Hypoxia Modulates A431 Cellular Pathways Association to Tumor Radioresistance and Enhanced Migration Revealed by Comprehensive Proteomic and Functional Studies*

Tumor hypoxia induces cancer cell angiogenesis, invasiveness, treatment resistance, and contributes to poor clinical outcome. However, the molecular mechanism by which tumor hypoxia exerts a coordinated effect on different molecular pathways to enhance tumor growth and survival and lead to poor clinical outcome is not fully understood. In this study, we attempt to elucidate the global protein expression and functional changes in A431 epithelial carcinoma cells induced by hypoxia and reoxygenation using iTRAQ quantitative proteomics and biochemical functional assays. Quantitative proteomics results showed that 4316 proteins were quantified with FDR<1%, in which over 1200 proteins were modulated >1.2 fold, and DNA repair, glycolysis, integrin, glycoprotein turnover, and STAT1 pathways were perturbed by hypoxia and reoxygenation-induced oxidative stress. For the first time, hypoxia was shown to up-regulate the nonhomologous end-joining pathway, which plays a central role in DNA repair of irradiated cells, thereby potentially contributing to the radioresistance of hypoxic A431 cells. The up-regulation of Ku70/Ku80 dimer, a key molecular complex in the nonhomologous end-joining pathway, was confirmed by Western blot and liquid chromatography/tandem mass spectrometry-MRM methods. Functional studies confirmed that up-regulation of glycolysis, integrin, glycoprotein synthesis, and down-regulation of STAT1 pathways during hypoxia enhanced metastastic activity of A431 cells. Migration of A431 cells was dramatically repressed by glycolysis inhibitor (2-Deoxy-d-glucose), glycoprotein synthesis inhibitor (1-Deoxynojirimycin Hydrochloride), and STAT1α overexpression that enhanced the integrin-mediated cell adhesion. These results revealed that hypoxia induced several biological processes involved in tumor migration and radioresistance and provided potential new targets for tumor therapy.

Tumor hypoxia induces cancer cell angiogenesis, invasiveness, treatment resistance, and contributes to poor clinical outcome. However, the molecular mechanism by which tumor hypoxia exerts a coordinated effect on different molecular pathways to enhance tumor growth and survival and lead to poor clinical outcome is not fully understood. In this study, we attempt to elucidate the global protein expression and functional changes in A431 epithelial carcinoma cells induced by hypoxia and reoxygenation using iTRAQ quantitative proteomics and biochemical functional assays. Quantitative proteomics results showed that 4316 proteins were quantified with FDR<1%, in which over 1200 proteins were modulated >1.2 fold, and DNA repair, glycolysis, integrin, glycoprotein turnover, and STAT1 pathways were perturbed by hypoxia and reoxygenation-induced oxidative stress. For the first time, hypoxia was shown to up-regulate the nonhomologous end-joining pathway, which plays a central role in DNA repair of irradiated cells, thereby potentially contributing to the radioresistance of hypoxic A431 cells. The up-regulation of Ku70/Ku80 dimer, a key molecular complex in the nonhomologous end-joining pathway, was confirmed by Western blot and liquid chromatography/ tandem mass spectrometry-MRM methods. Functional studies confirmed that up-regulation of glycolysis, integrin, glycoprotein synthesis, and down-regulation of STAT1 pathways during hypoxia enhanced metastastic activity of A431 cells. Migration of A431 cells was dramatically repressed by glycolysis inhibitor ( Tumor hypoxia arises when solid tumor expands too rapidly without a concomitant expansion of the supporting vasculature (1). Tumor hypoxia induces resistance to anticancer therapeutics, angiogenesis, genomic instability, invasion/metastasis, and overall poor clinical outcome (2,3). Given its importance in tumor development, the molecular pathways perturbed by tumor hypoxia are considered attractive targets to be exploited in oncology (4).
Extensive laboratory studies and clinical data have shown that hypoxic conditions contribute to therapeutic resistance in radiotherapy and limit the response of tumor cells to radiation therapy (5)(6)(7). DNA damage by irradiation is kept from repair with the presence of oxygen, but it can be repaired by hydrogen without oxygen under hypoxia and thus leads to increased radioresistance (7). Other biological molecules have also been reported to be involved in the radioresistance in tumor cells under hypoxic conditions (8). A major cellular response to hypoxia is the activation of hypoxia induced factor 1 (HIF-1) 1 , a transcription factor, which has been implicated in the regulation of tumor radiosensitivity (9). HIF-1␣ expression in uterine cervical cancer tissues positively correlates with the adverse effects of radiotherapy, and its expression in cervical cancer after radiotherapy is associated with increased risk for tumor-related death (10,11). DNA repair pathways are vital for cellular protection against radiation. Recent studies showed that the activation of DNA repair pathways in some tumors contributed to intrinsic resistance to radiotherapy (12,13). However, several reports have demonstrated that hypoxia induces the repression of DNA repairing genes at the mRNA level, for example, RAD51, BRCA1/2, XRCC3/4, Ku70, and Ligase IV, but the changes in mRNA level do not always correlate with a subsequent decrease in protein level (14 -17). These data thus highlight the importance of determining whether some of the functional consequences of hypoxia were mediated by DNA repair proteins.
Laboratory and clinical studies have also indicated that tumor hypoxia has been implicated in enhancing tumor cell metastatic potential (3, 18 -22). Metastasis is the major characteristics of malignant tumors and the main cause for cancer-related mortalities. As such, there have been major efforts to elucidate the molecular mechanisms underlying the distinct steps of cancer metastasis (23)(24)(25)(26). Some key molecules, such as adhesion molecules and MMPs, play important role in cancer cell metastasis (27). It has been reported that HIF-1 modulates the expression level of proteins, such as MMPs, plasminogen activator inhibitor (PAI-1), tissue factors, CapG, S100A4, filamentation 1, cadherin, and integrin alpha 5 (2, 28 -30), which play important roles in regulating the invasion or metastatic potential of tumor cells.
Hypoxia is now a well recognized cause of radioresistance and metastasis but its precise role in these processes is still poorly defined (24,31). A better understanding of the mechanisms behind this pathophysiology will lead to a more specific and efficient therapeutic outcome. The knowledge of hypoxia-regulated proteins from proteomic investigation will provide a better and global understanding of the molecular pathways perturbed by hypoxic tumor and give rise to novel biological insights and concepts for exploiting this factor (7).
Proteomic studies to elucidate the molecular events elicited by hypoxic stress (3,(32)(33)(34) are less extensive and have, to date, been limited to the use of 2-DE based methods (35). To further our proteomic knowledge of hypoxia-induced tumor evolution, shotgun based isobaric tag for relative and absolute quantification (iTRAQ) quantitative proteomics, which provides a global assessment of the proteins modulated, was used to analyze the cellular proteome changes of A431 epithelial carcinoma cells induced by hypoxia and reoxygenation. More than 4300 proteins with unused Protscore Ն1.68 and FDRϽ1% were identified, in which over 1200 proteins were significantly modulated by hypoxia (Supplemental Table S1). Further analysis of the quantitative proteomics data set indicated that DNA repair, glycolysis, integrin, glycoprotein turnover, and STAT1 pathways were perturbed by hypoxia. Most of them as listed in Supplemental Table S2 were confirmed and validated using Western blot (WB) or LC-MS/MS-multiple reaction monitoring (MRM) methods. Functional studies revealed the roles of the glycolysis, integrin, glycoprotein syn-thesis and STAT1 pathways in hypoxia-induced cancer metastasis.
Cell Culture and Hypoxic Condition-The A431 epithelial carcinoma cells were purchased from ATCC and maintained in Dulbecco's modified Eagle's medium (DMEM) with 10% fetal bovine serum. For all experiments, 2 ϫ 10 6 cells were seeded on 10 cm dishes in DMEM and incubated under 5% CO 2 at 37°C. On reaching 30 -40% confluency, the cells were washed with 1 ϫ PBS for three times and serum-free media for three times before incubating in serum-free media. They were subjected to normoxia (Nx: 21% O 2 /5% CO 2 ) for 72 h or hypoxia (Hx: Ͻ0.1% O 2 /5% CO 2 ) in a hypoxia chamber for 36 h or 72 h. For Reoxygenation (ReOx), the cells were exposed to Hx for 48 h, changed to a fresh medium, and then exposed to Nx condition for an additional 24 h. At least five independent biological replicates were pooled for the analyzed sample in each condition.
Sample Preparation, iTRAQ Labeling, and LC-MS-The cellular proteins were dissolved in 8 M urea in 50 mM triethylammonium bicarbonate (TEAB) (pH 8.5) solution with protease inhibitor mixture (1:50). Protein concentrations were measured using BCA assay. Two hundred micrograms of proteins from each condition were reduced with 5 mM TCEP at 30°C for 1 h, alkylated with 10 mM methyl methanethiosulfonate (MMTS) at room temperature, in the dark for 45 min. The samples were diluted to 1 M urea using 50 mM TEAB and digested by trypsin in a 1:50 mass ratio at 37°C overnight. The solution containing tryptic peptides was adjusted to pH 2-3 with TFA and desalted using Sep-Pak C18 cartridge (Waters, Milford, MA). The peptides were dried using speedvac concentrator and, dissolved in 1 M TEAB (pH 8.5) and labeled with the isobaric tags (Applied Biosystems, Foster City, CA) according to manufacturer's protocol as followed: the Nx-treated sample was labeled with 114; the samples under Hx for 36 h or 72 h were labeled with 115 or 116 respectively; and the ReOx-treated sample was labeled with 117.
The labeled sample was fractionated by a PolySULFOETHYL A column (PolyLC, Columbia, MD; 4.6 ϫ 200 mm, 5 m particle size, 200-Å pore size) using Shimadzu Prominence UFLC system (Kyoto, Japan). Fractionation was performed with a 60-min linear gradient of 0 -500 mM KCl (10 mM KH 2 PO 4 , pH 3.0, 25% acetonitrile) at a flow rate of 1 ml/min. A total of 25 fractions were collected and desalted using Sep-Pak C18 cartridges (Waters). Each fraction was dried and reconstituted in 30 l 3% ACN and 0.1% formic acid for LC-MS/MS analysis. All fractions were analyzed using a QSTAR Elite mass spectrophotometer coupled with an on-line Tempo nano-MDLC system. The sample was injected twice. For each injection, 15 l iTRAQ labeled peptide mixture was injected and separated on a homepacked nanobored C18 column with a picofrit nanospray tip (75 m inner diameterϫ15 cm, 5 m particles) (New Objectives, Woburn, MA). The detailed parameters were set as previously described (3).
Database Searching and Criteria-Protein identification and quantification were performed using ProteinPilot TM software 2.0.1 with Revision number 67476 (Applied Biosystems) by searching the combined raw data from the two runs of labeled sample against the concatenated "target" (Uniprot human database, downloaded on 12 March 2010, including 95624 sequences and 36307192 residues) and "decoy" (the reverse amino acid sequence for false discovery rate (FDR) estimation) databases. The Paragon and Pro Group algorithms in the ProteinPilot software were used for peptide identification and isoform-specific quantification. User-defined parameters were as follows: (i) sample type, iTRAQ 4plex (peptide-labeled); (ii) cysteine alkylation, MMTS; (iii) digestion, trypsin; (iv) instrument, QStar Elite ESI; (v) special factors, none; (vi) species, none; (vii) specify processing, quantitative; (viii) ID focus, biological modifications; and (ix) search effort, thorough ID. Two missed cleavage sites are permitted and nonspecific cleavage sites are prohibited. The Mass tolerance of precursor ions and fragment ions is both 0.2 Dalton.
To minimize false positive results, a strict cutoff for protein identification was applied with FDRϽ1% as listed in Supplemental Table  S1. The resulting data set was autobias-corrected to get rid of any variations imparted because of unequal mixing during combining different labeled samples. For iTRAQ quantitation, the peptide for quantification was automatically selected by the Pro Group algorithm (at least one peptide with 99% confidence) to calculate the reporter peak area, error factor (EF), and p value. The proteins for further analysis were filtered with p value Ͻ0.05, and 1.2-fold changes (Ͼ1.20 or Ͻ0.83) relative to the normoxic control unless the proteins were known to be of interest from our functional assay. The list of proteins used is shown in Supplemental Table S2.
Semi-quantitative RT-PCR and Western Blotting-Total cellular mRNAs were isolated using TRIzol reagent (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions. cDNA was reversely transcribed from 2 g total RNA in a final volume of 20 l using RTase (Promega, Madison, WI, USA). Primer sequences were listed in Supplemental Table S3. PCR was performed with KAPA Fast Taq polymerase in a DNA thermal cycler (PTC-200, Bio-Rad, Hercules, CA) according to a standard protocol as follows: 1 cycle of 95°C for 1 min; 28 cycles of 95°C for 10 s, annealing at 57°C for 10 s, 72°C for 1 s, a final extension at 72°C for 30 s and holding at 4°C. The amount of cDNA used for each PCR reaction was 20 ng in a 25 l reaction volume. The PCR products (5 l) were analyzed by electrophoresis through 2% agarose gels and visualized with ethidium bromide staining.
Before Western blot analysis, protein concentration of cell lysates dissolved in 1% SDS in 40 mM Tris-HCl (pH 8.0) was quantified using BCA assay, equal protein amount from each sample was separated on 12% SDS-PAGE. For native gel, the solution of 1% Triton X-100 in 20 mM Tris-HCl (pH 7.4), 150 mM NaCl with complete protein inhibitor mixture (1:50) was used in the extraction of protein complexes. Proteins were then transferred to nitrocellulose membrane and immunoblotted using antibodies against STAT1, p-STAT1, Ku70, Ku80, ITGA5 and ITGB1 respectively. The probe proteins were detected with Invitrogen ECL system according to the manufacturer's instructions.
Dynamic MRM Quantification-The same amount of proteins from both normoxic and hypoxic A431 cells were loaded and separated on 12% SDS-PAGE. Based on the proteins molecular weight, each lane was manually excised into six fractions and in-gel digested with trypsin. The peptides were loaded onto 6490 Triple Quadrupole LC-MS/MS system coupled with Infinity 1260 liquid phase separation system (Agilent Technologies Inc, Singapore). The information of all peptides and their transitions of target proteins were obtained from Peptide Atlas (www.peptideatlas.org) database and showed in Supplemental Table S4. All samples were subjected to conventional MRM analysis for confirmation of retention times of peptides. Then dynamic MRM was used to quantify the protein changes.
The mobile phase consisted of solvent A, 0.1% aqueous formic acid and solvent B, acetonitrile with 0.1% formic acid. Peptides were separated by Agilent chip (Large capacity chip II) packed with Zorbax 300SB C18 resin (5 m, separation: 150 mm ϫ 75 m; enrichment: 9 mm, 160 nL). Injections were 2 l of a sample solution and were followed by a 10 min wash period with 100% A, then by elution with a gradient of 0 -10% solvent B in 3 min, 10 -35% solvent B in 47 min, 35-80% in 2 min, 80 -82% solvent B in 3 min and followed by 3% solvent B in 5 min.
LC-MS/MS-MRM analyses of the peptide were done with an electrospray voltage of 1700 V, capillary temperature 185°C and skimmer offset Ϫ5 V. Both Q1 and Q3 were set at unit resolution (FWHM 0.7 Da) and collision gas (N2) pressure in Q2 was held at 1.5 mTorr. Scan width was 0.004 m/z and scan time was 20 ms for the peptide. Collision energy for each peptide was calculated based on the equation: CE ϭ (((m/z)/100.0)*3.6 -4.0), in which the m/z is the mass to charge ratio of the precursor ion. Peak areas for each peptide were extracted and integrated using Agilent Qualitative Analysis B.04.00 software.
Construction and Transfection of STAT1␣ Expression Vectors-The full length of STAT1␣ was obtained from A431 cDNA library prepared as described above using PCR. The primer information is listed in Supplemental Table S3. The gene with NotI and BamH I enzyme sites were inserted into the expression vector of p3xFLAG-CMV-10 and positive clones were confirmed by plasmid digestion with NotI and BamH I and DNA sequencing. For stable transfection, A431 cells were transfected using 20 l of lipofectamin 2000 (Invitrogen) and 4 g of p3xFLAG-STAT1␣ vector or control p3xFLAG vector according to manufacturer's instructions. Cells were then selected with standard medium with 1 mg/ml G418. 2-3 weeks later, the monoclones were picked up and further cultured in medium with 0.5 mg/ml G418. The positive clones were confirmed by WB using anti-Flag, STAT1 and phospho-STAT1 antibodies.
Cell Proliferation and Cell Adhesion Assay-Twenty-four-well plates were coated with 5 g/ml fibronectin in PBS. Uncoated wells were incubated with 0.5% BSA in DMEM to serve as negative control. Cells were trypsinized and seeded into coated 24-well dish at a density of 2.5 ϫ 10 5 live cells. After incubation for 20 mins to an hour at 37°C, the detached cells were removed by serum-free medium and attached cells were fixed with 95% ethanol for 10 mins and stained with 0.5% crystal violet for 25 mins. The plate was reimmersed in deionized water to remove excess dye and then allowed to air-dry for 5 to 10 mins. The adherent cells were photographed by optical microscopy under 100 ϫ magnification and a total of 200 l of 0.5% Triton X-100 was added to each well to solubilize the cells overnight at room temperature. The absorbance was measured at OD595 using a microplate reader (Tecan Magellan TM , Mä nnedorf, Switzerland). Cell proliferation was assessed by the ability to exclude 5 mg/ml MTT in PBS for 2 h. MTT uptake was quantified at A 540 nm , reference A 630 nm after solubulization in DMSO.
Cell Staining-To visualize ITGA5 distribution in cells, immunofluorescent staining with anti-integrin a5 antibody was performed. A431 cells were seeded into 20 ϫ 20-mM coverslips, treated for the indicated conditions, fixed by 4% paraformaldehyde, and then permeabilized with 0.2% Triton X-100 in PBS for 10 mins. The cells were incubated with 4% BSA in 0.1% Triton X-100 for 20 mins to block nonspecific binding sites before incubation with anti-ITGA5 in 1% Triton X-100 (Millipore; 1:100) for 2 h at 37°C. Subsequently, the cells were washed, and then incubated with Alexa Fluor® 488-conjugated secondary antibody for 1 h at room temperature. After that, cells were washed, mounted, and examined by fluorescence microscope.
Scratch-wound Assay-A total of 2 ϫ 10 6 cells were seeded in a 6-well plate. A wound was made by scratching the cells in a line with a sterile pipette tip in the middle of the well next day. Then cells were washed twice with 1 X PBS and three times with serum free media before incubating in serum-free media. Photographs were captured by a Nikon Eclipse TE2000-U microscope. After incubation in indicated conditions under hypoxia for specific time points shown in Fig.  2, 4, and 6, photographs were taken again. Hypoxic treatment was the same as above mentioned. For 1-DJ wound-scratch assay, the cells were treated by the drug for 48 h and then undergone scratch assay under hypoxia. After hypoxia, the cells were lysed with 1% SDS solution and used for WB to detect their glycoproteins by HRPlabeled lectin. Because 2-DG and 1-DJ are water soluble, no control solvents are used in the experiments. The migration of cells into the wound was recorded and the wound closure was determined.
Metabolite Analysis-For A431 cell metabolite extraction, MeOH/ H 2 O (80:20) solution was used as described in published paper because the solution had been proved to be a more suitable extraction solvent (36). Briefly, after the same amount of A431 cells were treated with or without 2-DG or hypoxia, the growth medium was discarded and cells were rapidly washed three times with cold PBS followed by adding cold MeOH/H 2 O (80:20), which caused protein precipitation and hence cell disruption. Wells were then lightly scraped with a rubber tipped cell scraper. The dishes were washed two times with MeOH/H 2 O solution and the wash and the extract solution were combined for centrifuging at 4°C and 5725 ϫ g for 5 min. The supernatant was collected for dry. All steps were performed on ice. The metabolites were solved in 0.1% formic acid for ESI-Q-TOF analysis. Metabolite levels were normalized to protein content in the pellet after metabolite extraction, which was determined by the 2D-Quant Kit (GE Healthcare, Munich, Germany) (36,37).
The same amount of metabolites normalized with protein level was directly injected to mass spectrometer with 10 l/min flow. Negativeion mode electrospray mass spectra were obtained using an Accurate-Mass Q-TOF LC/MS 6530 (Agilent Technologies). Spray voltage was -1.0 kV and the capillary temperature was set to 185°C. Spectra were collected in full scan mode in a scan range of 50 -400 m/z. The precursors with very close m/z with metabolites in glycolysis pathway were chosen for MS/MS identification. MS/MS spectra were obtained using isolation width 4 m/z, and scan rate 1 spectra/s, with collision energy -9. The m/z and intensity of MS/MS spectra was input to Massbank database for metabolite identification (http://www. massbank.jp/MetaboIdentif.html).
Statistical Analysis-The measurements were expressed as mean Ϯ S.D. The paired t test was used for statistical analysis between two groups. Significant level was set at p Ͻ 0.05.

RESULTS AND DISCUSSION
Proteomic Analysis of Normoxic, Hypoxic, and Reoxygenated A431 Cell Lysate-The iTRAQ reagent labeled peptides were quantitatively profiled by LC-MS/MS using QStar Elite MS. The expression level changes of proteins were indicated by the ratios of labeling tag (N72:114; H36:115; H72:116; H48ϩR24:117). Raw data were analyzed using ProteinPilot 2.0.1 software with autobias correction against the UniProt human database. A total of 4316 proteins were identified with unused score Ն 1.68 and FDR Ͻ1% (Supplemental Table S1). iTRAQ reporter ratios of 1.2 and 0.83 were set as the cutoff of protein changes. Totally 1253 proteins were found responsive to hypoxia. These proteins have at least one ratio with p valueϽ0.05, including 522 up-regulated and 743 down-regulated (Table I and Supplemental table S1). All of the data including wiff raw data from QStar, searched group file from ProteinPilot 2.0.1 and the search results of PeptideSummary and ProteinSummary can be downloaded from Proteome-Commons.org Tranche Network using the following hash (Q85ii60OE5SelJDghF4vp2Gjv/dJndWgdMF8yZlmbqXeYgO-CLVMBsRz840rGKEoyovesJyYL3rmak5baMgCRO5UBWKo-AAAAAAAAZ9Qϭϭ for wiff raw data set A; 1G-135KGPNpϩaETVXL7j5VSBfJDYGNshh6mDkHTpC7eXYzF-Wn5gq1xfx69LP/qHPaOpnyDd37dlBϩauqnmWCSwFSeczQ-AAAAAAAAZ9Qϭϭ for wiff raw data set B; LjkJ-yHVkOJU59YzP26G6Q5ozCw6Iy1XQQfNNGoDZaL8wrbE0e-av3TNatlUL2z3dhsqz17RXpx7v2pgxYiIHBe9MV0I8AAAAAA-AAB0Qϭϭ for searched group file; MAFBxhUOXAz2n-HHgcMZszGUCRewTyg0hoHJ/a5CWtGTr9/ZxYOeYc5-yJgrRW9W8jhJawUjdA9bNfdN0HUiwsXBnzYC8AAAAAA-AACGgϭϭ for PeptideSummary; StQPOjfbpRPzCc3AP4S-CSL6hzckua/anydKvaG0TSvim5DdFDFpXucqϩn6MMpvdn7-CFycO1R6xhLp7lOOx5OTpHSrNcAAAAAAAACDQϭϭ for ProteinSummary).
The high quality iTRAQ quantitative datasets provided invaluable biological insight on hypoxia induced tumor progression. To gain the global insights of these proteins, Ingenuity Pathway Analysis (IPA, Ingenuity® Systems, www.ingenuity. com) platform was used to analyze the data to reveal the molecular and cellular functions and canonical pathways induced by hypoxia stress. All the proteins listed in Supplemental Table S1 were used for IPA. The up-regulated and downregulated proteins were analyzed separately, and thus these proteins were mapped to different function, pathway and toxic lists. The summary of IPA was listed in Supplemental  Fig. S1 including top five function, pathway and tox lists that up-regulated and down-regulated proteins involved. The top changes in function and pathway were those regulating basic cell survival, for example, protein synthesis and degradation, post-transcriptional or translational modification, protein folding and trafficking and energy and material metabolism. The top toxic lists were mainly involved in mitochondrial dysfunction, oxidative stress and HIF signaling, which indicated the impact of hypoxia on cellular functions. The proteins involved in HIF signaling pathways were shown down-regulated after hypoxia because the top one pathway down-regulated under hypoxia is ubiquitination degradation pathway and the increased expression of HIF-1␣ was primarily because of increased protein stabilization via decreased ubiquitination degradation. The pathways with dominant difference in comparison analysis of down-regulated and up-regulated proteins were chosen for further studies (Supplemental Fig. S2 and Supplemental Table S2). The proteins mapped to STAT1 (down-regulated pathway) and N-glycosylation degradation (up-regulated pathway) were illuminated in Supplemental Fig. S3. Hypoxia Activated A431 NHEJ Pathway in DNA Repair-Ionizing radiation causes cell death primarily by DNA double strand breaks (DSBs). The two main pathways involved in the repair of DSBs are homologous recombination (HR) and nonhomologous end-joining (NHEJ) pathways. Of the two, NHEJ pathway generally considered to be an error prone pathway is the dominant repair pathway in radiation-induced DNA damage (38). Our iTRAQ data showed that hypoxia substantially increased A431 proteins involved in DNA repair, especially those responsible for DNA double strand break (DSB) repair in the NHEJ pathway (Fig. 1A). When DNA DSBs occurred, Ku70 and Ku80 bind rapidly with the DNA ends, promoting the inward sliding of Ku70 and Ku80 along the length of DNA, at the same time recruiting PRKDC to assemble a DNA-PK heterotrimer (39). Formation of this complex may also be promoted through interactions with heterodimeric ILF2 and ILF3 (40). Our proteomic data showed that all of these proteins (Ku70, Ku80, PRKDC, ILF2 and ILF3) were up-regulated under hypoxic conditions (Fig. 1A). Subsequent phosphorylation of DNA-PKcs destabilizes and causes the complex to dissociate, allowing the recruitment of several end-processing enzymes including ligase IV, NHEJ1 and Mre11/Rad50/Nbs1 proteins of the MRN complex to DNA. Two of them, NHEJ1 and Rad50 were shown to be up-regulated in our data (39). In addition, the proteins that favor DSBs repair were also upregulated under hypoxia. These include MDC1, SFPQ, NONO, SSRP1, SUPT16H, INTS3, APTX, and PRPF19. MDC1 and SFPQ-NONO heterodimer have been shown to facilitate the DNA repair in NHEJ pathway (41,42). SSRP1-SUPT16H dimers, INTS3, APTX, and PRPF19 also have been proven to stabilize DSB repair (43)(44)(45)(46). The expression level and heterodimer changes of Ku70 and Ku80 were validated by LC-MS/MS-MRM and Western blot (Figs. 1B and 1C). In dynamic MRM quantification results, TBB6 and ACTN1, housekeeping proteins, and ATP5B were used as internal controls as their expression levels assessed by iTRAQ were not modulated by hypoxia. The peptides of TBB6, ACTN1, and ATP5B with several strong transitions as shown in Supplemental Table S4 were quantified in dynamic MRM with ratios of 0.92 Ϯ 0.07, 0.90 Ϯ 0.07, and 0.96 Ϯ 0.09 respectively. The copy numbers of the two peptides from Ku70 with seven transitions showed significant increase to 2.86 Ϯ 0.78 folds under hypoxia stress. The intensity of four peptides of Ku80 under hypoxic condition was also higher (2.05 Ϯ 0.69) than that of peptides under normoxic condition (Fig. 1C and Supplemental Table S4). Their change ratios are much higher than those of Ku70 and Ku80 in iTRAQ quantification results (with ratios 1.1-1.2). In addition, the increase of another key component of DNA-PK heterotrimer, PRKDC1, and other components of the NHEJ pathway (ILF2, NONO, SP16H, PRP19 SFPQ, SSRP1, and APTX) during hypoxia was also verified by MRM method (Fig.  1C and Supplemental Table S4). The slight changes of Ku70 and Ku80 observed from iTRAQ data were further validated by SDS-PAGE WB that confirmed increased expression (with ratios less than 2.03) of both proteins as shown in Fig. 1B (the upper two panels). As formation of Ku70/Ku80 heterodimer are prerequisites for functional activity, protein extracts from the tumor cells were separated on native gels and analyzed for the presence of Ku70/Ku80 heterodimer using antibodies against either Ku70 or Ku80. Both antibodies detected the same single band on the native gel which indicated the Ku70/ Ku80 dimer position (the lowest panel in Fig. 1B). The WB results clearly showed obvious increase of heterodimer under hypoxia. To further confirm the result and avoid the interference from cellular proteins, we isolated chromatin of A431 under normoxia and hypoxia using sucrose gradient, and quantified the chromatin binding proteins using iTRAQ anal-ysis. Result showed that the key proteins in NHEJ pathway, such as Ku70, Ku80, PRKDC1, associated with chromatin increased under hypoxic condition (unpublished data). In 2008, for the first time, a decrease in Ku70 and Ku80 was found to be present in more hypoxic cervical carcinoma tissue by immunohistochemistry; however, the results were not further validated (38). Our quantitative proteomic data for the first time showed that hypoxia increased most components of the NHEJ pathways in A431. As these components have been implicated in mediating ionizing radiation resistance, our data also provided a possible molecular mechanism for hypoxiainduced radiation resistance.
Hypoxia Induced Cell Migration Through Metabolic Switch from TCA Cycle to Glycolytic Pathway-It has been well known that tumor cells exhibit increased rates of glucose uptake and glycolysis (47). Increased glycolysis in tumor cells is regarded as an effect of intratumoral hypoxia. Under hypoxic conditions, cells switch their form of glucose metabolism from the oxygen-dependent tricarboxylic acid (TCA) cycle to glycolysis, an oxygen-independent metabolic pathway. HIF-1 has been shown to activate transcription of genes that encode glucose transporters GLUT1, GLUT3, and all glycolytic enzymes (2,48). In our iTRAQ data, the enzymes that catalyze the four steps in glycolysis pathway, hexokinase (HK), glucose 6-phosphate isomerase (G6PI), fructose-bisphosphate aldolase (ALDOA and FBPA) and biphosphoglycerate kinase (PGK) were all found to be up-regulated after hypoxia. The enzyme that regulates pyruvic acid efflux, lactate dehygrogenase A (LDHA), was also significantly up-regulated under hypoxia ( Fig. 2A). Surprisingly, enzymes that are involved in the TCA cycle such as pyruvate dehydrogenase (PDHB), citrate synthase (CS), isocitrate degydrogenase (IDH2), a-ketoglutarate dehydrogenase (DHTKD1), fumarase (FH), malate dehydrogenase(MDH1) and pyruvate carboxylase (PC) were down-regulated under hypoxia (Fig. 2B). LC-MS/MS-MRM analysis results showed that the enzymes of HK, ALDOA, PGK1, and GPI in glycolysis pathway were also found obviously up-regulated as detected by iTRAQ analysis under hypoxia (Supplemental Table S4). Increased glycolysis has been associated with tumor invasiveness, metastasis and/or resistance to therapies (49). Recently, Cai et al. highlighted this relationship between glycolysis and metastasis in cancer cells by knocking down LDHA or overexpressing PDHB to force pyruvate into TCA cycle rather than the glycolysis process. This inhibited cellular growth and migration of gastric cancer cells (50). Our iTRAQ and MRM quantification data both indicated the increase of LDHA and the decrease of PDHB after hypoxia treatment (Figs. 2A, 2B, and Supplemental Table S4). In support of this link between glycolysis and metastasis, Izumi et al. also reported that knockdown of monocarboxylate transporters (MCT1 and MCT4) led to reduced invasiveness of human lung cancer cells (51). Because MCT1 and MCT4 transport monocarboxylate such as lactate across membranes and are important in maintaining continuous glycolysis, their knockdown would result in reduced glycolysis. In view of this, the up-regulation of monocarboxylate transporter, SLC16A1 (MCT1) under hypoxic conditions as observed in our study ( Fig. 2A) would enhance continuous glycolysis. In addition to MCT1, glucose transporter SLC2A1 (GLUT1) was also found up-regulated in iTRAQ and MRM quantification results ( Fig. 2A and Supplemental Table S4). Because down-regulation of GLUT1 inhibits cellular glucose uptake, growth, and invasiveness of MG63 cells (52), its up-regulation under hypoxia would enhance cellular glucose uptake, growth, and invasiveness. Based on these data, we hypothesized that cells increased proteins that will enhance glycolysis and promote metastasis during hypoxia.
2-Deoxy-D-glucose (2-DG) is one of the best characterized glycolysis inhibitors in animal model studies and human clinical trials (53,54). In scratch-wound assay, 2-DG at a higher concentration was used to treat A431 cells and suppress the glycolysis pathway. The cell migration was found to be significantly inhibited when compared with that of the control cells under hypoxia and this reduced migration appeared to show positive correlation to the concentration of 2-DG (Fig.  2C). After exposure of A431 cells to 10 mM or 15 mM 2-DG for 20 h, cell migration into the wounded region of A431 cells was reported to be 65.9 Ϯ 9.4% and 36.3 Ϯ 5.9% as compared with control cells, respectively (Fig. 2D). The inhibition of 2-DG to cell glycolysis pathway was confirmed by metabolite analysis using LC-ESI-Q-TOF. The TIC of control metabolite sample had a slight higher peak than that of 2-DG treated cells, which indicated the loading amount of control and 2-DG treated samples was similar and the normalization with pro-tein content was relatively accurate (Supplemental Fig. S4A). However, the EIC peaks of one intermediate of glycolysis pathway, Fructose 1,6-biphosphate (FBP, m/z:339.12), extracted from TICs had obvious difference. The area of EIC peak from control sample is about 12-fold of that of 2-DG treated sample (Supplemental Fig. S4B). FBP level in normoxic and hypoxic A431 was also detected using the same method. This glycolysis metabolite in hypoxia sample is upregulated compared with normoxia sample even with a lower TIC in hypoxia sample during MS analysis (Supplemental Figs.  S4C and S4D). Supplemental Fig. S4E is referred to the MS/MS fragments of the precursor 339.12. The m/z and intensity of these MS/MS fragments were input to Massbank database for metabolite identification (http://www.massbank. ;1jp/MetaboIdentif.html). Only the Fructose 1,6-biphosphate of metabolite was matched. These results clearly showed that 2-DG inhibited, but hypoxia activated glycolysis pathway. Though 2-DG also resulted in the inhibition of glycan biosynthesis of glycoproteins, thus reducing the glycoproteins in cells and cell adhesion, treatment with glycan biosynthesis inhibitor (1-deoxynojirimycin hydrochloride, a glucose analog that inhibits ␣-glucosidase I and II activity) during the scratch assay for 20 h had no effects in cell migration (data not shown). These results indicate that the activation of glycolysis under hypoxic conditions is associated with enhanced A431 cell migration ability.
Hypoxia Induced Overexpression of Intergrins and Their Ligands: Implication for Increased Metastasis-Integrins, a major family of adhesion proteins, are involved in numerous physiological and pathological processes, including inflammation, wound repair, proliferation, differentiation, and apoptosis (55,56). In particular, they play a crucial role in metastasis. It is well established that fibronectin-integrin complexes played direct and indirect roles in signaling pathways that are involved in adhesion, migration, apoptosis, and growth (57). Expression of fibronectins and integrins is up-regulated in migratory cells in promoting of cell migration (58). Each integrin is composed of two subunits, namely ␣ and ␤ subunits. 18␣ and 8␤ integrin subunits assemble into 24 distinct receptors in mammals (59). Our iTRAQ data revealed the up-regulation of several integrin subunits under hypoxic conditions. The changes were partially reversed by reoxygenation (Fig.  3A). ITG␣5␤1 is one of the important integrin molecules. The increase expression of ITGA5 and ITGB1 subunits under hypoxia was validated by WB and LC-MS/MS-MRM methods ( Fig. 3B and Supplemental Table S4). Both results clearly showed that hypoxia induced the expression of the two subunits. Cell staining results not only showed the up-regulation of ITGA5 but also their recruitment to the plasma membrane under hypoxic conditions. ITGA5 appears to be evenly distributed in the normoxic A431 cells, whereas it has been observed to reside in the plasma membrane of both hypoxic and reoxygenated cells (Fig. 3C). LC-MS/MS-MRM experiment also confirmed up-regulation of other integrin molecules (ITGA3, ITGA6, and ITGB4) in A431 cells under hypoxia, indicating hypoxia may increase cell adhesion (Supplemental Table S4). We thus examined the adhesion ability of A431 cells on matrix-coated dishes after pre-exposure to hypoxia. The adhesion assay proved that hypoxia induces increased adhesiveness of A431 cells pre-exposed to hypoxia within 20 mins of seeding onto the fibronectin-coated dishes (Fig. 3D). The adhesion of the hypoxic and reoxygenated cells to fibronectin is 134.5 Ϯ 4.2% and 121.4 Ϯ 5.5% of that of normoxic cells, respectively (Fig. 3E). This increase in cell adhesion was consistent with the increase of several integrin molecules and ITGA5 recruitment to the plasma membrane. It has been reported that up-regulation of ITGA5 and fibronectin mediated by hypoxia, is responsible for the increased invasiveness of cancer cells (60). In hypoxia, the loss of E-cadherin leads to the up-regulation of ITGA5, resulting in the increased migration of extravillous trophoblast cells during early implantation (29). Integrins have a wide binding range to matrix proteins, which include laminin, fibronectin, trombospondin, vitronectin, and the various types of collagen (61). The cell adhesion ligands including several lamimin and collagen subunits were also found to be up-regulated under hypoxia (Fig. 3F). Hypoxia induced factor 1 (HIF-1␣) directly regulates the promoter activity of laminin ␣3 chain of laminin-332 (formerly laminin-5) and contributes to the migration of keratinocyte by increasing the expression of laminin-332 (62).
Hypoxia Induced Higher Turnover Rates of Glycoproteins-Glycoproteins play important roles in a wide range of biological events, including bacterial and viral infections, the metastasis of tumor cells, immune responses, and fertilization and differentiation (63). There are two major types of glycoproteins in mammalian cells, N-linked and O-linked glycoproteins. The synthesis and breakdown of N-linked glycoproteins have been studied intensively. The biosynthesis of the various types of N-linked oligosaccharide structures involves two series of reactions: 1) the formation of the lipid-linked saccharide precursor, Glc3Man9(GlcNAc)2-pyrophosphoryldolichol, by the stepwise addition of GlcNAc, mannose and glucose to dolichyl-P, and 2) the removal of glucose and mannose by membrane-bound glycosidases and the addition of GlcNAc, galactose, sialic acid, and fucose by Golgi localized glycosyltransferases to produce different complex oligosaccharide structures (64). Glycosidases are key enzymes involved in the biosynthesis of N-linked glycoproteins as they hold critical role in the formation of hybrid and complex types of N-linked oligosaccharides (65). Glycoproteins are degraded in lysosomes. Once inside the lysosome, glycoproteins are broken down by a combination of proteases and glycosidases. We found that, for the first time, the turnover of glycoproteins is regulated and promoted by increase of enzymes involved in biosynthesis and breakdown of glycoproteins under hypoxia (Fig. 4A and 4B).
The iTRAQ data showed that the enzymes (ALG1, ALG2 and DPM1), involved in the formation of lipid-linked saccharide precursor by transferring mannose in N-linked oligosaccharide formation step 1, were up-regulated under hypoxia. The glycosidase (MA1B1and MAN2A1) responsible for the removal of mannose, MOGS for glucose removal and the glycosyltransferases (Oligosaccharyltransferase complex, OST) in the step 2 of N-linked glycan synthesis were also FIG. 3. Hypoxia increases the expression level of integrin subunits and their recruitment to plasma membrane and even induces the expression of their partners. A, Change patterns of integrin subunits because of hypoxia or reoxygenation from iTRAQ data; B, The higher expression level of ITGA5 and ITGB1 under hypoxia was confirmed by Western blot. C, Cell immunostaining of ITGA5 clearly indicated that the adhesion molecule was prone to localize at the cell membrane under hypoxia. Magnification, ϫ630; D and E, Images and quantification of adhesion assay of A431 to FN under Nx, Hx, and Reox condition. Magnification, ϫ100; F, Change patterns of integrin ligands because of hypoxia or reoxygenation from iTRAQ data.
We speculated that the quantitative changes in glycosylation enzymes during hypoxia would affect glycoproteome and that this change would be associated with cell metastasis because of the important roles of glycoproteins in tumor cell metastasis. 1-Deoxynojirimycin (1-DJ), an analog of glucose that contains an NH group substituting for the oxygen atom of the pyranose ring, inhibits Golgi alpha-glucosidase and thus inhibits glycoprotein synthesis. This drug was used to treat A431 and perturb the glycoprotein-turnover under hypoxia. Short term drug treatment (20 h) exhibited no prominent effect on cell migration in scratch assay and this result also was supported by the reported data (66). However, once A431 cells were treated with 100 M 1-DJ for 48 h, the cells displayed slower migration than nontreated cells under hypoxia (Fig. 4C). The migration into wound of A431 cells treated by 100 M 1-DJ is only 72.3% Ϯ 12.6% of that of control cells (Fig. 4E). The lectin binding result showed that the synthesis of glycoproteins in A431 was inhibited after 1-DJ and hypoxia treatment (Fig. 4D). These data indicate that hypoxia contributes to A431 enhanced migration by increasing glycoprotein synthesis and modulating glycoprotein turnover.
Hypoxia Promoted A431 Migration by Suppression of STAT1 Pathway-Thus far, studies determining the influence of hypoxia on protein expressions have been restricted to proteins with enhanced expressions. Only a few studies have been done on proteins whose expression is down-regulated by hypoxia (35). STAT1, the first described member of the STAT transcription factor family, is the master transcription factor for IFN-related intracellular signaling (67). STAT1 has two isoforms, STAT1␣ and STAT1␤. The ␤ isoform lacks the 38 residues encompassing the transcription activation domain at C-terminal and is transcriptionally inactive (68). STAT1␣ is phosphorylated by JAK1/2 kinases at the Tyr701 position and then translocates to the nucleus where it binds to GAS (IFNc-activated sequence) promoter elements, thereby activating several hundred genes (69). Phosphorylation of STAT1␣ at S727 is essential for activating transcription of target genes and interaction with proteins, such as BRCA1 and MCM-5, potentiating STAT1-mediated transcription (70 -73).
STAT1 and its downstream proteins (MX1, IFIT1, IFIT3, ISG15, and PSMB8) were significantly down-regulated under hypoxia as reflected in our proteomic data (Fig. 5A), the suppression of STAT1, PSMB8, and IFIT1 was confirmed by MRM quantification (Supplemental Table S4). The significant decline of total STAT1 was further confirmed by WB experiment. The active isoform of STAT1 was also shown to be highly suppressed by hypoxia because the anti-pSTAT1(S727) only recognizes STAT1␣ isoform with transcriptional activity (Fig.5B). The mRNA level changes of STAT1␣ detected by its specific primers and its downstream genes were validated by RT-PCR. Fig. 5C clearly indicated their suppressed transcription as a result of STAT1␣ downregulation after hypoxia. STAT1 dramatically increases the transcription of IRF-1 through the interaction with NF-B or in response to IFN (74,75). IRF-1 transcription was suppressed in response to decreased STAT1 in RT-PCR; however, as a specific inhibitor of STAT1, PIAS1 showed a reverse finding with STAT1 (76). The alteration of STAT1 expression level because of hypoxia at present remains controversial. Several studies have reported that hypoxia led to activation of STAT1, especially in cardiomyocytes (77)(78)(79)(80). Paradoxically, Ivanov et al. reported that STAT1 and its downstream genes could be repressed by a pVHL/HIF-1 target DEC1/STRA13 in carcinoma cells under hypoxia (81).
To get more evidence of STAT1 functions under hypoxia, two A431 stable cell lines with STAT1␣ overexpression were screened for further studies. The human cDNA of STAT1␣ was inserted into p3xFLAG-CMV-10 expression vector and transfected into A431. G418 was used to screen the cell line stably overexpressing STAT1␣. The positive clones were confirmed by WB using anti-FLAG and STAT1 antibodies. When compared with the control cell line transfected with empty p3xFLAG-CMV-10, FLAG-STAT1 was only detected in the STAT1␣ stable cell lines (A431-p3XFLAG-ST1a-1 and 2). The stable cell lines also showed higher expression level of STAT1 and phosphorylated STAT1␣, in which A431-p3XFLAG-ST1a-2 has a higher expression of STAT1␣ than A431-p3XFLAG-ST1a-1 (Fig. 5D). RT-PCR results showed up-regulation of STAT1 downstream genes (Fig. 5E). In summary, the results highly suggested that the stable cell lines with more active STAT1␣ expressed could be used for functional studies.
MTT was used to evaluate the proliferation/survival of the STAT1␣ stable cell lines under nomorxia and hypoxia. Under nomorxia, there was no significant difference in the proliferation between control and STAT1␣ overexpression cell lines. But under hypoxia, STAT1␣ cell survival suffered a greater decline as compared with control cells and A431-p3XFLAG-ST1a-2 with the highest STAT1␣ level had a maximum decline (Fig. 6A). These results indicated that A431 cells increased their survivability through the down-regulation of STAT1 under hypoxia. STAT1 has been implicated in the inhibition of cell proliferation in several cell systems by inducing the expression of cyclin kinase inhibitors and caspases in response to extracellular stimuli (82)(83)(84)(85)(86). It is also reported that STAT1 can repress HIF-1-mediated transcription after hypoxia (87).
Because hypoxia induces cancer cell migration and causes functional suppression of STAT1, we went on further to examine the effects of STAT1 overexpression on the migration ability of cancer cells under hypoxia. In the scratch-wound assay, statistical results showed that STAT1␣ overexpression resulted in the repression of A431 cell migration. The migration of A431-p3XFLAG-ST1a-1 and 2 cells is only 82.9 Ϯ 4.3% and 62.0 Ϯ 5.4% of that of A431-p3XFLAG control cells (Figs. 6B and 6C). The migration of A431-p3XFLAG-ST1a-2 is significantly lower than that of A431-p3XFLAG-ST1a-1 be- cause of its higher expression of STAT1␣, which shows negative regulation of STAT1 in tumor metastasis. STAT1 has been reported to suppress the metastasis of STAT1-reconstituted RAD-105 cells through the reduction of MMP-2 and MMP-9 in vivo and arrest monocyte migration by modulating RAC/CDC42 pathways (88,89). Our studies further revealed that STAT1␣ might suppress A431 cell migration by promoting cell adhesion. In adhesion assay, the adhesion ability of A431-p3XFLAG-ST1a-1 and 2 to fibronectin was much higher than that in control cell. After 30 mins incubation, the adhesion cell amount of A431-p3XFLAG-ST1a-1 and 2 is about 2.03 Ϯ 0.03 and 2.28 Ϯ 0.12 folds of that of A431-p3XFLAG (Figs. 6D and 6E). Cell adhesion receptors play important roles in promoting migration; however, the establishment of strong adhesion inhibits migration, which is fastest at optimum adhesion strength strong enough to support traction yet weak enough to allow rapid detachment of the rear of the cell (58). The above results provide evidences that hypoxia induces cell migration by down-regulation of STAT1 because overexpression STAT1 can inhibit migration by enhancing cell adhesion. CONCLUSION In this study, based on iTRAQ quantification data, the pathways associated with tumor radioresistance and migration induced by hypoxia were identified and validated. For the first time, most key molecules in NHEJ pathway, which occurs more often than homologous repair in radiated cells, were found up-regulated under hypoxia and confirmed by LC-MS/ MS-MRM or WB. Thus, the proteins in this pathway may provide new targets for molecular therapy in support to ra-diotherapy (90). In addition, we reported hypoxia perturbed the expression of enzymes in glycoprotein synthesis and degradation to modulate glycoprotein turnover. Inhibiting glycoprotein synthesis by small molecule inhibitors under hypoxia led to decrease in cell migration. Glycosylation undergoes a dramatic transformation during cancer malignancy. But only a few structural changes have been frequently correlated with tumor progression by altering specific enzymes for further modification of glycans in Golgi (91). It is an intriguing question to find whether improvement of glycoprotein turnover contributes to the glycosylation transformation during tumor malignancy. STAT1 and its downstream genes were significantly suppressed by hypoxia. The repression is believed to promote cell migration because overexpression of STAT1␣ can inhibit migration by enhancing cell adhesion. STAT1 has been reported to be inhibited by HIF-1 (81). How about the NHEJ and glycoprotein turnover pathway proteins? It is our future work to knockdown HIF-1 and to investigate its impact on the two pathways under hypoxia. Our findings add in new dimension to our present understanding to tumor hypoxia and provide potential new therapy targets for tumor therapy.