Article pubs.acs.org/jpr
Integrative Transcriptome and Proteome Study to Identify the Signaling Network Regulated by POPX2 Phosphatase Songjing Zhang,† Tiannan Guo,†,§ Hei Chan,† Siu Kwan Sze,† and Cheng-Gee Koh*,†,‡ †
School of Biological Sciences, Nanyang Technological University, Singapore 6387551 Mechanobiology Institute, Singapore 117411
‡
S Supporting Information *
ABSTRACT: POPX2 is a serine/threonine phosphatase belonging to the protein phosphatase 2C (PP2C) family that has been found to be elevated in invasive breast cancer cells. Silencing of POPX2 results in lower cell motility and invasiveness. The molecular mechanism of POPX2-regulated cell motility is not well understood. To identify the relevant signaling pathways, we investigated the transcriptome and proteome of POPX2-knockdown MDAMB-231 breast cancer cells. Our data suggest that POPX2 might be involved in the regulation of focal adhesions and cytoskeleton dynamics through the regulation of MAP kinase (MAPK1/3) and glycogen synthase kinase 3 (GSK3α/β) activities. Silencing POPX2 alters phosphorylation levels of MAPK1/3 and GSK3α/β and results in reduced activity of these kinases. Both MAPK and GSK3 are known to regulate the activities of transcription factors. MAPK1/3 are also implicated in the phosphorylation of stathmin. The level of phospho-stathmin was found to be lower in POPX2 knockdown cells. As phosphorylation of stathmin inhibits its microtubule severing activity, we observed less stable microtubules in POPX2 knockdown cells. Taken together, our data suggest that POPX2 might regulate cell motility through its regulation of the MAPK1/3, leading to changes in the cytoskeleton and cell motility. KEYWORDS: POPX2 phosphatase, proteome, SILAC, transcriptome, cancer cell motility, MAPK1/3
■
INTRODUCTION Cancer metastasis is a complex process. A cascade of events is involved, among which cancer cell migration is indispensable.1−3 Remodeling and coordination of the actin cytoskeleton, focal adhesion, and microtubule dynamics are required during the cell migration.4 We have earlier found that POPX2, a serine/threonine phosphatase of the Protein Phosphatase 2C (PP2C) family, is implicated in the maintenance of actin stress fibers in HeLa and breast cancer cells.5−7 The levels of POPX2 are also higher in invasive MDA-MB-231 breast cancer cells compared with noninvasive MCF7.7 Knocking down POPX2 in breast cancer cell lines results in lower motility and invasiveness, suggesting positive correlation between POPX2 levels and cell motility.7 In addition, POPX2 overexpression in NIH3T3 cells also results in enhanced motility in scratch wound assays.8 How POPX2 affects the actin cytoskeleton and cell motility is not well understood. So far, two known substrates of POPX2, PAK (p21-activated kinase) and CaMKII (calcium calmodulin kinase II), have been reported to mediate the downstream effects of POPX2.5,9 POPX2 dephosphorylates PAK’s kinase loop at Thr423 and other autophosphorylation sites of PAK to downregulate its kinase activity. It is conceivable that POPX2 can act via PAK to affect the actin cytoskeleton. However, we have observed other phenotypes associated with POPX2 that cannot be attributed to PAK or CaMKII. For instance, POPX2 © 2013 American Chemical Society
can affect transcription mediated by the serum response factor (SRF) through interference with mDia activities6 and therefore might be expected to regulate expression of a larger repertoire of genes. Moreover, the expression and activities of PAK have been reported to be elevated in breast and colon cancer cells.10−12 Therefore, it is of interest to study POPX2, the regulator of PAK, in cancer cell signaling. In order to decipher the roles of POPX2 in cancer cell migration, we embarked on an integrative genomic and proteomic study on POPX2knockdown MDA-MB-231 breast cancer cells, followed by indepth functional investigations.
■
MATERIALS AND METHODS
SILAC (Stable Isotope Labeling by Amino Acids in Cell Culture) and Cell Culture
A custom RPMI1640 medium lacking arginine, lysine, and methionine was purchased from Biowest (Maimi, FL). 13C6Arginine (heavy) and 13C6-lysine (heavy) were purchased from Cambridge Isotopes Laboratories, Inc. (Andover, MA). 12C6Arginine (light), 12C6-lysine (light), and 12C6‑methionine (light) were purchased from Sigma-Aldrich. Dialyzed fetal bovine serum was purchased from Invitrogen. The control and Received: November 28, 2012 Published: April 26, 2013 2525
dx.doi.org/10.1021/pr301113c | J. Proteome Res. 2013, 12, 2525−2536
Journal of Proteome Research
Article
trap (Agilent Technologies, Santa Clara, CA) and subsequently resolved in a capillary column (200 μm i.d. × 10 cm) packed with C18 AQ (5 μm particle size, 100 Å pore size, Michrom BioResources, Auburn, CA). The samples were ionized in an ADVANCE CaptiveSpray Source (Michrom BioResources) with an electrospray potential of 1.5 kV. The LTQ-FT Ultra was set to perform data acquisition in the positive-ion mode. The 10 most intense ions above a 500 count threshold were selected for fragmentation. A full MS scan (350−1600 m/z range) was acquired in the FT-ICR cell at a resolution of 100000 and a maximum ion accumulation time of 1000 ms. The automatic gain control target for FT was set at 1 × 106, and precursor ion charge state screening was activated. The linear ion trap was used to collect peptides and to measure peptide fragments generated by CID. The default automatic gain control setting was used (full MS target at 3.0 × 104, MSn at 1 × 104) in the linear ion trap. The 10 most intense ions above a 500-count threshold were selected for fragmentation in CID (MS2), which was performed concurrently with a maximum ion accumulation time of 200 ms and a dynamic range of 30 s. For CID, the activation Q was set at 0.25, isolation width (m/z) was 2.0, activation time was 30 ms, and normalized collision energy was 35%.
POPX2 knockdown (X2-shRNA1/2) MDA-MB-231 stable cells were described previously.7 The X2-shRNA1 cells were cultured in RPMI1640 medium containing heavy arginine, heavy lysine, and light methionine, along with 10% (v/v) dialyzed fetal bovine serum, 100 U of penicillin, and 100 μg of streptomycin per milliliter (Invitrogen, CA). The control cells were maintained in RPMI1640 medium containing light arginine, light lysine, and light methionine, along with 10% (v/v) dialyzed fetal bovine serum, 100 U of penicillin, and 100 μg of streptomycin per milliliter (Invitrogen, CA). All cells were grown in 37 °C humidified incubator supplemented with 5% CO2. The cells used in the experiments were grown up to seven passages to ensure full incorporation of heavy and light amino acids. Protein Extraction, Digestion, And Fractionation
Cells harvested from heavy and light media were washed with 1 × PBS five times and lysed in protein lysis buffer containing 8 M urea (Sigma-Aldrich), 20 mM HEPES (Merck), complete protease inhibitor cocktail (Roche), and phosphostop phosphatase inhibitor cocktail (Roche). Protein concentration was determined by Bicinchoninic Acid (BCA) protein assay. Equal amounts of protein (15 mg each) from heavy and light samples were mixed at a 1:1 ratio and stored at −80 °C for the following experiments. Proteins were reduced with 10 mM dithiothreitol at 33 °C for 1 h, followed by alkylation with 55 mM iodoacetamide at room temperature for 45 min. After diluting 8 times with 20 mM HEPES (Merck), mixed extracts were digested in-solution by trypsin (Sigma-Aldrich) at 1:100 (m/ m). Peptide samples were adjusted to pH 2 using 10% formic acid and desalted using SEP-PAK C18 cartridges (Waters, Milford, MA) and vacuum-dried. The resulting peptides were fractionated using ERLIC (electrostatic repulsion−hydrophilic interaction chromatography).13 Briefly, the peptides were dissolved in buffer A (72% acetonitrile + 0.5% formic acid) and injected into a PolyLC PolyWAX LP column (4.6 × 200 mm, 5 μm particle size, 300 Å pore size) mounted on a Shimadzu Prominence UFLC unit (Shimadzu Corporation, Kyoto, Japan). The 40 min gradient was generated using a combination of buffer A (72% acetonitrile + 0.5% formic acid) and buffer B (10% acetonitrile + 1.0% formic acid). The gradient was composed of 5 min 100% A, followed by 30 min gradient from 0% to 100% B, and ended with 5 min 100% B. The first five collections were combined into one tube. The subsequent collections, containing phosphopeptides, were combined into 15 fractions for phosphopeptide analysis. The combined first five fractions were further fractionated into 20 fractions for proteome analysis. For the proteome fractionation, a different 40 min gradient was generated using a combination of buffer A (85% acetonitrile + 10 mM ammonium acetate + 1% formic acid) and buffer B (30% acetonitrile + 0.1% formic acid). The gradient was composed of 5 min 100% A, followed by 30 min gradient from 0% to 100% B, and ended with 5 min 100% B. All 35 fractions were desalted using SEP-PAK C18 cartridges (Waters, Milford, MA), vacuum-dried, and stored at −80 °C prior to LC−MS/MS analysis.
Proteomic Data Analysis
The MS raw files were converted to mzXML format and mgf format using Trans-Proteome Pipeline. Protein database search was performed by uploading mgf files to an in-house Mascot cluster server (version 2.2.07) (Matrix Science, Boston, MA) against a concatenated target and decoy version of manually annotated nonredundant UniProt Knowledgebase protein sequence database (40516 sequences, downloaded on 8 October 2010). The decoy sequences were generated by reversing the UniProt protein sequences. The search was limited to a maximum of two missed trypsin cleavages; #13C of 2; mass tolerance of 20 ppm for peptide precursors; and 0.8 Da mass tolerance for fragment ions. Fixed modification was carbamidomethyl at Cys residues, while variable modifications include oxidation at methionine residues, phosphorylation at serine, threonine and tyrosine, heavy lysine, and arginine. False discovery rates, calculated as 2 times the percentage of decoy matches in total matches, at both peptide and protein levels were below 1%. An extracted ion chromatograph (XIC) of the identified peptides was generated using in-house scripts. A tolerance of 0.05 Th was used for XIC generation. The ratio for SILAC pairs was calculated based on the area of XIC of each peptide. Ratios calculated from both replicate experiments were average in log scale, and error factors, where applicable, were calculated in log scale as well. RNA Extraction and DNA Microarray Data Analysis
The microarray experiment was performed by Genotypic (Bangalore, India). Briefly, total RNA was extracted from two sets of MDA-MB-231 (control) and MDA-MB-231-POPX2 knockdown cells using Qiagen RNeasy kit. The RNA was labeled with Cy3 or Cy5 using an Agilent Low Input RNA Amplification Kit. Agilent human whole genome 4 × 44k array slide was used for the hybridization. Array normalization was done using GeneSpring GX (Agilent) with the recommended Per Spot and Per Chip intensity dependent (lowness) normalization. Fold change analysis and gene ontology analysis were done using GeneSpring GX10 and Biointerpreter (Genotypic Technology) respectively.
LC−MS/MS Analysis
Digested peptides were analyzed in an LTQ-FT Ultra mass spectrometer (Thermo Fisher, Waltham, MA) coupled to a Prominence HPLC unit (Shimadzu, Kyoto, Japan) as described previously.14 Briefly, peptide samples were injected from an autosampler (Shimadzu) and concentrated in a Zorbax peptide 2526
dx.doi.org/10.1021/pr301113c | J. Proteome Res. 2013, 12, 2525−2536
Journal of Proteome Research
Article
Figure 1. Characterization of stable POPX2 knockdown MDA-MB-231 cell lines and our experimental workflow. (A) Western blot analysis using anti-POPX2 antibody was performed for protein lysates collected from Control, POPX2-KD1 and POPX2-KD2 cell lines. Actin was used as a loading control. (B) Workflow of combined transcriptome and proteome analysis.
1% Triton/PBS at 4 °C overnight and secondary antibodies diluted in 1% Triton/PBS at room temperature for 1 h. Images were captured using an Axio Observer microscope (Carl Zeiss, Germany) and analyzed by Image J software.
Western Blot and Antibodies
A total of 50 μg of protein per well was loaded onto 12% SDSPAGE. After gel electrophoresis, the protein was transferred to a nitrocellulose membrane. The membrane was blocked in 5% skimmed milk for 1 h at room temperature. Then, the blot was incubated in primary antibody 1:1000 diluted in 5% skimmed milk at 4 °C overnight (phospho-antibodies were diluted in 2% BSA), followed by secondary antibody 1:5000 diluted in 5% skimmed milk for 1 h at room temperature. POPX2 (rabbit, polyclonal) antibody was raised by us and targets N-terminus of POPX2. Antibodies targeting MAPK1/3, phospho-MAPK1/3 (T185/Y187/T202/Y204), GSK3α, phospho-GSK3α (S21), GSK3β, phospho-GSK3β (S9), stathmin, and phosphostathmin (S38) antibodies were from Cell Signaling Technology. α-Tubulin, α-acetylated-tubulin, and Flag antibodies were from Sigma-Aldrich. Actin antibody was from Millipore.
Scratch Wound Assay
Similar numbers of cells (6.5 × 105) were seeded in six-well dishes. When cells were grown to a monolayer (24 h), five scratch wounds were generated per well using white pipet tips. The cells were washed with 1 × PBS, incubated in fresh medium and allowed to migrate into the wound gap for 24 h. Phase contrast images were acquired at time-0 and time-24 and analyzed using Image J software. Three independent experiments were performed. Bioinformatics
Pathway Analysis. The differentially expressed genes and proteins were submitted to DAVID (Database for Annotation, Visualization and Integrated Discovery)15 separately using their Agilent ID (transcriptome) and official_gene_symbol (proteome). The functional annotations of each gene or protein were downloaded and their KEGG pathway annotations were manually distributed. The fisher exact P value was adopted to examine the pathway enrichment. Genes or proteins involved in
Immunofluorescence
Cells cultured on glass coverslips were washed with 1 × PBS once and fixed in 4% paraformaldehyde for 20 min at room temperature. After fixing, the cells were permeabilized with 0.2% Triton/PBS for 10 min and blocked in 4% BSA for at least 30 min. Cells were incubated in primary antibodies diluted in 2527
dx.doi.org/10.1021/pr301113c | J. Proteome Res. 2013, 12, 2525−2536
Journal of Proteome Research
Article
the same KEGG pathway were counted. For each pathway, the fisher extract P-value was calculated based on the following numbers: (1) Hit number: the number of genes (or proteins) in our query list involved in this pathway. (2) Hit total number: the number of genes (or proteins) in our query list mapped to KEGG pathway database. (3) Background hit number: in human genes background, the number of genes annotated to this pathway. (4) Background hit total number: in human genes background, the number of genes annotated to KEGG pathway database. Motif Analysis. Motif-X16 was used for the kinase motif analysis. Parameters were set up as follows: the significance was set as 0.000001 with an occurrence of 5. The IPI human proteome database was used as background. The sequences of all phospho-peptides were submitted and centered on phosphoserine with at least ±6 amino acids extended from the phosphorylation sites. No phospho-threonine or phosphotyrosine motifs were extracted, which might be due to the low numbers of phospho-threonine or phospho-tyrosine peptides in our data set. STRING Analysis. Protein network was analyzed using the STRING database.17 The differentially expressed phosphoproteins were submitted to the STRING database and a medium confidence setting (0.400) was used. Post-translational modification (pink) and binding (blue) functional link between two proteins were used to visualize protein relationships.
■
Figure 2. Distribution of proteome and transcriptome expression ratio. Both the transcript and protein ratios were calculated as POPX2 knockdown vs control. The Log2 Ratio was calculated for every gene and protein. The frequencies (vertical axis) of the Log2 Ratio (horizontal axis) were plotted in histograms for transcriptome (A) and proteome (B).
RESULTS
Identification of genes Perturbed by POPX2 Knockdown by Transcriptome analysis
We first investigated transcriptome changes induced by stable POPX2-knockdown in a MDA-MB-231 breast cancer cell model that we have established previously.7 Briefly, two independent POPX2 knockdown cell lines were derived from MDA-MB-231 cells infected with two separate lentiviruses delivering short hairpin shRNA targeting different sequences of POPX2 (X2-shRNA1 and 2). Cells infected with scrambled shRNA lentivirus (Control) were also generated as controls. The lentiviral vector also expressed GFP (green fluorescence protein) as a marker.18 The knockdown efficiency was shown in Figure 1A. An Agilent DNA microarray system was employed for the analysis of the transcriptome (Figure 1B). Total RNA from the control and two independent POPX2 knockdown cells, X2-shRNA1, and X2-shRNA2, were used. Two independent data sets were obtained from the DNA microarray analysis. A total of 22604 genes at the RNA level were obtained from the two independent DNA microarray data sets (X2shRNA1 vs Control and X2-shRNA2 vs Control) (Supplementary Table 1_transcriptome, Supporting Information). The distribution of transcriptome expression ratio was plotted in the histogram shown in Figure 2A. Typically, genes with more than 2-fold change would be chosen as the significantly perturbed set.19 Accordingly, a total of 1163 genes were regarded as upregulated hits (Log2 Ratio >1), and 639 genes were regarded as down-regulated hits (Log2 Ratio < −1).
grown cells were mixed, and two technical replicates were carried out. A total of 7159 peptides corresponding to 1775 proteins from experiment 1 and 7790 peptides corresponding to 1894 proteins from experiment 2 were identified from the proteome analysis (Supplementary Tables 2−7_proteome, Supporting Information). The number of total unique quantified proteins was 2146. A 1.8-fold change cutoff was applied for the proteome analysis. Accordingly, a total of 110 proteins were regarded as up-regulated hits (Log2 Ratio >0.85), and 797 proteins were regarded as down-regulated hits (Log2 Ratio < −0.84) (Figure 2B). Combined KEGG Pathway Analysis Implicated POPX2-Related Signaling Pathways
To determine POPX2 related signaling pathways in breast cancer, we translated the differentially regulated genes and proteins into affected pathways. A combined pathway analysis was performed using the public KEGG (Kyoto Encyclopedia of Gene and Genomes) pathway database20 and the DAVID (Database for Annotation, Visualization and Integrated Discovery)15 for both differentially regulated transcriptome and proteome data sets (Table 1). We found signaling pathways related to the regulation of “focal adhesion” and “cell cycle” to be commonly and significantly enriched using the KEGG pathway analysis (P < 0.05). Among these overrepresented pathways, “focal adhesion” was most related to cell migration,21 a phenotype we associate with invasive cancer cell types. Although cell cycle regulatory pathways are over-
Proteomic Alterations in POPX2 Knockdown Cells
Changes in the proteome in response to POPX2 silencing were investigated by SILAC-based quantitative proteomics (Figure 1B). Proteins harvested separately from multiple independently 2528
dx.doi.org/10.1021/pr301113c | J. Proteome Res. 2013, 12, 2525−2536
Journal of Proteome Research
Article
several related signaling pathways, including those regulating the actin cytoskeleton, MAPK signaling, and the cell cycle. This is consistent with our KEGG pathway analysis done independently using the proteome and transcriptome data sets (Table 1). We then asked whether the affected signaling pathways are associated with phenotypic variations we observed for the control and POPX2 knockdown cells. Since we have previously reported that the levels of POPX2 were higher in the more invasive breast cancer cells,7 we proceeded to determine whether POPX2 was linked to the regulation of actin cytoskeleton and focal adhesion which might lead to the increased cell motility and invasiveness in MDA-MB-231 cells. A decrease in stress fibers was observed in both POPX2 knockdown cell lines compared with control cells (Figure 3A). The number of focal adhesions, especially those at the center of the cells, was also reduced in the POPX2 knockdown cells (Figure 3B). Similar phenotypes were observed in another
Table 1. KEGG pathway enrichment. Enriched pathways with P-value
