Phosphatase of Regenerating Liver 3 (PRL3) Provokes a Tyrosine Phosphoproteome to Drive Prometastatic Signal Transduction*

Phosphatase of regenerating liver 3 (PRL3) is suspected to be a causative factor toward cellular metastasis when in excess. To date, the molecular basis for PRL3 function remains an enigma, making efforts at distilling a concerted mechanism for PRL3-mediated metastatic dissemination very difficult. We previously discovered that PRL3 expressing cells exhibit a pronounced increase in protein tyrosine phosphorylation. Here we take an unbiased mass spectrometry-based approach toward identifying the phosphoproteins exhibiting enhanced levels of tyrosine phosphorylation with a goal to define the “PRL3-mediated signaling network.” Phosphoproteomic data support intracellular activation of an extensive signaling network normally governed by extracellular ligand-activated transmembrane growth factor, cytokine, and integrin receptors in the PRL3 cells. Additionally, data implicate the Src tyrosine kinase as the major intracellular kinase responsible for “hijacking” this network and provide strong evidence that aberrant Src activation is a major consequence of PRL3 overexpression. Importantly, the data support a PDGF(α/β)-, Eph (A2/B3/B4)-, and Integrin (β1/β5)-receptor array as being the predominant network coordinator in the PRL3 cells, corroborating a PRL3-induced mesenchymal-state. Within this network, we find that tyrosine phosphorylation is increased on a multitude of signaling effectors responsible for Rho-family GTPase, PI3K-Akt, STAT, and ERK activation, linking observations made by the field as a whole under Src as a primary signal transducer. Our phosphoproteomic data paint the most comprehensive picture to date of how PRL3 drives prometastatic molecular events through Src activation.

Protein-tyrosine phosphatases play critical regulatory roles during signal transduction and when deregulated cause aberrant tyrosine phosphorylation that lies at the heart of many human diseases, including cancer (1)(2)(3). The Phosphatase of Regenerating Liver (PRL) 1 phosphatases represent a unique sub-family of prenylated protein-tyrosine phosphatases comprised of three members (PRL1, 2, and 3) that share Ͼ75% of amino acid sequence identity (4 -6). A ground-breaking observation that PRL1 was an immediate early gene induced before the regeneration period of the rat liver following resection brought attention to the PRL-family as potential protooncogenes (7). Over more than two decades, research continues to provide evidence that the PRLs may play causative roles in tumorigenic and metastatic processes when aberrantly overexpressed (8).
PRL3 (Ptp4a3) was first cast into the spotlight as a potential causative factor of metastasis when its transcript was found to be consistently and massively overexpressed in colorectal cancer (CRC) metastases found in the liver, whereas its expression in non-metastatic primary tumors and normal colorectal epithelium was undetectable (9). Subsequently, PRL3 transcript was found to be elevated in all metastatic lesions derived from CRC, regardless of the site of metastasis (liver, lung, brain, or ovary) (10 -11). To date, PRL3 transcript as well as protein has been found to be overexpressed in a variety of advanced neoplasms or metastases originating from a multitude of physiologically distinct tissues (8,(12)(13), suggesting a fundamental role for this phosphatase in driving cellular behaviors that are necessary to gain selective advantage toward metastatic dissemination when in excess.
In an effort to elucidate a mechanism by which PRL3 potentiates cellular metastasis, investigators have used a candidate approach in a variety of ectopic PRL3 expressing cell culture models. Consistent with gross morphological alterations akin to an epithelial-to-mesenchymal transition (EMT) and bio-functional responses such as increased proliferation, migration or invasion, and survival, data from these studies provide molecular evidence supporting the modulation of key proto-oncogenes and tumor suppressors known to govern these processes including: Src tyrosine kinase, Crk-associated substrate (p130Cas), C-terminal Src kinase (Csk), extracellular signal-regulated kinase 1/2 (ERK1/2), and signal transducer and activator of transcription 3 (STAT3) (14 -15), integrin receptors (␣1, ␣5, ␤1) (16 -18), Rho-family GTPases (Cdc42, Rac1 and RhoA/C) (19 -20), matrix metalloproteinase 2 (MMP2) (17), nuclear factor kappa-light-chain-enhancer of activated B cells (NFB) (21), p53 (22)(23), E-cadherin, phosphatase and tensin homolog (PTEN), phosphoinositide 3-kinase (PI3K)-Akt, and Snail (24). Similar to the pleiotropic response elicited by various transmembrane growth factor, cytokine, and integrin receptors following extracellular ligand stimulation, these studies reveal that the consequences of PRL3 overexpression are broad and directed toward the intracellular activation of a multitude of cellular proto-oncogenes. How does the overexpression of a phosphatase "tilt" the natural regulation of tyrosine phosphorylation toward a stimulus that would support the sustained intracellular activation of a multitude of proto-oncogenes? Additionally, what transmembrane receptors, molecular adaptors, enzymes, and transcription factors are commissioned to integrate this pleiotropic response? Unfortunately, the molecular basis for PRL3 function remains an enigma, predominately because of a complete lack of knowledge regarding a putative substrate, making efforts at distilling a concerted mechanism that would potentiate metastatic dissemination very difficult.
We have previously documented that ectopic PRL3 expression activates the Src tyrosine kinase by down-regulating Csk, a major negative regulator of Src, in squamous-epithelial-HEK293 cells (14). Consistent with the aberrant activation of a tyrosine kinase such as Src, a pronounced increase in "global" tyrosine phosphorylation was also observed in the PRL3 cells. We hypothesized that identification of the complete repertoire of proteins experiencing enhanced levels of tyrosine phosphorylation in the PRL3 cells would enable us to better define the "PRL3-mediated signaling network." Here we take an unbiased mass spectrometry-based approach toward identifying these phosphoproteins. Phosphoproteomic data support the intracellular activation of an extensive signaling network normally governed by extracellular ligand-activated transmembrane growth factor and integrin receptors in the PRL3 cells. Interestingly, our data support a mitogenic and chemotactic transmembrane receptor array known to be used by migratory mesenchymal cells as being the predominant network coordinator in the PRL3 cells. Additionally, the data implicate the Src tyrosine kinase as the major intracellular kinase responsible for "hijacking" this network and provide strong evidence that aberrant Src activation is a major consequence of PRL3 overexpression. Within this network, we find that tyrosine phosphorylation is increased on a multitude of signaling effectors responsible for Rho-family GTPase, PI3K-Akt, STAT, and ERK activation, linking obser-vations made by the field as a whole under Src as a primary signal transducer. Our phosphoproteomic data paint the most comprehensive picture to date of how PRL3 drives pro-metastatic molecular events through Src activation.
Cell Culture and Stable Clone Selection-HEK293 cells were grown in DMEM supplemented with 10% FBS, penicillin (50 units/ml), and streptomycin (50 g/ml) under a humidified atmosphere containing 5% CO 2 . Human PRL3 was inserted into pCDNA3 (14) and v207 expression vectors (v207 as described in (25)). Transfection and stable clone selection for pCDNA3 as described in (14). HEK293 cells were seeded so that 40 -50% confluence would be achieved following an over-night incubation period. v207-PRL3 constructs were transfected into HEK293 cells maintained in antibiotic-free medium using Poly(ethylenimine) (PEI). Twenty-four hours after transfection, Puromycin (1 g/ml) was added to the culture medium to initiate stable clone selection. Stable clones were picked after 2 weeks of selection under Puromycin.
Immunoblotting and Immunoprecipitation-Cells were grown in DMEM supplemented with 10% FBS, penicillin (50 units/ml), and streptomycin (50 g/ml) under a humidified atmosphere containing 5% CO 2 to 70 -80% confluence, washed with ice-cold phosphatebuffered saline (PBS), and lysed on ice for 30 min in 500 L-1 ml of lysis buffer (100 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% Triton X-100, 5% Glycerol, PhosSTOP phosphatase inhibitor mixture tablet (Roche), and a Complete EDTA-free protease inhibitor mixture tablet (Roche)). Cell lysates were cleared by centrifugation at 15,000 rpm for 15 min. Lysate protein concentration was assayed using the BCA protein assay kit (Pierce, Waltham, MA) (measurements for the standard series and experimental groups were kept under 5% coefficient of variation-CV). For immunoprecipitation, 10 g antibody was added to 1 mg protein lysate and incubated at 4°C for 4h to over-night by end-over-end rotation. Twenty microliters of protein A/G-plus agarose beads was then added and incubated with sample for an additional 2 h at 4°C using end-over-end rotation. After extensive washing, protein complex was boiled with Laemmli (SDS)-sample buffer, separated by SDS-PAGE, transferred electrophoretically to a nitrocellulose membrane, and immunoblotted with appropriate antibodies followed by incubation with horseradish peroxidase-conjugated secondary antibodies. The blots were developed by the enhanced chemiluminescence technique using the SuperSignal West Pico Chemiluminescent substrate (Pierce). Data shown is a representation of multiple repeat experiments.
Imaging-Stable RFP-PRL3-WT-and RFP-vector-expressing HEK293 cells were generated and selected, using methodology documented in the Cell Culture and Stable Clone Selection section. Cells were grown to subconfluence and RFP was visualized by confocal microscopy from live cells. Confocal images were acquired on Zeiss Axio ObserverZ1 as structured light via an Apotome and processed with Axiovision 4.7.
Label-free Quantitative Mass Spectrometry-Stable PRL3-HEK293 cells and their vector control HEK293 cell counterparts were grown to 80% confluence in DMEM supplemented with 10% FBS, penicillin (50 units/ml), and streptomycin (50 g/ml) at 37°C under a humidified atmosphere containing 5% CO 2 . This study was comprised of two groups with seven biological replicates per group allowing for 14 randomized HPLC injections. A detailed account of the label-free LC/MS-based protein quantification method used in this study, including MS-data acquisition and analysis can be viewed in (26 -28). This "label-free" approach is a quantitative assessment of protein abundance obtained from mass spectral data derived following single-dimension reverse-phase HPLC. This methodology does not have the capacity to represent the natural protein abundances of phosphotyrosine-containing proteins derived by the double enrichment strategy highlighted in the Phosphopeptide Enrichment Using Phosphotyrosine-Immunoprecipitation and PolyMAC-Ti Reagents section.
SILAC-based Quantitative Mass Spectrometry-SILAC (Stable Isotope Labeling of Amino acids in Cell culture) DMEM without L-Lysine or L-Arginine was supplemented with 7.5% dialyzed FBS and 2.5% undialyzed FBS, penicillin (50 units/ml), streptomycin (50 g/ml), and either ( 12 C 6 -L-Lysine monohydrochloride; 12 C 6 -L-Arginine monohydrochloride) or ( 13 C 6 -L-Lysine monohydrochloride; 13 C 6 -L-Arginine monohydrochloride) to create SILAC-"Light" or SILAC-"Heavy" media, respectively. PRL3-HEK293 and their vector-control HEK293 counterparts were grown in both SILAC-Heavy and SILAC-Light media, respectively, for a total of five passages before testing the labeling efficiency of the SILAC-Heavy media ( 13 C 6 -Lys/Arg amino acids). Tryptic-peptides from the Heavy-labeled PRL3-HEK293 cell lysate were prepared as documented in the Phosphopeptide Enrichment Using Phosphotyrosine-Immunoprecipitation and PolyMAC-Ti Reagents section. Two (2) "quantitative" biological replicates were carried out using SILAC to complement the eight (8) and nine (9) "qualitative" biological replicates carried out from pTyr-peptide-enriched samples derived from the vector-and PRL3-cells, respectively. Data for 2725 tryptic-peptides were acquired following a single-dimension reverse phase HPLC separation. Of the 2725 total peptides, 2613 peptides were completely labeled with 13 C 6 -Lysine and/or 13 C 6 -Arginine, whereas just 112 peptides contained no labeled amino acids (96% labeling efficiency). This degree of labeling efficiency was deemed sufficient for quantitative mass spectrometry to be carried out. Quantitative data analysis was carried out by Proteome Discoverer V1.3. See the Mass Spectrometry (LTQ-Orbitrap) Analysis and Phosphopeptide Data Acquisition and Analysis sections for methodology following cell culture.
Phosphopeptide Enrichment Using Phosphotyrosine-Immunoprecipitation and PolyMAC-Ti Reagents-PRL3-HEK293 and vectorcontrol HEK293 cells were grown in either normal DMEM supplemented with 10% FBS for qualitative analysis or SILAC-Heavy/Light DMEM supplemented with 7.5% dialyzed FBS/2.5% undialyzed FBS for quantitative analysis. On reaching 80% confluence, cells were lysed in ice cold lysis buffer (100 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% Triton X-100, 5% glycerol, PhosSTOP phosphatase inhibitor mixture (Roche), and Complete EDTA-free protease inhibitor mixture (Roche)). Lysate protein concentration was quantified using the BCA assay (measurements for the standard series and experimental groups were kept under 5% coefficient of variation-CV). 2.5 mg lysate protein/experimental group were used for subsequent steps (at this point if SILAC-based quantitation was performed, the PRL3 and vector-control lysates were consolidated to a single master lysate) (for qualitative assessment lysate protein from both experimental groups was held at an equivalent concentration and total volume). Lysate protein was denatured using 0.1% RapiGest surfactant (Waters, Milford, MA) in 50 mM trimethylammonium hydrogen carbonate (bicarbonate) (TMAB). Protein disulfides were reduced using 10 mM DTT in a 30 min incubation at 50°C. Reduced sulfhydryls were then alkylated using 20 mM iodoacetamide (IAA) in a 1 h incubation in the dark at ambient temperature. 1 M TMAB was used to adjust the sample pH to 8.0 before the trypsin digestion reaction. Proteins were then subjected to an over-night (12 h) digestion at 37°C by the trypsin endoproteinase at a ratio of 1:100 (trypsin/lysate protein). RapiGest was removed by reducing the pH to Ͻ3.0 using 1 M HCl (final concentration 100 -120 mM in the sample) and incubating the sample at 37°C using a water bath for 40 min. The supernatant was transferred to a new nonstick/low-binding OmniSeal tube (Life Science Products, Inc.). 1 M Tris was used to adjust the sample pH to 7.5 and 100 l of the anti-pTyr-antibody PT66-agarose conjugate slurry was added to the sample. Phosphotyrosyl-tryptic peptides were immunoprecipitated using the "pan" anti-pTyr-PT66 antibody over-night (12 h) at 4°C by end-over-end rotation. The PT66-agarose beads were extensively washed using ice cold lysis buffer and H 2 O. The phosphopeptides were eluted from the PT66-agarose beads using a series of elution steps with 0.1% trifluoroacetic acid (TFA), 0.1% TFA/50% acetonitrile, and 100 mM glycine pH 2.5. The consolidated eluent was dried down using vacuum centrifugation. The dried down product was solvated in 150 mM HEPES, pH 6.8. A secondary-phosphopeptide enrichment was then performed using a PolyMAC-Ti reagent as per the protocol documented in (29). A complete documentation regarding the above methodology can be viewed in (29).
Mass Spectrometry (LTQ-Orbitrap) Analysis-Peptide samples were solvated in 8 l of 0.1% formic acid and injected into an Agilent nanoflow 1100 HPLC system. The reverse phase C 18 -based chromatography was performed using an in-house C 18 -capillary column packed with 5-m C 18 Magic bead resin (Michrom; 75 m innerdiameter and 12-cm bed length) on an 1100 Agilent HPLC system (30). The mobile phase buffer consisted of 0.1% formic acid in ultrapure water (Sigma, MS-grade) with the eluting buffer of 100% acetonitrile run over a shallow linear gradient over 60 min with a flow rate of 0.3 l/min. The electrospray ionization emitter tip was generated on the prepacked column with a laser puller (Model P-2000, Sutter Instrument, Novato, CA). The Agilent 1100 HPLC system was coupled on line with a high resolution hybrid linear ion trap orbitrap mass spectrometer (LTQ-Orbitrap XL, Thermo Fisher Scientific). The mass spectrometer was operated in the data-dependent mode in which a full MS scan (from m/z 300 to 1700 with a resolution of 30,000 at m/z 400) was followed by four MS/MS scans of the most abundant ions meeting a 1000 signal threshold count mark. Ions with a charge state of 1ϩ were excluded. The mass exclusion time was 180s.
Phosphopeptide Data Acquisition and Analysis-LTQ-Orbitrap raw files were searched directly against a Homo-sapiens database with no redundant entries (67,250 entries; human International Protein Index (IPI) version 3.83) using the SEQUEST and Mascot algorithms as part of Proteome Discoverer software V1.3 (Thermo-Fisher Scientific, San Jose, CA). Search parameters were user specified in the following manner. Peptide mass tolerance was set at 10 ppm, and MS/MS tolerance was set at 0.8 Da. Search criteria included a static modification of cysteine residues of ϩ57.0214 Da, variable modifications of ϩ15.9949 Da to include potential oxidation of methionine residues, and a modification of ϩ79.996 Da on tyrosines for identification of phosphorylation. Searches were performed with full tryptic digestion and allowed a maximum of two missed cleavages on the peptides analyzed from the sequence database. The parameters for false-discovery-rate (FDR) were set for 1% for each analysis. Proteome Discoverer software generated a reverse "decoy" database from the chosen database, and any peptides passing the initial filtering parameters that were derived from this decoy database are defined as false positive identifications. The minimum cross-correlation factor (Xcorr) filter was then re-adjusted for each individual charge state separately to optimally meet the predetermined target FDR of 1% based on the number of random false-positive matches from the reversed decoy database. Thus, each dataset had its own passing parameters. The Percolator semi-supervised machine learning algorithm, as part of the Proteome Discoverer V1.3 software package, was used to assist in the generation of the 1% FDR threshold (results for each PSM were given as q-values in the data analysis). The most likely phosphorylation site localization from CID mass spectra was determined by PhosphoRS algorithm within the Proteome Discoverer 1.3 software. The number of unique phosphopeptides and nonphosphopeptides identified were then counted and compared. Unless otherwise specified the phosphoprotein documented is the primary isoform (1 or A) as the phosphopeptide(s) identified are redundant among the isoforms. SILAC quantitation was carried out using Proteome Discoverer software V1.3, which uses the MS peak areas of the light and heavy peptides and reports light/heavy (L/H) ratios. The significance threshold for quantitation parameters were determined by Proteome Discoverer V1.3 internal statistical algorithm, resulting in at least twofold passing significance ratio for L/H ratio after the removal of outliers. Search criteria included variable modifications of ( 13 C 6 )-Lys and -Arg residues of ϩ6.02Da. Quantification method was (SILAC 2plex (Arg6, Ly6) (Custom). RT tolerance of isotope pattern multiplets was set to 0.2 min. Biological reproducibility of identified phosphopeptides was assessed through eight (8) vector and nine (9) PRL3 qualitative sample preparations/MS runs and two (2) consolidated (vector/PRL3) quantitative sample preparations/MS runs. All data associated with the results reported in this study including: MS raw files, result output files, and annotated MS/MS spectra can be found at the following data repository link (http://www.peptideatlas. org/PASS/PASS00206). Follow the 'Description and Usage Information for Archived Files' document as an instruction and usage reference to the files archived in this repository.
Ingenuity Pathway Analysis (IPA)-All proteins from the PRL3 and vector-control datasets that possess a tyrosine-phosphorylated peptide(s) and their corresponding SILAC-based quantification values (1% FDR data following Sequest and Mascot searches of the IPI human v3.83 database) were uploaded to Ingenuity Pathway Analysis (IPA) software (Ingenuity Systems, Inc.). A new core analysis was created that included: Ingenuity knowledge base (genes only) reference set, direct and indirect relationships to target proteins, and a filter summary which included: (species: human, mouse, rat; confidence: experimentally observed; data sources: Ingenuity expert findings). The top scoring network (Network 1) and top 10 predicted canonical pathways and bio-functions (using a B-H (Benjamini-Hochberg) p value adjustment) were used to represent the current data set.

PRL3 Expression Induces Enhanced Global Tyrosine
Phosphorylation-To begin to investigate the tyrosine phosphoproteome following stable ectopic PRL3 expression, we employed squamous-epithelial human embryonic kidney 293 (HEK293) cells. In addition to the PRL3-WT expressing clone (WT1) used in our previous study (14), we also generated a second stable PRL3-WT expressing HEK293 cell clone (WT2), which displays similar morphological and molecular characteristics. We documented the degree of PRL3 expression in WT1 by quantitative real-time PCR (qRT-PCR) in our previous study (14). Here, we corroborate this data by showing enhanced PRL3 expression in WT1, WT2 and the catalytically inactive mutant PRL3-C104S cells, relative to endogenous levels of PRL3 observed in vector counterparts by RT-PCR using the same reagents/methodology (Fig. 1A). Furthermore, HEK293 cells harboring a stably expressed RFP (Red Fluorescent Protein)-tagged PRL3-WT fusion protein were generated to validate the expression and proper localization of PRL3 on endomembranes (Fig. 1A). PRL3-WT cells have a "spindle-like" fibroblast morphology as compared with the "squamous-like" epithelial morphology of their PRL3-C104S and vector counterparts, consistent with an epithelial-tomesenchymal transition (EMT) occurring on PRL3-WT expression in these cells (Fig. 1B). PRL3-WT cells have enhanced global tyrosine phosphorylation, a markedly less latent pTyr527-Src population, and constitutive phosphorylated/activated ERK1/2 and STAT3, relative to vector counterparts (Fig. 1C). These results are in agreement with our previous findings (14) and establish a framework that can be built on to more extensively define the tyrosine phosphoproteome of PRL3-WT (PRL3) cells.
PRL3-WT (WT1) and its vector (Vec1) HEK293 counterpart were used to provide material for our phosphoproteomic investigation. The sample handling and analysis flow-chart that was used to acquire both qualitative and SILAC (stable-isotope labeling of amino acids in cell culture)-based quantitative phosphoproteomic data from these cell lines can be seen in (Fig. 2). Qualitative and SILAC-based quantitative data are additive in this work and used to represent the entire PRL3 and vector phosphoproteomic datasets. Briefly, normalized total protein from the two experimental groups was either processed separately (qualitative assessment) or consolidated to one master sample (quantitative assessment). Tyrosine phosphorylated tryptic peptides were purified using a tandem enrichment strategy based on a primary pan-pTyrantibody immunoprecipitation followed by a secondary enrichment using a novel soluble nanopolymer multifunctionalized with Titanium (Ti) (29) and subsequently sequenced by high resolution/high mass accuracy tandem mass spectrometry. Quantitative data representing tryptic peptides from the "house-keeping" proteins ␤-Actin and ␤-Tubulin, validate a ϳ1:1 mixing ratio between the SILAC-Light ( 12 C 6 -Lys/ 12 C 6 -Arg) and SILAC-Heavy ( 13 C 6 -Lys/ 13 C 6 -Arg) protein lysates before sample processing (supplemental Fig. S1).
Using this approach, we were able to validate our immunoblot findings that proteins from the PRL3 cells experience enhanced levels of tyrosine phosphorylation by qualitatively identifying 172 phosphotyrosyl-residues on 123 tyrosine phosphorylated proteins from the PRL3 cells and 78 phosphotyrosyl-residues on 61 tyrosine phosphorylated proteins from vector cells with an overlap of 32 phosphotyrosinecontaining proteins between the two data sets. Using SILAC, we were able to quantify the relative abundance of 121 of 169 (71.6%) and 66 of 77 (85.7%) phosphotyrosine-peptides observed from the qualitative analysis of phosphopeptides acquired from the PRL3 and vector cells, respectively. The entire curated phosphoproteomic dataset organized into bio-functional categories can be seen in (supplemental Table S1). Out of the 250 total phosphotyrosyl-residues identified in this study, 226 (90%) have been previously identified by studies aimed at identifying tyrosine phosphorylation events downstream of tyrosine kinase activation, a strong testament to the quality and credibility of the data presented in this work. Raw fragmentation spectra depicting phosphopeptides representing pTyr187 of ERK2, pTyr705 of STAT3, and pTyr417 of PAG1 were chosen based on relevance to data presented in our previous study to represent the general quality of spectra used for SILAC-based quantitative assessment of tyrosine phosphorylation in this study (Fig. 3). Additionally, raw fragmentation spectra depicting phosphopeptides representing pTyr187 of ERK2 and pTyr783 of PLC␥1 were chosen to show the general quality of spectra used for qualitative assessment FIG. 1. Ectopic PRL3 expression induces aberrant regulation of tyrosine phosphorylation. A, PRL3 transcript is significantly enhanced in the PRL3-WT and -C104S (phosphatase dead) expressing HEK293 clones, relative to endogenous levels of PRL3 transcript in vector counterparts as observed through RT-PCR using PRL3-specific oligonucleotides. 18S-rRNA was used as control. RFP (red fluorescent protein)-tagged PRL3-WT protein is localized on endomembranes following stable ectopic expression in HEK293 cells as assessed by confocal microscopy. B, PRL3-WT cells have an unmistakable "spindle-like" fibroblast/mesenchymal morphology as compared with their "squamous" epithelial PRL3-C104S "phosphatase-dead" and vector counterparts. C, PRL3-WT clones have enhanced "global" tyrosine phosphorylation ("pan" pTyr-␣), a markedly less latent pTyr527-Src population (pTyr527-␣), and constitutive phosphorylated/activated ERK1/2 (pThr202/ pTyr204-␣) and STAT3 (pTyr705-␣), relative to vector counterparts as measured by phospho-specific immunoblotting.
of tyrosine phosphorylation in this study (supplemental Fig.  S2).
PRL3 Provokes an Aberrant Tyrosine Phosphoproteome to Drive Prometastatic Signal Transduction-Investigators have used a candidate approach in a variety of ectopic PRL3 expressing cell culture models in an attempt to shed light on the mechanism by which PRL3 potentiates prometastatic cellular behaviors when in excess. Collectively, data show that the consequences of PRL3 overexpression are pleiotropic, supporting a general notion that PRL3 potentiates the aberrant activation of a multitude of proto-oncogenic signal transducers downstream of an unknown substrate(s). We postulated that the various conclusions proposed regarding the downstream consequences of PRL3 overexpression collectively resembled the pleiotropic response elicited by transmembrane growth factor, cytokine, and integrin receptors following extracellular ligand stimulation. A central phenomenon of transmembrane receptor activation is phosphotyrosine-mediated assembly of signal transduction networks responsible for driving diverse cellular functions. Interestingly, a common theme among the various candidate approaches reported to date is the intracellular activation of oncogenic signaling by PRL3 overexpression alone. Our phosphoproteomic data afforded us the opportunity to hypothesize that a transmembrane receptor-mediated signal transduction network is being assembled and hijacked by an aberrantly activated intracellular tyrosine kinase(s) downstream of PRL3. In light of this hypothesis, we set out to resolve our understanding of the "PRL3-mediated signaling network." As a primary method of data analysis we organized the entire phosphoproteomic data set (supplemental Table S1) into bio-functional categories and created a summary of the comparative analysis between the PRL3 and vector datasets (Fig. 4A). This analysis reveals that the "Cellular Communication and Signal Transduction" bin encompassing protein kinases, protein phosphatases, adaptor/scaffolds, G-proteins, and lipases shows the most prominent difference with regard to the sheer number and identities of phosphoproteins between the two data sets. Strikingly, within this group, 75% (92 of 123) of the phosphoproteins identified in the PRL3 cells are unique from those observed from vector. Consistent with our hypothesis, this observation provides evidence that a transmembrane receptor-mediated phosphotyrosine-dependent signal transduction network is exploited in the PRL3 cells. Additionally, we took an unbiased approach toward understanding how our data best fit into canonical signaling networks by employing Ingenuity Pathway Analysis (IPA). We queried the Ingenuity Knowledge Base using proteins from our phosphoproteomic dataset. The top 10 bio-functions and canonical pathways predicted to be significantly represented from our data set are shown in Fig. 4B. PDGF-, Neuregulin-, PAK (p21/Cdc42/Rac1-activated kinase)-, Ephrin-, Interleukin-, and Integrin-signaling (all with -log(B-H p values) ϳ 12.5-15), collectively encompass a set of signal transducers well known to drive promigratory, proliferative, and survival decisions including: Cav1, ERK1/2, FAK, Jak1, JNK1, Nck2, p85⅐PI3K, PLC␥1, Ras, Shp2, Src, STAT3, STAT5B, and N-WASP. Using information gleaned from IPA as well as a manual literature review, we accumulated a list of phosphoproteins experiencing significant changes in tyrosine phosphorylation as a result of PRL3 expression from our data set that have characteristic molecular-/bio-functions deemed to be most relevant to PRL3-biology. This list of phosphoproteins, selected from the total dataset (supplemental Table S1) is summarized in Table I. Based on information presented in Table I, we constructed a model of the PRL3-mediated signaling network (Fig. 5). The model shows the molecular signature of a competent phosphotyrosine-dependent signaling network including: transmembrane receptor tyrosine kinases (RTKs) and integrin receptors, molecular adaptor/scaffolds, signal transducers/enzymes (e.g. kinases, phosphatases, Gproteins, and lipases), and transcription factors. The PDGF (␣/␤)-, Eph (A2/B3/B4)-, and Integrin (␤1/␤5)-receptors are the only transmembrane receptors shown to be phosphorylated on tyrosine in the PRL3 cells, suggesting that signaling coordinated by this receptor array is being hijacked by an aberrantly activated intracellular tyrosine kinase(s) downstream of PRL3. Interestingly, tyrosine phosphorylation present on this receptor array defines the signal transduction downstream of PRL3 as being mesenchymal in nature as these receptors are well known to be used by migratory mesenchymal cells during development as well as during acute wound healing in the adult animal (31)(32)(33).
Our data suggest that the PDGF␣-receptor is exclusively phosphorylated on Tyr-613, -720, -731, -742, and -988, whereas the PDGF␤-receptor is phosphorylated on Tyr-683, -692, -857, and -970 in the PRL3 cells. PDGF␣-receptor pTyr720 has been shown to coordinate Grb2, Shb, and Shp2 (34 -36), pTyr731 and 742 have been shown to coordinate the p85 regulatory subunit of phosphatidylinositol-3 kinase (PI3K) (37), and pTyr988 has been shown to be an autophosphorylation site and to coordinate phospholipase-C gamma 1 (PLC␥1) (38). Intra-and intermolecular regulation by the observed PDGF␤-receptor pTyr-residues remain enigmatic to date, with exception to pTyr970, a recognized c-Abl phosphorylation site (39). Consistent with the presence of a coordination site for Grb2, Shb, and Shp2 at pTyr720 and PLC␥1 FIG. 3. Quality of mass spectra used for SILAC-based quantitative assessment of tyrosine phosphorylation. Representative fragmentation spectra for phosphopeptides representing pTyr187 of ERK2 ( 173 VADPDHDHTGFLTEY[PO 3 2-]VATR), pTyr705 of STAT3 (K. 686 YCRPESQEHPEADPGSAAPY[PO 3 2-]LK), and pTyr417 of PAG1 ( 414 ENDY[PO 3 2-]ESISDLQQGR) showing the quality of mass spectra used for SILAC-based quantitative assessment of tyrosine phosphorylation. Raw fragmentation "sequencing" spectra including parent ion abundances (SILAC-based quantitative spectra: SILAC-"Light" (vector; blue points); SILAC-"Heavy" (PRL-3; red points)). b-ion series is colored in "red" and y-ion series in "blue." at pTyr988 of the PDGFR␣, strong evidence exists for downstream ERK1/2 activation in the PRL3 cells. Phosphopeptides representing pTyr157 of both highly transformative N-and K-Ras are up 22-fold and exclusively present in the PRL3 cells, respectively. Though the physiological relevance of pTyr157 remains unclear, this residue is part of the membrane-binding switch region that is postulated to govern activation state-dependent membrane engagement of this Gprotein. Furthermore, mutation of the highly conserved Phe156 residue, immediately adjacent to Tyr157, is shown to significantly increase Ras-GTP levels as well as Ras transforming activity, in vivo (40), suggesting that modification of Tyr157 would engender similar gain-of-function consequences. In addition, exclusive representation of phosphopeptides encompassing pTyr246 of the adaptor protein Shb and the activating pTyr542 modification of the proto-oncogenic Shp2 tyrosine phosphatase, provide further evidence in support of a Ras-dependent mechanism of ERK1/2 activation in the PRL3 cells. Although Shb has been shown to contribute to the coordination of Grb2 (41)(42), for SOS1-mediated direct Ras activation, Shp2 contributes to indirect Ras activation by attenuating the activities and/or localization of negative regulators of Ras by various mechanisms (43). Alternatively, evidence in support of a Ras-independent mechanism of ERK1/2 activation in the PRL3 cells comes from the exclusive representation of phosphopeptides encompassing pTyr-771, -977, and -1253 of proto-oncogenic PLC␥1, whereas a phosphopeptide representing pTyr783 of this enzyme is up 10fold. Through the hydrolysis of PI(4,5)P 2 (PIP 2 ), PLC␥1 drives the activation of "classic" (c) and "novel" (n)-type PKC-family members (44 -47). PKC may then induce activation of ERK1/2 through the phosphorylation/activation of Raf1 (48 -49) in the PRL3 cells. Finally, corroborating previous immunoblot-derived evidence for ERK1/2 activation, phosphopeptides representing pTyr204 of ERK1 and pTyr187 of ERK2 are up fourand fivefold in the PRL3 cells, respectively.
Consistent with the presence of coordination sites for the p85 regulatory subunit of PI3K at pTyr-731 and -742 of the PDGFR␣, phosphopeptides representing pTyr-73, -150 (exclusively present), and -580 (up 31-fold) in p85 provide further evidence in support of PI3K activation in the PRL3 cells. p85 has been shown to be phosphorylated in response to a variety of stimuli including being phosphorylated by Src-family kinases (SFKs), an event that relieves the SH2 domain-mediated inhibitory interaction that p85 makes with the p110 catalytic domain of PI3K in a latent state, leading to PI3K activation (50). p85 has also been shown to bind Shb (51), further validating coincident presence of these two phospho-

FIG. 4. Ectopic PRL3 expression induces aberrant activation of mitogenic and chemotactic signal transduction.
A, Phosphoproteomic study summary in pie-chart format (sub-categories comprising each bio-functional bin can be seen in the entire curated dataset represented in supplemental Table S1). 75% of phosphoproteins are unique to the PRL3 expressing cells and concentrate within "Cellular Communication and Signal Transduction." B, Ingenuity Pathway Analysis (IPA) results showing the Top 10 biofunctions and canonical pathways predicted from the phosphoproteomic data set. Data significance is represented using a B-H (Benjamini-Hochberg) p value adjustment to the false-discoveryrate (FDR; q-value). Phosphoproteomic data provide evidence in support of mitogenic and chemotactic signal transduction being prominently affected following ectopic PRL3 expression including significant representation of networks associated with PDGF, Neuregulin, p21/Cdc42/ Rac1-activated kinase (Pak), Ephrin (Eph) receptor, Interleukin, and Integrin signaling. Table S1) providing evidence in support of signal transduction responsible for driving pro-metastatic molecular events following ectopic PRL3 expression. A graphical model representing this data is presented in Figure 5.

PRL3-Mediated Signaling Network
proteins within the proposed network. Phosphorylation of p85 provides a plausible mechanism whereby PI(3, 4, 5)P 3 (PIP 3 ) production and concomitant Akt activation can be potentiated in the PRL3 cells. Evidence in support of STAT activation is also prominent in the PRL3 cells. Phosphopeptides representing pTyr1034 of Jak1 as well as pTyr705 of STAT3 and pTyr699 of STAT5B are up sixfold and exclusively present in the PRL3 cells, respectively. Though specific pTyr-based coordination sites on the PDGF␣/␤ receptors for Jak1, STAT3, or STAT5B remain uncharacterized, Src-dependent activation of STAT3 downstream of PDGF-stimulation and PDGF␤-receptor association is well-recognized (52)(53)(54)(55). Additionally, STATs are activated by Src-mediated Jak-dependent/-independent mechanisms, including the constitutive tyrosine phosphorylation and activation of Jak1 as well as STAT3 in v-Src transformed cells (56 -64).
In addition to the tyrosine phosphorylation initiated from RTKs, such as the PDGFRs, integrin receptor complexes induce tyrosine phosphorylation following dynamic engagement of the ECM. Our phosphoproteomic data provide strong evidence in support of integrin receptor complex activation including the downstream phosphorylation/activation of Rhofamily GTPases in the PRL3 cells (Fig. 5). Phosphopeptides representing pTyr783 of integrin-␤1 and pTyr774 of integrin-␤5 shown to be exclusively present and up 15-fold in the PRL3 cells, respectively, support these fibronectin receptors as the predominant integrin receptors associated with the PRL3-mediated signaling network. Tyr783 of integrin-␤1 and Tyr774 of integrin-␤5 are part of "NPxY" motifs that are known coordination sites for talin-1, a key molecular scaffold that links integrin receptors to the actin cytoskeleton (65)(66). Consistent with the presence of coordination sites for talin-1 on the integrin-␤1/␤5 receptors, phosphopeptides representing pTyr-70 and -71 of talin-1 are up 14-fold in the PRL3 cells, supporting the critical integrin receptor-actin cytoskeleton link that is essential to cellular migration.
We previously documented Src-dependent tyrosine phosphorylation of p130Cas in the PRL3 cells (14). p130Cas represents the major molecular scaffold of activated integrin receptor complexes that integrates both integrin-and RTKmediated signals toward Rho-family GTPase and MAPK activation when hyperphosphorylated (67)(68)(69). Phosphopeptides representing the activating pTyr397 modification of FAK as well as pTyr570 shown to be up 12-fold and exclusively present in the PRL3 cells, respectively, provide evidence of the establishment of the FAK-Src kinase complex responsible for p130Cas phosphorylation (70 -72). Additionally, a phosphopeptide representing pTyr798 (orthologous to pTyr789 of isoform 2) of receptor protein tyrosine phosphatase-␣ (RPTP␣) is up twofold in the PRL3 cells, a modification known to be associated with Src activation and the potentiation of cellular migration downstream of active integrin receptors and the Src-FAK complex (73)(74)(75). Directly from p130Cas, a Crk-

PRL3-Mediated Signaling Network
dedicator of cytokinesis-180 (DOCK180; DOCK1) Rac1-GEF complex responsible for driving Rac1 activation (69 -71, 76) is proposed to be relevant in the PRL3 cells. A phosphopeptide representing pTyr522 of DOCK7, a member of the DOCK180family of Rac1-GEFs expressed at high levels in the brain and heart, shown to be exclusively present in the PRL3 cells, provides evidence in support of this proposal. Additional evidence supporting the assembly of a p130Cas-mediated Rac1-activation complex comes from phosphopeptides representing pTyr246 of Shb and pTyr50 of Nck2 (Grb4), shown to be exclusively present and up 17-fold in the PRL3 cells, respectively. Relevant to the activation of the FAK-Src complex, the adaptor protein Shb is known to bind Src and FAK and become phosphorylated in a Src-dependent manner (51,77). Phosphorylated Shb has also been shown to bind the SH2 domain of Crk (78), providing additional evidence supporting the assembly of a p130Cas-Crk-DOCK180 (DOCK7) Rac1-activation complex. Nck2, like Shb, is an SH-domain-containing adaptor protein that plays a major role in driving cytoskeletal re-organization and cell movement following chemotactic-RTK activation (79 -81). Nck2 is known to be tyrosine phosphorylated following growth factor stimulation and in Src transformed cells (82). Additionally, this phospho-adaptor protein binds p130Cas, DOCK180, and the p21/Cdc42/ Rac1-activated protein kinase (Pak1) (79,(83)(84)(85)(86), providing further precedent for p130Cas-mediated Rac1 activation in the PRL3 cells. Adding to the relevance of Nck2 and Rac1 activation in the PRL3 cells is the exclusive presence of phosphopeptides representing pTyr-588 and -594 of Ephrin receptor A2 (EphA2), phosphorylation sites known to coordinate the Nck adaptor proteins as well as Vav2/3 Rac1-GEFs (32,87).
Though the functional relevance of pTyr792 of EphB3 and pTyr774 of EphB4, remains unclear, phosphorylated B-type Eph-receptors also mediate Nck2 association toward the cytoskeletal re-organization observed during ephrinB forward and reverse signaling (32,83,88). Finally, evidence suggest- Graphical model depicting phosphoproteomic data present within the PRL3 dataset that supports a mechanism by which PRL3 potentiates pro-metastatic signal transduction by exploiting "oncogenic" Src kinase activity. Src activates a signal transduction network coordinated by a mitogenic and chemotactic PDGF (␣/␤)-, Eph (A2/B3/B4)-, and Integrin (␤1/␤5)receptor array in the PRL3 cells. Proteins highlighted within green ovals represent phosphoproteins shown by both qualitative and SILACbased quantitative mass analysis to be either significantly up-regulated (pTyr-residue; blue circles) or exclusively present (pTyr-residue; red circles) within the PRL3 data set. Nonhighlighted proteins represent members of canonical signal transduction pathways assumed to be activated based on data present within the PRL3 phosphoproteomic dataset. pTyr-residue number designation is based on documented (NP and GI numbers) in supplemental Table S1.

PRL3-Mediated Signaling Network
ing Rac1 activation comes from a phosphopeptide representing pTyr185 of the c-Jun N-terminal kinase (JNK1), shown to be exclusively present in the PRL3 cells. JNK1 activation is well documented downstream of Rac1 activation via Pak1 kinase activity (89 -90).
Evidence in support of the activation of Cdc42, another member of the Rho-family GTPases, is present in our phosphoproteomic dataset. A phosphopeptide representing pTyr827 of the activated Cdc42 kinase-1 (ACK1) is up 4-fold in the PRL3 cells. ACK1 acts downstream of a multitude of transmembrane receptors following its activation by SFKs (most notably Src) with the predominant role of keeping Cdc42 in an active GTP-bound form (91)(92)(93). Furthermore, evidence in support of ACK1 activation and a strong biomarker of dynamic actin-cytoskeletal re-organization comes from a phosphopeptide representing pTyr256 of the putative ACK1 substrate, neural-Wiskott-Aldrich Syndrome protein (WASL/N-WASP), which is up 9-fold in the PRL3 cells. N-WASP is fundamental to filamentous actin growth and stabilization through its association with both the Arp2/3 actin nucleation complex and activated members of the Rho-family GTPases (94 -95). Nck2 also associates with both ACK and N-WASP (96 -98) providing further credibility to complex formation responsible for Rho-family GTPase activation and actin polymerization in the PRL3 cells.
Taken together, our phosphoproteomic data detail a rich signal transduction network downstream of a mitogenic and chemotactic PDGF (␣/␤), Eph (A2/B3/B4), and Integrin (␤1/␤5) receptor array in the PRL3 cells (Fig. 5). Importantly, the data support the collective tyrosine phosphorylation and plausible activation of ERK1/2, PI3K-Akt, Rho-family GTPases (Cdc42 and Rac1), and STAT3/5 under a "master" kinase signal transducer responsible for driving ligand-independent intracellular activation of these effectors. To ensure that the observed phosphorylation changes represent different modification stoichiometry, not changes in expression of the modified proteins, we have measured the level of protein expression for several key signaling molecules including integrin-␤1, Jak1, FAK, p85, Shp2, Ras, and Caveolin-1, which display increased tyrosine phosphorylation in PRL3 expressing cells. As shown in supplemental Fig. S3, the levels of protein expression for integrin-␤1, Jak1, FAK, p85, Shp2, Ras, Caveolin-1 in PRL3 expressing cells are similar to those in the vector control cells. This finding together with our early data showing no significant difference in relative protein abundance on a global scale between the PRL3 cells and vector counterparts (15) and similar protein expression observed in both PRL3 and vector control cells for Src, ERK1/2, STAT3, PDGFR␣, PDGFR␤, and PLC␥ ( Fig. 1 and Fig. 6), lend strong support that the reported variations in phosphopeptides are not because of alterations in protein abundance but do reflect changes in modification stoichiometry.
Src Kinase is the Major Intracellular Kinase That Hijacks the PRL3-mediated Signaling Network-In our previous work (14) we provided strong evidence of the impact that Src kinase activity has on the significantly increased proliferative and migratory/invasive response of PRL3 cells, relative to vector counterparts using both the Src kinase chemical inhibitor (SU6656) and Csk-rescue. In an effort to expand on observations made at the molecular level with regard to phosphorylation mediated by the Src kinase that would aid in potentiating the aforementioned bio-functional responses in the PRL3 cells including: pTyr187 of ERK1/2, p130Cas, and pTyr705 of STAT3 (14), we chose to further highlight the penetrance of Src activity in the PRL3 cells from data generated from this phosphoproteomic investigation.
A primary observation made from the PRL3 phosphoproteomic data set is the exclusive or increased presence of phosphopeptides derived from 10 putative Src substrates including: pTyr100 of calmodulin, pTyr14 of caveolin-1, pTyr397 of FAK, pTyr44 of ␥-enolase, pTyr783 of integrin-␤1, pTyr774 of integrin-␤5, pTyr783 and pTyr1253 of PLC␥1, pTyr798 of RPTP␣, pTyr705 of STAT3, and pTyr699 of STAT5B as assessed through both information present at phosphosite.org and manual literature review. Among these, RPTP␣ is known to be a direct positive regulator of Src-family kinase (SFK) signaling through dephosphorylating the inhibitory phosphotyrosine-residues of these enzymes (73)(74). As stated earlier, phosphorylation of Tyr798 has been shown to be Src-FAK-dependent and be an activating event to this phosphatase attributed to the potentiation of cellular migration (75).
Intriguingly, while analyzing phosphoproteins exclusively present in the vector data sets to gain information about possible PRL3 substrates, we consistently came across Cskbinding protein-phosphoprotein associated with glycosphingolipid microdomains (Cbp-PAG1), a tumor suppressor known to negatively regulate Src activity by both Csk-dependent and -independent mechanisms when in a tyrosine phosphorylated state (99 -101). Absence of 6 phosphotyrosyl residues including the pTyr317 Csk binding site of PAG1 provides strong evidence, in addition to the previously documented down-regulation of Csk (14), of the endogenous Src population existing in an activated state in the PRL3 cells. Though PAG1 is the most attractive candidate with regard to our current model of Src activation to be a substrate of PRL3, the absence of six pTyr-residues within totally distinct tyrosine phosphorylation motifs, suggests that this phosphoprotein is most likely down-regulated and not dephosphorylated by a single phosphatase (PRL3) in the PRL3 cells. Ongoing effort in our laboratory is directed to uncover the relationship between PRL3 and PAG1.
Here, we chose to further validate the penetrance of Src kinase activity within the PRL3-mediated signaling network on what we postulate to be two novel and highly significant phosphoproteins associated with PRL3 biology (PDGFR␣/␤ and PLC␥1). Because of the importance of the PDGF-receptors in regulating both mitogenic and chemotactic signal transduction including the prominent role that Src plays in PDGFR-mediated signal transduction, we chose to validate the tyrosine phosphorylation state of these receptors (Fig.  6A). In validation of our mass spectrometry-based data, IP-Western blot data show the constitutive tyrosine phosphorylation of both PDGF-␣ and -␤ receptors in the PRL3 cells. Surprisingly, although the levels of PDGF␣-receptor are equivalent in both the PRL3 and vector cells, the PDGF␤receptor is selectively expressed and/or stabilized in the PRL3 cells. Furthermore, treatment of these cells with a small molecule Src kinase inhibitor (SU6656) at 2.5 M, a concentration shown to be selective for inhibition of Src kinase activity over the kinase activity of the PDGFR (52), revealed that the tyrosine phosphorylation of the PDGF␣-receptor is independent of Src activity, whereas the tyrosine phosphorylation of the PDGF␤-receptor is dependent on Src activity as measured by this assay. This data is in agreement with previous published results showing a functional link between the PDGF-BB ligand, PDGF␤-receptor, and Src (SFKs) (102)(103)(104)(105). We postulated that the most likely mechanism responsible for the constitutive phosphorylation of the PDGF␣-receptor would be trans-autophosphorylation, because of a heightened sensitivity for serum-derived PDGF ligands driven by the significantly altered signal transduction present in the PRL3 cells, relative to vector counterparts. The observed selective expression and/or stabilization of the PDGF␤-receptor in the PRL3 cells is interesting in that signaling from the PDGF␤-receptor has major implications in developmental neo-vascularization and in tumor-angiogenesis, processes absolutely required for late stage tumorigenesis and the survival of disseminating meta-  6. Ectopic PRL3 expression induces selective expression and/or stabilization of the PDGF␤-receptor and Src-dependent constitutive tyrosine phosphorylation of the PDGF␤-receptor and PLC␥1. A, The PDGF␤-receptor is selectively expressed and/or stabilized and is constitutive phosphorylated in the PRL3 cells as assessed by PDGFR␤ antibody-specific immuoprecipitation (IP) followed by "pan"-pTyr immunoblotting. PDGFR␤ phosphorylation is dependent on the activity of the Src kinase as assessed by the use of the Src kinase chemical inhibitor (SU6656) at 2.5 M for 2h. B, Phospholipase-C gamma 1 (PLC␥1) is constitutive phosphorylated in the PRL3 cells as assessed by phosphotyrosine-specific immunoblotting against pTyr-783. Phosphorylation of PLC␥1 on Tyr-783 is dependent on the activity of the Src kinase. The attenuation of ERK1/2 phosphorylation using SU6656 is used as a positive control for SU6656-mediated Src kinase inhibition based on our previous published results (14). C, Data validation and spectral quality for pTyr783-PLC␥1 ( 779 NPGFY[PO 3 2-]VEANPMPTFK): Raw fragmentation sequencing spectra including parent ion abundances (SILAC-based quantitative spectra: SILAC-Light (vector; blue points); SILAC-Heavy (PRL-3; red points)). b-ion series is colored in red and y-ion series in blue.

PRL3-Mediated Signaling Network
static cells (33). Data show that the relative abundance of the PDGF␤-receptor is not affected by inhibition of Src activity during the experimental time course (2h), suggesting that the selective expression and/or stabilization of this receptor could be upstream of Src. Further research is required to understand the relationship between PRL3 and the PDGF␤-receptor. Additionally, because of the importance of PLC␥1 in the biology downstream of the PDGF-receptors including the activation of proto-oncogenic PKC enzymes, we chose to validate our mass spectrometry data with regards to the activation state of this enzyme. Through phospho-specific immunoblotting, we show that PLC␥1 is exclusively phosphorylated on the putative Src substrate, Tyr783, an event essential for lipase activation, in the PRL3 cells (Fig. 6B). We show that the phosphorylation of Tyr783 is completely dependent on Src activity by treatment of the PRL3 cells with the Src kinase inhibitor, SU6656. This data provide insight into a possible mechanism by which PLC␥1 becomes fully activated by Src following its localization to tyrosine phosphorylated PDGF␣/␤-receptors. Raw fragmentation spectral data showing the extensive coverage of the phosphopeptide representing pTyr783 of PLC␥1 as well as SILAC-data showing the 10-fold increase in abundance of this phosphopeptide in the PRL3 cells relative to its abundance in the vector counterparts, can been seen in (Fig.  6C). Collectively, these data provide strong evidence in support of Src kinase activation being a prominent consequence of PRL3 expression. DISCUSSION Despite a large number of circumstantial observations implicating PRL3 as a causative factor toward cellular metastasis when in excess, mechanistic evidence is lacking at the molecular level with regard to a putative signaling network that could be responsible for governing pro-metastatic molecular events downstream of PRL3. Here, we expand on a previously published observation that PRL3 is able to induce the aberrant activation of the Src tyrosine kinase when ectopically expressed in squamous epithelial-derived HEK293 cells (14) and present the most extensive and comprehensive model to date of how PRL3 drives pro-metastatic signal transduction.
Following PRL3 expression, we do not know whether network-assembly or Src activation comes first, but can attest to Src activity being the driving force behind the enhanced migration/invasion and proliferation observed in this cell culture model (14). This makes our tyrosine-phosphoproteomic network relevant as a causative factor toward these enhanced bio-functional responses and provides us with a more thorough understanding of PRL3-biology. Our model of the PRL3mediated signaling network depicted in (Fig. 5), highlights canonical signal transduction normally propagated following extracellular ligand stimulation of tyrosine kinase-and integrin-receptors. We postulated that these receptors act predominately as signal coordinators rather than signal activators; used by Src to drive the intracellular activation of critical effectors of migration/invasion, proliferation, and survival including: FAK-Src, Jak-STAT3/5, JNK1, Ras-ERK1/2, Rhofamily GTPases (Cdc42 and Rac1), PI3K-Akt, and PLC␥1. Importantly, this model reveals a plausible mechanism whereby PRL3 potentiates pleiotropic bio-functional activities characteristic to disseminating cells by leveraging "oncogenic" Src as a primary signal transducer. To date, only integrin receptors have been linked to the downstream consequences of PRL3 activity (16 -18). We propose that a PDGF (␣/␤)-, Eph (A2/B3/B4)-, and Integrin (␤1/␤5)-receptor array is the "master" signal coordinator of the PRL3-mediated signaling network. Collectively, these receptors are critically important to the biology of migratory mesenchymal cells during development as well as during acute wound healing in the adult animal (31)(32)(33). This corroborates previously published results, including those of our own, implicating PRL3 as a factor capable of driving morphological and molecular alterations akin to an EMT (14, 17, 19, 24, 106 -108), an essential prerequisite to metastatic dissemination. In fact, we acquired additional proteomic data that support the mesenchymal state of the PRL3 cells, including the exclusive presence of a phosphopeptide representing pTyr117 of the mesenchymal marker, vimentin as well as an increased abundance of vimentin and related intermediate filament proteins including: peripherin (ϩ3.35-fold), desmin (ϩ2.64-fold), alpha-internexin (ϩ2.64-fold), glial fibrillary acidic protein (ϩ2.64-fold), and vimentin (ϩ2.0-fold) as assessed by a "label-free" quantitative tandem mass spectrometry approach (26 -28) (data not shown).
The data presented in this work is highly relevant to PRL3 biology. We present strong evidence that PRL3 is a phosphatase capable of exploiting mesenchymal signal transduction and toward that end, highlight a novel PRL3-effector, PDGFR␤, an RTK known to drive developmental neo-vascularization and tumor-angiogenesis. We provide a plausible mechanism by which PRL3 induces Src activation through the down-regulation or dephosphorylation of the PAG1 tumor suppressor. Importantly, this mechanism has been implicated in causing the metastatic progression of colorectal cancer, a cancer type where aberrant Src activity is prominent and the expression of PRL3 has been shown to be highly correlative with cancer grade and prognosis. Finally, we provide the most extensive and comprehensive model to date depicting how the various proto-oncogenic effectors shown through candidate approaches to be exploited downstream of PRL3 become collectively activated through Src as a primary signal transducer. Further research is needed to understand the true nature of PRL3-mediated Src activation and how a network, such as the one depicted in (Fig. 5), becomes exploited by Src to drive the pro-metastatic bio-functions known to be linked to PRL3. teomic data were collected on an LTQ Orbitrap Velos mass spectrometer purchased through a generous NIH high-end-instrumentation grant, S10RR025044 (W.A.T.). □ S This article contains supplemental Figs. S1 to S3 and All data associated with the results reported in this study including: MS raw files, result output files, and annotated MS/MS spectra can be found at the following data repository link (http://www.peptideatlas. org/PASS/PASS00206). Follow the 'Description and Usage Information for Archived Files' document as an instruction and usage reference to the files archived in this repository.