Introduction
It has been known for decades that most tumor cells and tissues enhanced glucose metabolism by glycolysis [
1,
2]. Although its causal relationship with cancer cell proliferation is still unclear, the phenomenon has been developed a reliable technique for detecting and classifying tumors by fluorodeoxyglucose positron emission tomography (FDG-PET) [
3,
4]. In recent years, this metabolic alteration of malignant cells has been observed in multiple cancer cells, and it has become an important aspect for design of anticancer drugs that inhibits glycolysis and other relevant metabolic processes. Several small molecules, including 2-deoxyglucose, lonidamine, 3-bromopyruvate, imatinib and oxythiamine (OT), have shown the effectiveness in anticancer activity in vitro and in vivo [
5‐
16]. They are currently in the clinical and pre-clinical phase. Some other compounds also exhibit potential anticancer activity by modulating glucose metabolism [
16].
OT is a thiamine antagonist and inhibits transketolase (TK) which is an enzyme of the pentose phosphate pathway in animals. As transketolase reaction plays a vital role of the pentose phosphate pathway, inhibition of transketolase will suppress the pentose phosphate pathway and interrupt the synthesis of these important coenzymes ATP, CoA, NAD(P)+, FAD, and genetic material, RNA and DNA in cancer cells. OT can suppress the nonoxidative synthesis of ribose and cause cell apoptosis by inducing a G1 phase arrest
in vitro and
in vivo
[
14,
15,
17,
18]. Although the exactly molecular mechanism is not clear, it has been accepted that the decreased biological macromolecular synthesis can inhibit cell proliferation and induces cell apoptosis. Therefore, these features of metabolism are actually used for cancer therapeutic approach known as “metabolic therapy” [
19,
20].
In the present study, a dynamic proteomic method was adapted to analyze the effects of antimetabolite OT on dynamic changes of protein expression in pancreatic cancer cells, thus to understand the molecular mechanism underlying antimetabolite interference.
Materials and methods
Chemicals and regents
15 N enriched algal amino acid mixture (15 N enrichment, 98%) was purchased from Cambridge Isotope Laboratory Inc. (Andover, MA). Fetal bovine serum (FBS) was purchased from Irvine Scientific (Santa Ana, CA). Dulbecco’s modified Eagles’s medium and antimycotic were from Gibco (Calsbad, CA). Sequence grade trypsin solution was from Promega (Madison, WI).
Acetonitrile was purchased from Thermo Fisher Scientific (Rockford, IL). Materials employed for gel electrophoresis were purchased from BioRad. Water was prepared using a Milli-Q system (Millipore, Bedford, MA). Other chemicals employed were purchased from Sigma (St. Louis, MO). This project was approved by Creighton University Institutional Review Board.
In vitro cytotoxic activity
The cell cytotoxicity of OT against the MIA PaCa-2 cells was determined by MTT assay [
21,
22]. The cells at exponential phase were dispensed in 96-well plates at a density of 1 × 10
4 cells per well. The cells were stimulated with different concentrations of OT for 2 days. The cells were then incubated in 20 μl MTT (3-(4, 5-dimethylthiazol-2-yl)-2, 5-diphenyl tetrazolium bromide) (Sigma, USA) in growth medium at 37°C for 4 h lysed in 100 μl of dimethyl sulfoxide (Sigma, USA) for 10 min. The absorbance in each well was measured at 490 nm by an ELx800 Absorbance Microplate Reader (Biotek, CA). The cell viability and IC
50 value were calculated by the following equations: cell viability = mean optical density of experimental group/mean of the control × 100%; IC
50 value = concentration of OT at 50% cell viability.
Cell culture
Human pancreatic carcinoma cell line MIA PaCa-2 was maintained in MEM supplemented with 10% fetal bovine serum and 1% antibiotic antimycotic at 37°C in 5% CO2 until 80% confluence when the experiment started [
23,
24]. Experiments were set up in two groups: dose- and time-dependent groups. For the dose-dependent group, the cells were stimulated with 5, 50 and 500 μM OT for 48 hours, respectively. The unstimulated cells were considered as control. For the time-dependent group, the cells were stimulated with 50 μM OT in MEM containing natural amino acids or 50% of
15 N algal amino acid mixture (
15 N enrichment, 98%) for 12 and 48 h. The unstimulated cells were considered as the zero time point. Each treatment was repeated four times with 10 mL/flask. The cell pellets were then collected for further analysis.
Protein sample preparation
The cell pellets were immediately washed three times with ice-cold PBS. Cells were harvested in 2-DE lysis buffer with protease inhibitor set III and phosphatase inhibitor set II (Calbiochem, La Jolla, CA). The suspension was sonicated at 100 Watt for 3 × 5 s and centrifuged at 20,000 × g for 30 min. Protein concentration was measured by Bradford assay using bovine serum albumin as the standard. The samples were stored at -80°C until analysis.
Two-Dimensional Gel Electrophoresis (2-DE)
Two-DE was performed as previously described [
23,
24]. Briefly, five hundred micrograms of proteins were mixed with a rehydration solution (Bio-Rad, Hercules, CA) containing 7 M urea, 2 M thiourea, 4% CHAPS, 50 mM DTT, 0.2% biolyte 3–10, 0.1% biolyte 4–6, and 0.1% biolyte 5–8 and a trace of bromophenol blue to a total volume of 300 μL. The mixtures were pipetted into IPG strip holder channels. After 14 h of rehydration, the strips, pH 3–10 NL, were transferred to the isoelectric focusing (IEF) holders (Bio-Rad, Hercules, CA). Prefocusing and focusing were performed on the IPGphor platfor (Bio-Rad, Hercules, CA) (500 V hold 2.5 h, linear 500–1000 V increase 1 h, 1000 V hold 1 h, linear 1000-8000 V increase 1.5 h, and 8000 V hold 60,000 KV h). Following IEF separation, the gel strips were equilibrated twice for 15 min each with equilibration buffer I and II (37.5 mM Tris-Cl, pH 8.8, 20% glycerol, 2% SDS, 6 M urea, with 2% DTT in buffer I and 2.5% iodoacetamide in buffer II, respectively). The equilibrated gel strips were then placed onto 8-16% Tris–HCl gel, and sealed with 0.5% agarose in a Protean Plus Dodeca cell (Bio-Rad, Hercules, CA) until the bromophenol blue reached the bottom of the gels.
After 2-DE, the gels were stained with Pro-Q Diamond [
25,
26]. Then the gels were stained using SYPRO-Ruby (Molecular Probes, Eugene, OR) or visualized with the Coomassie Brilliant Blue R-250 (Merck, Germany) overnight at room temperature. Following 2-DE and protein staining, stained gels were scanned with a Pharox FX molecular imager (Bio-Rad) with a 532 nm laser excitation and a 580 nm band-pass emission filter. Spot detection, quantification and matching were identified using PDQuest 8.0 software (Bio-Rad). The intensity of each protein spot was normalized to the entire gel intensity of all spots detected. Quantitative analysis was performed using the Student’s t-test. The confidence level was 95%. Only those proteins of intensity difference > 2-fold change were selected for MALDI-TOF/TOF MS.
In-gel Trypsin digestion
Protein spots of interest were excised from the gels and in-gel digested with trypsin as previously described [
27]. Briefly, gel pieces were destained with 100 mM ammonium bicarbonate in 30% ACN and dried in a vacuum centrifuge. Ten ng of modified trypsin (Promega, Madison, WI) in 25 mM ammonium bicarbonate was added, followed by incubation 20 h at 37°C. The supernatant was collected, and then the peptides were further extracted three times from the gel pieces with 0.1% trifluoroacetic acid (TFA), 60% ACN with vortexing for 45 min at room temperature. Peptides extracts were vacuum-dried.
MALDI-TOF-MS
For mass spectrometric analysis, the peptides extracts were brought up in 10 μL of 0.1% TFA and cleaned using C18 ZipTip (Millippore, MA). Typically, 2 μL of a-cyano-4-hydroxycinnamic acid (HCCA) matrix in 50% ACN/0.1% TFA was used to elute peptide onto the ground steel plate (Bruker, Germany). The internal standard from Bruker Bruker (MH1: angiotensin II, 1046.5420 Da; angiotensin I, 1296.6853 Da; substance P, 1347.7361 Da; bombesin, 1619.823 Da; ACTH clip 18–39, 2465.199 Da) were used for mass scale calibration. The resulting peptides were extracted and analyzed by MALDI TOF/TOF mass spectrometer (Ultraflex III, Bruker, Germany) in the reflector mode and for sequence analysis in the “lift” mode.
Protein identification and spectral data analysis
The MS/MS spectrum from MALDI measurements were then searched against the Mus musculus subset (16235 sequences) of UniProt KB/Swiss-Prot/TrEMBL database (database version 57.15; 515203 sequences) using the Mascot v 2.2 search program (Matrix Science, London, United Kingdom) (
http://www.matrixscience.com). Search parameters for the database search with Mascot were set as follows: enzyme, trypsin; allowance of up to one missed cleavage peptide; fixed modification parameter, carbamidomethylation (C); variable modification parameters, oxidation (at Met); mass tolerance for precursor ions was ±1.2 Da; mass tolerance for fragment ions, ±0.6 Da.
Mascot scores of proteins or peptides were used for protein identification (p < 0.05). In the case of peptides matching to multiple members of a protein family, the positive identified protein was selected based on both the highest score and the highest number of matching peptides. These peaks were externally calibrated with peptide standards from Bruker (MH1: angiotensin II, 1046.5420 Da; angiotensin I, 1296.6853 Da; substance P, 1347.7361 Da; bombesin, 1619.823 Da; ACTH clip 18–39, 2465.199 Da).
The synthesis rates of the differential proteins were calculated according to our in-house algorithms [
24,
25]. One-way ANOVA with the Tukey’s adjustment was used for multiple comparisons in SPSS 13.0 (SPSS Inc., Chicago, IL).
Pathway analysis
Ingenuity Pathway Analysis (IPA) (Ingenuity Systems, Inc., Redwood City, CA,
http://www.ingenuity.com) was used for pathway, network and functional analyses of differential proteins in the present study.
K-means clustering
Protein ratios were transformed to the log scale (base 2) before clustering [
28]. The Cluster 3.0 freeware software package was used for clustering analysis (
http://bonsai.hgc.jp/~mdehoon/software/cluster/software.htm). Repeated (10–100) K-means clustering of proteins was based on Pearson correlation coefficient of their expression profiles [
29].
Western blotting analysis
Western blotting analysis was performed as described previously [
29]. Briefly, after SDS-PAGE separation, proteins were then transferred to PVDF membranes (Millipore, CA) according to the manufacture’s protocol, and antibody labeling was visualized using ECL reagent (Pierce Biotech Inc., Rockford, IL). Western blot score was a fraction of β-actin or β-tubulin, and measured in Quantity One (Bio-Rad).
To examine whether Annexin A1 is expressed in pancreatic cancer, we separately tested a set of serum samples from patients (pancreatic tumor; pT2/pT3) and healthy volunteer subjects with age-matched. The patient serum samples (n = 7) and the healthy volunteer blood samples (n = 12) were collected by City of Hope National Medical Center and NCI-designated Cancer Center (Duarte, CA) with proper informed consent according to a protocol approved by the Institute Review Board. All samples used in this study were further approved by the Institutional Review Board at Creighton University. The blood samples in BD Vacutainer® Blood Collection Tubes (BD Ventures, L.L.C., NJ) were fractionated by centrifuging at 1,000 × g for 10 min. The serum samples were immediately divided into aliquots and frozen at -80°C. The mean (±SD) age for the tumor patients was 61.3 (±8.1) years, and for the healthy volunteer group, 60.3 (±5.4) years. We measured serum levels of Annexin A1 by using Western blotting analysis. All the experiments were performed in triplicates.
Discussion
Cancer cells utilize glucose maximally as a main source of energy supply and substrates for proliferation through glycolytic metabolic pathways [
1,
2]. Inhibition of the activity of the key enzymes (e.g., transketolase/transadolase) in these metabolic networks, resulting in significant limitation of glucose utilization, provides an ideal strategy for an effective therapy of cancer. A number of our previous studies have shown that inhibition of activity of either transketolase in the pentose phosphate cycle, or glycogen phosphorylase causes cell cycle arrest leading to cancer cell apoptosis [
17,
23,
47,
48]. In this study, we found that transketolase inhibitor OT altered dynamics of cellular protein expression in MIA PaCa-2 cells by interrupting the rates of protein
de novo synthesis. This study provides 1) an important clinical implication for identifying novel cellular protein signals/targets that are associated mechanistically with cancer treatment; 2) a novel approach for detecting signal molecules that initiate drug resistance.
Small molecule antimetabolites are among the more effective chemotherapeutic agents in use today. Currently, gemcitabine, 5-fluorouracil (5-FU), and imatinib, are commonly used for the treatment of pancreatic cancer [
49‐
51]. However, the response rate to either gemcitabine, or imatinib, and patient survival, are poor [
52,
53]. There is an urgent need to discover additional chemotherapeutic targets such as metabolic enzymes that play a crucial role in controlling the growth of cancer cells. In this study, we found that OT caused protein expression in a time dependent fashion. Peroxiredoxin-6 of cluster 1, which can suppress TRAIL-mediated cell death in human cancer cells by binding to death effector domain caspase [
54], was constantly down-regulated by the duration of OT treatment (from 0 hr to 48 hrs) (Figure
2B). It implicated that OT induced cell death by hperoxiredoxin-6 related TRAIL-induced pathway. However, peroxiredoxin-2 and peroxiredoxin-4 of cluster 2, the same ubiquitous family of peroxiredoxin-6, which were up-regulated in many cancers [
38‐
40], were shown in an upright “V” shape (Figure
2B). It suggested that the two proteins might be the only early response molecules upon OT-treatment comparing with peroxiredoxin-6. Calreticulin of cluster 3 was shown in a downright “
V” shape (Figure
2B). It suggested calreticulin might have an opposite function of OT-induced cell apoptosis as the early time response molecular comparing with peroxiredoxin-2 and peroxiredoxin-4. The protein expression pattern in three clusters suggested that metabolic dynamic changes (dynamic changes of transketolase activity) in MIA cancer cells in response to OT treatment caused dynamic changes of cellular protein signals. Protein expression pattern indicates that dynamics of these protein expressions differs in MIA cells in response to OT treatment. Interestingly, expression of these proteins in cluster 2 and 3 was no significant difference in MIA cells treated with OT for 48 hrs, but significantly changed in MIA cells treated with OT for 12 hrs, compared to that in MIA cells at basal line (0 hrs). Early response proteins (e.g., Glyceraldehyde-3-phosphate dehydrogenase, S100A8) in clusters 2 and 3 may play an important role in MIA cells against OT treatment. Several studies revealed that expression of both Glyceraldehyde-3-phosphate dehydrogenase and S100A8 are suppressed in Raw264.7 cells [
25,
26,
55]or bone cells [
56,
57] in response to toxins or tobacco smoke, suggesting that these proteins play an important role in cell defense (survival). Suppression of the protein expression may be part of cell emergent response mechanism since OT treatment cut off supplies of substrates and energy for cancer cell proliferation.
Because of a wild range of protein concentration in cells, it is the most difficult to study in a truly comprehensive manner. Standard proteomics usually compares amounts of proteins in cells in two different states (e.g. disease vs. normal) or conditions (e.g. treatment vs. non treatment) [
58]; it does not address the dynamics of the proteome in the different biological states that are being compared, nor does it provide information about the mechanisms whereby the system changes from one state to the other. Thus, data obtained from this study may provide unique cell survival mechanism.
Previous study showed that dynamic changes of metabolic enzyme activity determined the metabolic sensitivity of cancer cells to the treatment [
47], therefore, the early responsive protein signals upon OT treatment may be indicatives for the sensitivity of pancreatic cancer cells to the treatment in molecular level.
The dynamic changes of the cellular molecules (mRNAs, proteins, and metabolites) depend upon the physiological, developmental, or pathological state of living cells [
59]. A change in the proteome may be the most important outcomes of a cellular response, such as autophagy, to exogenous stimuli. Autophagy is a constitutive, catabolic process leading to the lysosomal degradation of cytosolic proteins and organelles. Dynamic changes of proteins identified in Clusters 3 and 4 may reflect the cellular autophaging phenomena.
Because of a wild range of protein concentration in cells, it is the most difficult to study changes of protein expression in a truly comprehensive manner. Standard proteomics usually compares amounts of proteins in cells in two different states (e.g. disease
vs. normal) or conditions (e.g. treatment
vs. non treatment) [
58]; it does not address the dynamics of the proteome in the different biological states that are being compared, nor does it provide information about the mechanisms whereby the system changes from one state to the other. This study provides more dynamic information of cellular protein signals than our previous studies and others [
23], which some dynamic information of proteins in clusters 2 and 3 may be missed (or so-called “false negative” errors) when standard proteomics approach is used in our previous studies.
Protein turnover is the balance between protein synthesis and protein degradation (or breakdown), which is believed to decrease with age in all senescence organisms including humans. This results in an increase in the amount of damaged protein within the body. It is unknown if this is a cause or consequence of aging but it seems likely that it is in fact both. The damaged protein results in a slower protein turnover which then results in more damaged protein causing an exponential increase in damage to all protein within the body and to aging. Protein turnover is being considered as a missing dimension in proteomics for biomedical research [
59]. The dynamics of protein turnover is one of key features to the understanding of regulation of protein expression and protein-protein interaction in cells [
24,
60]. The level of expression of a protein depends on the rates of its synthesis and degradation. Thus the turnover of a protein is an important indicator of its functional significance in cells. Despite its evident importance, the role of protein turnover has not previously been considered in analyses of the proteome. Protein turnover can be quantified on a protein-by-protein basis. With the established method [
24], in this study we were able to quantitatively measure the rates of newly synthesized proteins. Among 41 proteins measured, 7 proteins are with a turnover rate of < 45%, 5 proteins are with a turnover rate of > 65%, and 29 proteins are with a turnover rate between 45% and 65%. The turnover rates of the proteins with extreme high or low levels are related to specific status of cell physiology (e.g., suppression of metabolic signaling).
Intriguingly, we did not observe
de novo synthesized peptides in MIA cells treated with OT in 12 h. This is reasonable since it takes more than 24 hr for a specific gene translating to the cognate protein. The explanation for detection of the differential expressed proteins at 12-h treatment may be 1) because OT directly or indirectly activated pathways of degradation resulting in rapid degradation of cellular functional proteins; 2) because OT directly or indirectly turned on pathways of posttranslational modification leading to the increased amount of proteins. This notion can be demonstrated from Figure
6B/C, obviously,
de novo synthesis of proteins cannot be detected.
Previous study revealed that some active metabolic pathways, TCA, glycolysis, oxidative phosphorylation and the pentose phosphate pathway, were interconnected with the critical signaling pathways in proliferating cells [
61]. Inhibiting of these metabolic pathways could affect the biological macromolecular synthesis and suppress cell proliferation [
23,
47]. In previous studies it had been shown OT could cause the inhibition of nucleic acid synthesis metabolic through increasing imbalances in pentose phosphate cycle [
18]. In our study, the expression profile patterns of cellular phosphorylated proteins of MIA PaCa-2 were significantly inhibited by OT treatment (Figure
3). This phenomenon might be caused by inhibition of biological macromolecular synthesis or some critical signaling pathways related phosphorylation.
Certainly, this is a principal proof study examining whether interference of the interactive metabolic and cell signaling pathways alters expression of protein signals associated with cellular abnormal activity of protein turnover. Although information obtained in a single pancreatic cancer cell line MIA PaCa-2 may be limited to a study in other cell lines, protein signals identified in this study may provide useful information for further development of novel biomarkers and/or drug targets for administration of pancreatic cancer.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
JW and DM, carried out rat experiments. DM, drafted the manuscript. GGX were heavily involved in experimental design, and also mainly involved in scientific correction of the draft manuscript. JX, YZ, and XZ were involved in sample collection and measurements of proteins, WNL, VLG, QW, YY, and RR, were involved in project discussion. All authors were involved in drafting the manuscript and revising it for critically important content. All authors have read and approved the final manuscript.