect of tachyplesin I treatment on U251 gliomaspheres were identified using DAVID analysis (https://david. ncifcrf.gov/). The protein-protein interaction network of all the differentially expressed proteins was established using String (http://string-db.org/), and then the data was exported as .net file and imported into pajek software for degree based partition of the proteins in the network. The correlation of the possible key proteins involved in the effects of tachyplesin I in our proteomic analysis with its mRNA transcript level and clinical prognosis in GBM patients based on per TCGA data was analyzed by cBioPortal tools (http://www.cbioportal.org/) [38].

Parallel Reaction Monitoring (PRM) Mass Spectrometry
We applied PRM to validate the major protein changes observed in the dimethyl labeling analyses. Proteins were extracted from another batch of differently treated U251 gliomaspheres (biological replicate) and digested to peptides. These unlabeled peptides were fractionated and identified as described above with only difference in database searching (no dimethylation as dynamic modifications). For PRM analysis, 2 µg of non-fractionated peptides from each group were separated using the same LC system. Linear gradient ranging from 4% to 35% buffer B over 60 min was used. For each target protein, two unique precursor peptide ions were monitored in the inclusion list. The settings for MS full scan were the same as in the DDA mode with only different in m/z scan range (300-900). The following MS/MS PRM scan parameters were set: orbitrap resolution of 35,000, AGC target value of 5 × 10 5 , auto maximum IT, isolation window of 2 m/z, HCD collision energy of 27, and starting mass of m/z 110. The PRM raw files were analysed using Skyline [39] to extract the peak areas of six to seven most intense transitions for each peptide. Then the data was imported to GraphPad for statistical analysis. Differences between two groups were analyzed by the Student's t-test and statistical significance was considered when p < 0.05.

Western Blot Assay
Total proteins were extracted from different groups of U251 gliomaspheres with the same treatment as described in the DIGE analysis, and protein concentrations were quantified by BCA kit. Western blot procedures were carried out as we previously described [40], with minor modifications. Namely, after boiling for 5 min with loading buffer, the same amount of proteins from each groups were separated by SDS-PAGE and transferred onto PVDF membranes. The membranes were incubated with mouse monoclonal anti-ECE-1 antibody (sc-376017, Santa Cruz, CA, USA), mouse monoclonal anti-alpha-enolase (sc-101513, Santa Cruz, CA, USA), mouse monoclonal anti-cathepsin A (sc-73766, Santa Cruz, CA, USA) and mouse monoclonal anti-GAPDH (sc-32233, Santa Cruz, CA, USA) at 1:500 dilution. The immunoblots were developed by incubation with goat anti-mouse IgG-HRP (sc-2005, Santa Cruz, CA, USA) as the secondary antibody followed by ECL detection (GE Healthcare).

Conclusions
In our study, we combined a gel-based 2D-DIGE approach and a dimethyl labeling LC-MS-based shotgun proteomic strategy to identify the proteome expression alterations in U251 gliomaspheres treated with different doses of tachyplesin I. Our results demonstrate complementary advantages of these two techniques. We show that tachyplesin I alters the cellular metabolism, especially glycolysis process and changes the expression of several cytoskeleton proteins and lysosomal acid hydrolases. Moreover, the important role of DNA topoisomerase 2-alpha (TOP2A) in the signal cascades of tachyplesin I was suggested. Further, parallel reaction monitoring (PRM) mass spectrometry confirmed that the major protein of lysosomal acid hydrolases including cathepsin A, cathepsin B and cathepsin D were down-regulated and the possible target-related protein TOP2A was up-regulated by tachyplesin I treatment. In conclusion, we propose that tachyplesin I may down-regulate cathepsins in lysome and up-regulate TOP2A to inhibit migration and promote apoptosis in glioma, thus contributing to its anti-tumor activity. Further work including functional analyses is needed to elucidate the mode of action of tachyplesin I in tumor cells. As far as we know, there is no previous report that reveals the effect of tachyplesin I on proteome of gliomaspheres and our findings imply that tachyplesin I could serve as a promising candidate in the combined therapy against glioma.
Supplementary Materials: The following are available online at www.mdpi.com/1660-3397/15/1/20/s1, Table S1: Total identified peptides information in the forward dimethyl labeling experiment, Table S2: Total identified peptides information in the reverse dimethyl labeling experiment, Table S3: Total identified proteins in the forward dimethyl labeling experiment, Table S4: Total identified proteins in the reverse dimethyl labeling experiment, Table S5: Transitions obtained in Parallel Reaction Monitoring (PRM).