Global Effects of BCR/ABL and TEL/PDGFRβ Expression on the Proteome and Phosphoproteome

Many leukemic oncogenes form as a consequence of gene fusions or mutation that result in the activation or overexpression of a tyrosine kinase. To identify commonalities and differences in the action of two such kinases, breakpoint cluster region (BCR)/ABL and TEL/PDGFRβ, two-dimensional gel electrophoresis was employed to characterize their effects on the proteome. While both oncogenes affected expression of specific proteins, few common effects were observed. A number of proteins whose expression is altered by BCR/ABL, including gelsolin and stathmin, are related to cytoskeletal function whereas no such changes were seen in TEL/PDGFRβ-transfected cells. Treatment of cells with the kinase inhibitor STI571 for 4-h reversed changes in expression of some of these cytoskeletal proteins. Correspondingly, BCR/ABL-transfected cells were less responsive to chemotactic and chemokinetic stimuli than non-transfected cells and TEL/PDGFRβ-transfected Ba/F3 cells. Decreased motile response was reversed by a 16-h treatment with STI571. A phosphoprotein-specific gel stain was used to identify TEL/PDGFRβ and BCR/ABL-mediated changes in the phosphoproteome. These included changes on Crkl, Ras-GAP-binding protein 1, and for BCR/ABL, cytoskeletal proteins such as tubulin, and Nedd5. Decreased phosphorylation of Rho-GTPase dissociation inhibitor (Rho GDI) was also observed in BCR/ABL-transfected cells. This results in the activation of the Rho pathway, and treatment of cells with Y27632, an inhibitor of Rho kinase, inhibited DNA synthesis in BCR/ABL-transfected Ba/F3 cells but not TEL/PDGFRβ-expressing cells. Expression of a dominant-negative RhoA inhibited both DNA synthesis and transwell migration, demonstrating the significance of this pathway in BCR/ABL-mediated transformation.

Chronic myeloid leukemia (CML) 1 is a disease with a characteristic t(9:22) chromosomal translocation giving rise to the Philadelphia chromosome (1). This translocation results in the juxtaposition of the BCR (breakpoint cluster region) gene and the c-ABL oncogene resulting in the constitutive expression of a BCR/ABL fusion oncoprotein (2). Another chimeric leukemogenic oncogene product, TEL/PDGFR␤, isolated from chronic myelomonocytic leukemia (CMML) patients bearing a t (5,12) translocation, also has constitutive tyrosine kinase activity.
The transforming ability of BCR/ABL and TEL/PDGFR␤ resides in their protein-tyrosine kinase activity. A number of signaling proteins activated by BCR/ABL have been identified including Ras, STAT5, protein kinase C, and phosphatidylinositol (PI) 3-kinase (3)(4)(5)(6). BCR/ABL is also known to affect cell adhesion and motility, in part via its ability to bind actin (7), as well as altering the expression or activity of focal adhesion proteins (8). A detailed review of these pathways and downstream effectors can be found in Ref. 9. TEL/PDGFR␤ can also activate signaling proteins such as PI 3-kinase, STAT1, STAT5, and stress-activated protein kinase (10 -12). As with BCR/ABL, Myc is an essential element in transformation by TEL/PDGFR␤ (13).
Recently, an inhibitor of the BCR/ABL protein-tyrosine kinase activity has shown great promise for the treatment of CML. STI571 (also known as Imatinib mesylate or Gleevec TM ) can inhibit BCR/ABL kinase activity both in vitro and in intact cells. STI571 can inhibit proliferation and induce apoptosis in BCR/ABL-and TEL/PDGFR␤-expressing cells (14,15). Clinical trials using STI571 have demonstrated impressive hematologic and cytogenetic responses in CML patients (16) and also where rearrangement of the PDGFR␤ gene has occurred (17). However, despite this success, there are cases of STI571-resistant BCR/ABL-positive CML (18). Thus the mechanistic detail of CML and CMML development remains a significant research objective to identify potential (common) targets for therapy.
Microarray experiments on primary CML samples have shown that the expression of BCR/ABL can alter transcript levels of genes involved in a wide variety of cellular processes, with poor disease prognosis in chronic phase being associated with changes in DNA repair, cell cycle, and cell adhesion path-ways, as well as STAT5 and Myc pathway targets. Whether these are directly attributable to BCR/ABL activity or secondary changes within the tumor was not determined (19). Studies using transfected cells plus screening for altered transcript levels by subtractive hybridization show alterations in genes involved in, for example, the MAP kinase (TOPK, or T-LAK cell-originated protein, a MAP kinase kinase), ubiquitination (HSPC150), or protein transport (NUP98 and RAN) pathways (20). A recent study using an inducible p210 BCR/ABL system screened using cDNA microarrays revealed that BCR/ABL upregulates a range in interferon-inducible genes, as well as transcription factors (STAT1, JUN), and cell growth and differentiation-related genes (PCNA, REL, Stathmin) (21). The situation with TEL/PDGFR␤-expressing cells remains relatively uninvestigated.
BCR/ABL can also initiate key changes that occur only at the proteome level. However, this phenomenon has also not been explored for TEL/PDGFR␤. BCR/ABL is known to decrease expression of the Abl inhibitory proteins via a proteasomemediated mechanism (22). Other proteins whose expression alters with no apparent change in mRNA abundance as a consequence of BCR/ABL action include p53 (down-regulated as a result of post-translational-mediated MDM2 overexpression) (23,24) and C/EBP␣, whose mRNA is bound by an inhibitory poly(rC)-binding protein hnRNP E2 (25). These proteins have important functions in hematopoietic cell survival and differentiation; thus, post-translational regulation of protein levels can be seen to have a role in transformation processes.
A model system for comparing the transforming effects of oncogenes is the Ba/F3 murine cytokine-dependent cell line transformed with BCR/ABL and TEL/PDGFR␤, respectively (26). We have used two-dimensional gel electrophoresis with the objective of comparing the effects of BCR/ABL and TEL/ PDGFR␤ on Ba/F3 cells by studying protein expression and phosphoprotein profiles with or without treatment with STI571. Very different outcomes are generated in Ba/F3 cells expressing these chimeric tyrosine kinases indicating pleiotropic mechanisms lead to or are associated with transformation by these 2 leukemogenic kinases.

EXPERIMENTAL PROCEDURES
Cell Lines-Ba/F3 cells were transduced with either an empty MSCV (murine stem cell virus) retroviral vector or MSCV containing the BCR/ABL or TEL/PDGFR␤ gene, respectively. The resultant Ba/F3-BCR/ABL or Ba/F3-TEL/PDGFR␤ cells were maintained in suspension in culture in RPMI with 10% (v/v) fetal bovine serum. Ba/F3-MSCV cells were grown in RPMI with 10% (v/v) fetal bovine serum supplemented with 1 mg/ml G418 (Sigma) and 1 ng/ml IL-3 (R&D Systems, Minneapolis, MN). Prior to lysis, cells were either starved for 4 h in RPMI with 1% (w/v) bovine serum albumin (Sigma) and, where appropriate, 5 M STI571, or were treated with 50 M Y27632 (Merck Biosciences, Nottingham, UK) for 6 h. Cells were washed twice in ice-cold PBS and once in ice-cold 250 mM sucrose with 0.4 mM sodium orthovanadate (Sigma), and cell pellets were stored at Ϫ80°C until use. Further transduction with Rho N19 was performed as described previously (27). Successfully transduced cells were selected for expression of green fluorescent protein marker using a flow cytometer (BD Biosciences). Cells were stained using May-Grunwald-Giemsa (MGG) stain as described previously (27).
Two-dimensional Gel Electrophoresis-Cells were lysed in 9 M urea, 2 M thiourea, 4% (w/v) CHAPS, 1% (w/v) dithiothreitol, and 2% IPG buffers (Amersham Biosciences, Little Chalfont, UK), and protein concentration was determined using the Bio-Rad modified Bradford protein assay. For silver-stained gels, 100 g of protein (1 mg for gels used for spot identification) was loaded by in-gel rehydration onto 24 cm, pH 3-10 nl IPG strips (Amersham Biosciences) in a total volume of 450 l of lysis buffer with a trace of Orange G (Sigma). For ProQ® Diamond (Molecular Probes, Leiden, The Netherlands)-stained gels, 500 g of protein were loaded onto 18-cm pH 4 -7 IPG strips (Amersham Biosciences) in a 350-l final volume. Strips were rehydrated at room temperature overnight, transferred to a Multiphor II apparatus (Am-ersham Biosciences), and protein focused over 2 days for a total of 115 kV h. Second dimension separation was carried out on 10%T SDS-PAGE gels with 4%T stacking gel using a Hoeffer vertical electrophoresis system (Amersham Biosciences) at 18 mA/gel overnight until the dye front reached the end of the gel. Detailed protocols can be found at www.lrf.umist.ac.uk.
Silver Staining-Analytical gels were stained using a silver staining kit from OWL separation systems (Portsmouth, NH) employing a modified protocol, as described in Ref. 28. Preparative grade gels were stained with the mass spectrometry-compatible silver stain of Shevchenko et al. (29).
ProQ® Diamond and Coomassie Blue Staining-Gel staining with ProQ® Diamond was carried out according to the manufacturer's instructions. Briefly, gels were fixed in 50% (v/v) methanol, 10% (w/v) trichloroacetic acid overnight, followed by a second fix for 1 h, washed in water 4ϫ 15 min, and then stained with ProQ® Diamond for 4 h in the dark. Gels were then destained with 50 mM sodium acetate (pH 4.0) with 4% (v/v) acetonitrile for 2 ϫ 1 h, and a third wash overnight. Gels were imaged on a Typhoon TM 8600 scanner (Amersham Biosciences) using 532-nm excitation and 610-nm emission filters, with photomultiplier tube voltage set at 600 V. Selected gels were subsequently stained with Coomassie Blue to visualize the pattern of total protein and determine the specificity of the ProQ® Diamond stain. Gels were washed in water for 30 min, then stained in 10% (w/v) ammonium sulfate/2% (v/v) phosphoric acid with 0.1% Coomassie Blue G (Sigma) for 48 h. Staining solution was made up at least 24 h before use and diluted 4 parts stain to 1 part methanol before use. The gels were destained briefly in 50% (v/v) methanol and scanned on a Molecular Imager FX (Bio-Rad). Gels were stained with ProQ® Diamond, imaged, and stored until analysis had been completed. Spots were then excised from these gels using as a template a 1:1 scale image of the gel. Following cutting, gels were rescanned to ensure that the correct spot had been excised.
Image Analysis-All gel analysis was performed using Progenesis (Non-linear Dynamics, Newcastle, UK) software. Changes in spot intensity were deemed significant where the average normalized volume altered by greater than 1.5-fold between samples, with p Ͻ 0.05 from a Student's t test on 3 or more replicate gels. Spot-normalized volume defines the volume of a given spot as a percentage of the total volume of all spots in the gel.
In-gel Digestion and Protein Identification-Spots excised from silver-stained gels were first destained using a 50:50 mixture of 30 mM potassium III ferricyanide and 100 mM sodium thiosulfate. All spots were washed twice in water, then equilibrated three times with 25 mM ammonium bicarbonate, and dried by three washes in acetonitrile. Dried gel pieces were rehydrated in 25 ng/l trypsin (Promega, Southampton, UK) in 25 mM ammonium bicarbonate on ice for 20 min. A further 20 l of 25 mM ammonium bicarbonate was added to prevent drying out, and the gel was incubated at 37°C overnight. Supernatant was then dried to around 2 l in a SpeedVac centrifuge. Remaining peptides were extracted from the gel piece by addition of 50% (v/v) acetonitrile, 0.1% (v/v) trifluoroacetic acid and sonication for 15 min. Extracted peptides were added to the concentrated supernatant and dried in a SpeedVac centrifuge. Peptides were reconstituted in 50% (v/v) acetonitrile, 0.1% (v/v) trifluoroacetic acid and spotted along with 0.5 l of 10 mg/ml ␣-cyano 4-hydroxycinnamic acid (CHCA, Sigma) in 50% (v/v) acetonitrile, 0.1% (v/v) trifluoroacetic acid onto a MALDI target for analysis using a reflectron MALDI-ToF mass spectrometer (M@LDI, Micromass, Manchester). Spectra were internally calibrated using trypsin autolysis fragments at m/z 842.509 or 2211.104. Tandem MS experiments were performed with an ABI QSTAR-Pulsar XL mass spectrometer (ABI/Sciex, Thornhill, Ontario, Canada) using an on-line liquid chromatography system (Dionex, Amsterdam, Netherlands). Dried peptides were reconstituted in 2% (v/v) acetonitrile, 0.1% (v/v) formic acid and were separated in a C18 reverse phase column on-line with the Q-STAR using a gradient of 2-60% acetonitrile over 15 min. Protein identities were obtained by searching against either the SWISS-PROT or NCBI non-redundant data bases using MASCOT (Matrix Science, London).
Western Blotting-To confirm changes in the expression of cytoskeletal/motility-associated proteins, 20 g of cell lysate was separated on a 10% one-dimensional SDS-PAGE gel (this was increased to 50 g of lysate on a 15% SDS-PAGE gel for stathmin detection), and transferred to a nitrocellulose membrane using an Electro-Blot apparatus (Web Scientific, Crewe, UK) in 192 mM glycine, 25 mM Tris, 20% (v/v) methanol at 100 V for 45 min. Membranes were blocked in PBS with 0.1% Tween-20 (PBS-T) with 5% (w/v) dried nonfat milk (Marvel) at 4°C overnight. The membrane was then washed once in PBS-T, and incu-bated with primary antibodies for 1 h as follows; anti-stathmin rabbit polyclonal antibody (30) (a gift from E. Sobel, INSERM, Paris) at 1:10,000, rabbit anti-Arp2 (Chemicon International, Chandlers ford, UK) at 1:1,000, and rabbit anti-gelsolin serum (31) (a gift from D. Kwiatkowski, Harvard Medical School, Boston, MA) at 1:10,000. Membranes were washed three times, incubated in peroxidase-conjugated mouse anti-rabbit Ig (Amersham Biosciences) at 1:10,000 for 1 h, and washed four times. All antibody dilutions were in PBS-T with 1% w/v milk. All washes were performed with PBS-T. Detection was carried out using Supersignal West Pico chemiluminescent substrate (Pierce), and signal detected on Kodak x-ray (Sigma) film. All blots were subsequently stained with Coomassie Blue to ensure equal loading.
Chemotaxis and Cell Motility Assay-The migration of Ba/F3 cells in response to agonists was assessed using a 24-well transwell plate (Costar, Corning, New York). These consisted of two wells separated by a membrane containing 5-micron pores. Cells (1-2 ϫ 10 5 in 100 l) were placed in the top well, and agonists were added to the top and/or bottom wells (bottom well volume 600 l) in Fishers medium plus 20% (v/v) batch-tested horse serum. After 6 -8 h of incubation at 37°C in a 5% CO 2 humidified incubator, viable cells in the lower well were counted using trypan blue (Sigma). In no experiment was cell viability less than 98% in either top or bottom well after incubation.

BCR/ABL-and TEL/PDGFR␤-transfected Ba/F3 Cells Display Different Changes in Protein Expression
That Are Partially Reversed by STI571-Ba/F3 cells expressing either BCR/ ABL or TEL/PDGFR␤ exhibit a number of behavioral changes. Most notable among these, is that transduced cells no longer require the presence of growth factor (IL-3) for their survival or proliferation (12,26). Protein from Ba/F3 cells transfected with an empty vector (MSCV), and cells transfected to express either BCR/ABL or TEL/PDGFR␤, (with or without treatment with STI571 for 4 h) were separated using a pH 3-10 immobilized  Table I. Results shown are representative of at least five similar gels. pH gradient gel in the first dimension followed by SDS-PAGE. The spot patterns (from three gels) were analyzed to detect significant differences in levels of expression of specific proteins between the control Ba/F3 cells, Ba/F3-BCR/ABL cells and TEL/PDGFR␤-transfected Ba/F3 cells (Fig. 1). The latter two samples were also treated with STI571 and then compared with non-treated controls using two-dimensional electrophoresis (Fig. 1). The major changes in protein levels observed were quantified using Progenesis software. Proteins shown to alter in their expression according to the gel patterns were identified where possible using mass spectrometry, as shown in Table I.
Ba/F3-BCR/ABL cells showed significant changes in 32 proteins compared with Ba/F3 cells. Several protein functional groups can be identified in the list of proteins that change as a consequence of BCR/ABL expression, for example signaling proteins such as 14-3-3 proteins and Grb2, metabolic enzymes, and cytoskeleton regulators (see Table I). TEL/PDGFR␤ induced 13 significant changes in protein expression, and these TABLE I Identities of proteins whose expression is altered by expression of BCR/ABL or TEL/PDGFR␤, and/or treatment with STI571 Proteins whose expression level changes with BCR/ABL or TEL/PDGFR␤ with a significance greater than 0.05 and a fold change greater than 1.5 times are shown below. M and pI given are those predicted from amino acid sequence. In all cases the molecular mass and pI correlated to the region of the gel where the protein was found, helping confirm identity. Mean normalized spot volume is the average value from five replicate gels for each condition. showed some overlap to those seen with BCR/ABL. However there are a significantly fewer changes induced by TEL/ PDGFR␤ than by BCR/ABL: each tyrosine kinase, while inducing growth factor independence is not affecting the proteome in a similar fashion.

BCR/ABL but Not TEL/PDGFR␤ Induces Changes in Cytoskeletal Proteins and Regulators of the Motile Response-
When compared with control Ba/F3 cells, changes in cytoskeletal-associated proteins such as decreased levels of gelsolin, adseverin, and actin-related protein 2 (Arp2), and increased expression of stathmin are seen in BCR/ABL expressing Ba/F3 cells but not TEL/PDGFR␤-transfected cells. Gelsolin and adseverin are closely related proteins that play a role in actin filament severing and capping, and are therefore closely involved in cell motility (33). Arp2 is part of the Arp2/3 complex that is also associated with formation of short, branched actin filaments in the leading edge of motile cells (34). In contrast, tropomyosin, several forms of which were increased by both BCR/ABL and TEL/PDGFR␤ in this study, has been shown to cause annealing of gelsolin capped actin filaments (35), and can inhibit actin branch formation and nucleation by the Arp2/3 complex (36). TEL/PDGFR␤-transfected cells carry only one of these changes in proteins known to regulate motility and morphology, namely tropomyosin. This has been confirmed by Western blotting for gelsolin, actin-related protein, and stathmin (Fig. 2, A-C).
A notable change seen in both TEL/PDGFR␤-and BCR/ABLtransfected Ba/F3 cells compared with control Ba/F3 cells is proliferating cell nuclear antigen (PCNA). As all three populations are actively cycling (the latter via the inclusion of IL-3 to cultured Ba/F3 cells) this is a significant oncogene-induced difference in the proteome. Again, this finding was confirmed by Western blot analysis (Fig. 2D). 14-3-3 proteins are also altered, both in TEL/PDGFR␤-and BCR/ABL-transfected cells.
The treatment of Ba/F3-BCR/ABL cells with STI571 resulted in alterations in the two-dimensional-PAGE pattern, with the increased expression of 5 proteins (including adseverin) and the decrease of seven proteins following drug treatment (see Table I). Cells were treated with STI571 for only 4 h (as the cells began to undergo drug-induced apoptosis after this time). The full extent of the changes in the specific proteins listed above in viable cells may not therefore be represented in this 4-h incubation. Nonetheless the involvement of a caspase-mediated cleavage of gelsolin in apoptosis (37) gives a clear reason for avoiding examination of the proteome of STI 571-stimulated apoptosis in Ba/F3 cells. In fact we view it as indicative of the success of this strategy that gelsolin levels did not fall as a consequence of the 4-h incubation with STI571. In the case of Ba/F3-TEL/PDGFR␤ cells no spots were increased in expression levels by STI571 while seven spots were decreased. Thus STI571 can reverse some of the effects of TEL/PDGFR␤-or BCR/ABL transduction after just 4 h. Since STI571 has no effect on untransduced Ba/F3 cells (see Ref. 38 and phosphoprotein analysis described in this report), this combination was not analyzed.
BCR/ABL but Not TEL/PDGFR␤-expressing Ba/F3 Cells Display Altered Response to Chemotactic Stimuli-The weight of changes driven by BCR/ABL associated with the cytoskeleton and its regulation and the paucity of such effects in TEL/ PDGFR␤-transfected cells led us to consider the response of these cell lines to chemotactic factors such as stromal-derived factor 1 (SDF 1).
Importantly, oncogene-transduced cells show no changes in cell morphology as compared with mock-transfected cells (Fig.  3A). We assessed the response of Ba/F3 control, Ba/F3-TEL/ PDGFR␤, and Ba/F3-BCR/ABL cells to the chemokine SDF-1 (which induces a chemotactic response) in the presence and absence of the stem cell chemokinetic factor, lysophosphatidic acid (LPA). LPA has been shown to potentiate motile responses via the Vav guanyl nucleotide exchange factor that is activated by BCR/ABL (39). BCR/ABL expression decreased the response to both SDF-1 and LPA, and these effects were not relieved by addition of STI571 during the assay (Fig. 3B), although prior treatment with STI571 did abrogate this partial loss of motile function (Fig. 3B). The incubation with STI571 had no effect on cell viability. The lack of ability of the BCR/ABL kinase inhibitor to reverse the decreased motile response of the Ba/F3 cells correlates with the inability of STI571 to affect a change in the levels of all the cytoskeletal proteins whose levels are modulated by BCR/ABL over a short incubation time. We have previously shown that decreased gelsolin and adseverin expression is associated with a low level motile response in hematopoietic stem cells (40). TEL/PDGFR␤ showed no alteration in motility, correlating with the findings from the proteomics screen where changes in cytoskeletal proteins, with the exception of tropomyosin, were found only in BCR/ABL-expressing cells. Altered expression patterns of these proteins can therefore provide an explanation for the observed decrease in motility in the chemotaxis assays displayed by the BCR/ABL-expressing cells.
Identification of Alterations in Phosphoprotein Levels Using Two-dimensional Electrophoresis and Phosphoprotein Staining-As a rapid and objective means of identifying major phosphorylation events governed by BCR/ABL and TEL/PDGFR␤ we carried out an analysis of two-dimensional gels from control and transfected cells using a phosphoprotein stain, ProQ® Diamond. Triplicate gels containing lysate from Ba/F3-MSCV (empty vector) Ba/F3-TEL/PDGFR␤, and Ba/F3-BCR/ABL cells with or without treatment with STI571 were produced with a pH 4 -7 gradient in the first dimension. Preliminary experiments have shown that the majority of phosphoproteins are resolved on this gradient. Furthermore the majority of changes in the silver staining experiments described above were in this region and using a narrower pH range allowed more protein to be loaded, improving the probability of detecting lower abundance phosphoproteins. The use of known quantities of enriched phosphoproteins confirmed that there is an ability to detect the phosphoproteins casein and ovalbumin at the low picomole level after gel electrophoresis, with little or no detection of non-phosphorylated proteins such as albumin or alcohol dehydrogenase (data not shown).
When 500 g of protein from Ba/F3 cells were loaded in two-dimensional electrophoresis experiments and stained for phosphoproteins, an average of 296 Ϯ 7 spots were observed on the pI 4 -7 gels for Ba/F3 cells, 290 Ϯ 17 for Ba/F3-BCR/ABL cells, and 308 Ϯ 14 for Ba/F3-TEL/PDGFR␤ cells (Fig. 4). The addition of STI571 reduced the number of spots in BCR/ABL cells by 23.1% and in TEL/PDGFR␤ cells by 11%. This approach detects changes in major phosphoproteins and as such the similarities in numbers of stained proteins is to be anticipated, as 30% of the proteome is predicted to be made up of phosphoproteins and we observe ϳ1000 proteins from these samples with a total protein stain (silver). To detect differences between the molecular mode of action of BCR/ABL and TEL/ PDGFR␤ analysis of these spot patterns was followed by mass spectrometry where sufficient protein was available to enable detection of the protein in question (Table II). No significant differences were identified between the control Ba/F3 cells in the presence or absence of STI571, supporting the observation that STI571 treatment has no effect on untransfected Ba/F3 cells (38). BCR/ABL-transfected Ba/F3 cells displayed significant differences in phosphoprotein-specific staining as compared with control cells and TEL/PDGFR␤-transfected cells.
Expression of BCR/ABL affected the phosphoprotein staining of 32 spots. It caused a significant increase (greater than 2-fold, p Ͻ 0.05) in the fluorescence of 14 protein spots, and this effect is reversed with 4 h of STI571 treatment in 5 cases. BCR/ABL also causes decreased staining (greater than 2-fold decrease in signal intensity, p Ͻ 0.05) of 18 protein spots compared with Ba/F3 cells, with 7 of these changes being reversed affected by STI571. Interestingly, 1 spot appeared to alter upon STI571 treatment although it was not significantly different following BCR/ABL expression. This spot was found to contain a Rab GDP dissociation inhibitor ␤2, which is increased by BCR/ABL in a distinct (adjacent) spot implying an increase in a phosphorylated form in BCR/ABL expressing cells that is reversed upon treatment with STI571. The fact that only one form increases phosphorylation in BCR/ABL yet both are decreased by STI571 possibly suggests that addition of STI571 has a secondary effect on Rab GDI ␤2, which cannot be determined using the current experimental design.
Where the relative intensities of the silver-or Coomassie Bluestained and ProQ Diamond stained spots could be compared, the data implied that the majority of changes detected by this staining method are caused by alterations in specific phosphorylation per se. Only 60% of the changes could confidently be assessed in this manner due to the differences in spot patterns produced by the two stains. Likewise, the majority of proteins identified as being altered in expression in the silver staining experiments (Fig. 1) do not show any/altered phosphoprotein staining.
The most notable differences seen in BCR/ABL Ba/F3 cells include Crkl and enolase, both well known downstream targets of BCR/ABL. Each protein spot increased in ProQ Diamond staining in BCR/ABL expressing cells compared with control cells (Fig. 4 and Table II). This validates the method employed. Noteworthy is the fact that BCR/ABL decreased phosphorylation in a large number of proteins as well as increasing phosphorylation in others. Proteins whose fluorescence staining was modulated included those involved in protein synthesis (eukaryotic initiation factor 4A-1), chaperone proteins (Hsp90), proteasomal proteins (sumo 1-activating enzyme subunit 2, 26 S protease regulatory subunit), cytoskeletal proteins (tubulin ␤5, tubulin ␣6), and signal transduction proteins, including Toll-like receptor 2, Ras-GTPase-activating protein-binding protein, and Rho GDI. The only proteins commonly affected by TEL/PDGFR␤ and BCR/ABL were Ras-GTPase-activating protein-binding protein and eukaryotic initiation factor 4A-I. On three occasions two spots, one identified as being increased by oncogene and the other decreased, were found to contain the same protein (Rab GDP dissociation inhibitor ␤-2 in the BCR/ ABL analysis, Ras GAP-binding protein, and tubulin ␣-6 in the TEL/PDGFR␤ analysis). Presumably these spots contained the different phosphorylated forms of these proteins.
The dephosphorylation of Rho GDI1 in Ba/F3-expressing cells is of interest, as the Rho GTPase is known to be constitu- tively activated in BCR/ABL-expressing Ba/F3 cells (41). We confirmed the increase in GTP-bound Rho in BCR/ABL Ba/F3 cells (data not shown). As a principal target of Rho is Rho kinase we have analyzed the effect of Rho kinase inhibition on BCR/ABL-transfected Ba/F3 cells using the Y27632 inhibitor.
The Rho Kinase Inhibitor Y27632 Inhibits BCR/ABL-mediated Proliferation-Rho kinase is involved in motile behavior in many cell types (42). As expected, the Rho kinase inhibitor Y27632 markedly decreased the motile response of control, BCR/ABLand TEL/PDGFR␤-transfected Ba/F3 cells in transwell migration assays (data not shown) demonstrating the role of the kinase in agonist-stimulated motile responses in these cells (43).
The potential contribution of Rho kinase to the BCR/ABLand TEL/PDGFR␤-mediated mitogenic response was also analyzed. TEL/PDGFR␤-stimulated DNA synthesis was inhibited by 15 Ϯ 2% (mean Ϯ S.E., n ϭ 5) by 50 M Y27632, consistent with the role of Rho in cell cycle progression. BCR/ABL-medi-ated effects on DNA synthesis were inhibited more markedly, by 35 Ϯ 11% by addition of 50 M Y27632 (Fig. 5). Decreased inhibition was seen in the presence of IL-3. No additive or synergistic effects of the BCR/ABL inhibitor STI571 and the Rho kinase inhibitor Y27632 were observed at either suboptimal or optimal doses. This suggests that inhibition of BCR/ABL is also leading to Rho kinase inhibition, because no further effects on the Rho pathway can be observed upon addition of a Rho inhibitor when STI571 is present. This is consistent with data reported by Harnois et al. (41). Thus Rho kinase represents a novel target downstream from BCR/ABL.  Table II. Results shown are representative of at least three similar gels.

Dominant-negative Rho Decreases BCR/ABL-mediated Proliferative Responses-To
S.E., n ϭ 4) compared with Ba/F3-BCR/ABL cells. In the presence of IL-3 the difference between the Ba/F3-BCR/ABL-N19 and the Ba/F3-BCR/ABL cells was reduced to 26 Ϯ 3%. The effect of RhoA N19 on control Ba/F3 cells was a 34 Ϯ 7% decrease in IL-3-stimulated rate of DNA synthesis. Thus the BCR/ABL-mediated events leading to proliferation are more dependent on the Rho pathway.
RhoA N19 also decreased the rate of SDF-1-stimulated tran-swell migration in Ba/F3 cells by 36 Ϯ 8% (mean Ϯ S.E., n ϭ 3) and in Ba/F3-BCR/ABL cells by 60 Ϯ 12%. The chemotactic response is significantly affected by inhibition of Rho either in the presence or absence of BCR/ABL activity.

DISCUSSION
Many aspects of BCR/ABL-mediated signal transduction events have now been elucidated. In the case of the TEL/ TABLE II Identities of phosphoproteins whose expression or phosphorylation status is altered by expression of BCR/ABL or Tel/PDGFR␤, and/or treatment with STI571 M and pI given are those predicted from amino acid sequence. In cases where predicted pI/M do not match the gel position, the identity has been confirmed by repeated MALDI/PMF and by MS/MS. Mean normalized spot volume is the average value from three replicate gels for each condition. PDGFR␤-mediated events in hematopoietic cells far less is known than for BCR/ABL. The initial objective of the work described here was to compare and contrast downstream effects of these 2 oncogenes on the proteome and major effects observed on the phosphoproteome. We have shown that, while both TEL/PDGFR␤ and BCR/ABL can induce growth factor independence in Ba/F3 cells, the effects on the major elements of the proteome and the phosphoproteome are very different. Some of the BCR/ABL-and TEL/PDGFR␤-mediated effects on the proteome we observed have been seen previously in leukemic cells. For example, the Grb2 signaling protein, found to be up-regulated by BCR/ABL expression, is known to be a key component of BCR/ABL signaling (3), and is essential for the growth of BCR/ABL-expressing cells and the development of CML (44,45). Likewise, stathmin (oncoprotein 18), a microtubule-destabilizing protein, has been previously shown to be up-regulated in ALL and AML samples, which can both contain a BCR/ABL fusion protein (46,47). Stathmin is a key regulator in proliferation, with phosphorylation (occurring in response to cell proliferation) reducing its binding to tubulin and promoting the formation of microtubules key in cell division.
Another protein that has been linked to BCR/ABL signaling and was shown here to be regulated by BCR/ABL and TEL/ PDGFR␤ is the 14-3-3 family of proteins. 14-3-3 family members can associate with, and be phosphorylated by, BCR/ABL (48), and are involved in a wide variety of processes. Of particular interest in terms of the development of leukemias, is that 14-3-3 proteins can interact with protein kinase C (also found to be increased by BCR/ABL in the current study) (49) and provide a link to Raf-1, a downstream component of the Ras signaling pathway, which is activated in BCR/ABL-expressing cells. Phosphatidylethanolamine-binding protein, which was shown to be down-regulated by BCR/ABL and Tel/PDGFR␤, is an inhibitor of Raf-1 (50), implying that not only is Ras activation essential to transformation by BCR/ABL but also maintenance of the Raf 1 downstream effector activity via "secondary" effects. Protein kinase C (PKC)␦ is elevated in BCR/ABL-expressing cells. After BCR/ABL transformation, inhibition of this protein, but not other PKC isoforms, reduces anchorageindependent growth (5). There is also evidence that 14-3-3 proteins can inhibit apoptosis, either by direct interaction with the pro-apoptotic regulator Bax (51) or by regulation of MAP kinase cascades (52). Thus our proteomic survey reveals mechanistic detail for process of oncogenic transformation.
Only BCR/ABL stimulated the dysregulation of many cytoskeletal-associated proteins. BCR/ABL induced decreased levels of gelsolin, adseverin, and actin-related protein 2, and increased expression of tropomyosin, whereas of these only tropomyosin was affected by TEL/PDGFR␤. This observation is important, as altered adhesive and motile properties of CML cells has been linked to their escape from the marrow and loss of regulation by the marrow microenvironment (53).
Gelsolin and adseverin play a key role in cell motility (54) via actin filament severing and end capping (33). Their activity can be modulated by tropomyosin, which anneals gelsolin-capped actin and forms long actin filaments (35). The expression of adseverin in a megakaryoblastic leukemia model induces differentiation and inhibits proliferation and transforming ability (55). Arp2 also has a role in the formation of short, branched actin filaments in motile cells (34). Arp2 is part of the Arp2/3 complex that is key in the formation of new actin branches (56). Interestingly, tropomyosin can also impact on this process by inhibiting actin branch formation and nucleation by the Arp2/3 complex (36).
To confirm that BCR/ABL did indeed have an effect on motility that was not seen in TEL/PDGFR␤, transwell migration assays were performed to assess the response to the chemokine SDF-1 and LPA. BCR/ABL-expressing cells displayed a lower rate of motility than both control and TEL/PDGFR␤-expressing cells, and this effect could be reversed by the addition of STI571 to the cells. This confirms that not only does BCR/ABL affect motile behavior (see Ref. 43), but also that TEL/PDGFR␤ does not, providing further proof that the two oncogenes affect different pathways.
By comparison to BCR/ABL, TEL/PDGFR␤ had little effect on the proteome, and is certainly not affecting the panoply of cytoskeletal proteins potentiated by BCR/ABL expression. This correlates with the lack of effect of TEL/PDGFR␤ on motile responses to chemokines (see Fig. 3). The only proteins identified whose expression is altered by both oncogenes are the 14-3-3 family members, tropomyosin, phosphatidylethanolamine-binding protein, and PCNA, a component of the DNA polymerase (57). The presence of both the 14-3-3 proteins and phosphatidylethanolamine-binding protein in both lists may indicate that both oncogenes have an effect on Raf.
A potentially important protein affected by TEL/PDGFR␤ that did not appear to change when BCR/ABL is expressed is follistatin-related protein 1. This TGF-␤-inducible protein (58), which can also be up-regulated upon protein kinase C activation (59), has been previously demonstrated to down-regulate the activity of activin A (60), a TGF-␤ superfamily member. Activin A has been shown to have an important role in hematopoiesis, and can inhibit colony formation by peripheral blood granulocyte-monocyte colony-forming units when stimulated with IL-3. Activin A also abolishes the increased DNA synthesis seen when these cells are treated with IL-3 (61), suggesting that it plays a role in the maintenance of differentiation. Inhibition of this protein, for example by up-regulation of follistatin-related protein, may be important in leukemogenesis (62).
Whereas TEL/PDGFR␤ Is Not Affecting Motility, It Most Certainly Affects Growth of Ba/F3 Cells-To provide information about the relative effects of BCR/ABL and TEL/PDGFR␤ on processes that may govern proliferation, we utilized the ProQ® Diamond phosphoprotein-specific gel stain (63,64). Our objective was to identify major phosphorylation events associated with expression of the BCR/ABL and TEL/PDGFR␤, respectively.
The phosphoprotein alterations observed between the control and BCR/ABL-transfected samples again include a number of proteins, which are already known to be targets of the ABL kinase. Crkl, known to be one of the major proteins that binds to and is phosphorylated by BCR/ABL (65)(66)(67)(68). Heat shock protein 90 (Hsp90), is apparently highly phosphorylated in control cells but this is not so in BCR/ABL-expressing cells. Hsp90 has been identified as a potential therapeutic target in CML. It binds to BCR/ABL in a complex with p23 (69) and protects the BCR/ABL protein from proteasome-mediated degradation. Cells containing mutated BCR/ABL, resistant to STI571 treatment, are still responsive to Hsp90 inhibitors and show increased rates of BCR/ABL degradation (70). Phosphorylation of Hsp90 appears to release the Hsp90 substrate binding (71), and so the dephosphorylation seen in these cells is probably key to the stabilization of the BCR/ABL protein. The above proteins are unaffected by TEL/PDGFR␤ expression.
The Ras-GTPase-activating protein-binding protein 1 was dephosphorylated by expression of BCR/ABL and TEL/ PDGFR␤. It has been shown that this protein is hyperphosphorylated in quiescent cells (72). A key point here is that the Ba/F3 control cell population had been maintained in an actively proliferating state by maintaining the cells in IL-3 thus the difference between BCR/ABL-transfected and mock-transfected Ba/F3 will consist of more than the difference between dividing and non-dividing populations. Ras-GAP has been implicated in Rho-mediated remodeling of the cytoskeleton. Nedd5, a septin GTPase family member implicated in actin binding and cytokinesis (73) was also shown to be dephosphorylated in response to BCR/ABL expression, as was Rho-GDI1.
Another protein whose phosphorylation status is decreased by BCR/ABL (spot MB4) contains a tubulin tyrosine ligase (TTL) family domain (amino acids 341-638, Pfam PF03133). This domain catalyzes the post-translational addition of a tyrosine to the C terminus of ␣-tubulin and may be controlled by phosphorylation of the enzyme (74). In normally cycling cells, the tyrosinated form of tubulin predominates. In breast cancer cells, however, the detyrosinated form is more common and correlates to tumor aggressiveness (75). Loss of TTL activity also correlates to tumor progression (76). Plainly the phosphoprotein staining approach offers opportunities to understand further the processes of leukemogenesis.
Given the data from the total protein expression analysis where BCR/ABL was shown to have profound effects on components of the cytoskeleton and cell motility (see Table I), we were interested to note that BCR/ABL appeared to dephosphorylate Rho GDI1. Dephosphorylation of the Rho GDI1 protein destabilizes its interaction with RhoA in hematopoietic cells (77). Rho GDI association with Rho proteins maintains them in an inactive cytosolic form, when released from this constraint the G-proteins become membrane-associated and activated. Rho proteins are activated in BCR/ABL-transfected Ba/F3 cells (41). In part this has been ascribed to the p210 bcr/abl guanyl nucleotide exchange factor activity. The loss of Rho GDI inhibitory activity contributes to Rho activation. Rho GDI1 interacts directly with RhoB, Rac 1, Rac2, and Cdc42 (78). Activation of Rho GTPases result in the formation of actin stress fibers and membrane ruffling (79), features previously observed in BCR/ ABL-transfected cells (80). The significance of the observed change in Rho GDI phosphorylation was further investigated using the Rho kinase inhibitor Y27632, and a dominant-negative Rho transgene, Rho N19. We showed that BCR/ABL-stimulated DNA synthesis was sensitive to Rho kinase inhibitor whereas IL-3-stimulated cells were not. This suggests that the BCR/ABL-mediated dephosphorylation of Rho GDI and subsequent increase in active Rho and activation of Rho kinase play key roles in the growth and development of leukemias. Expression of the dominant-negative Rho confirmed that Rho activity is at least partially required for these processes, with N19 expression reducing thymidine incorporation and transwell migration. Pharmacologic agents are available for inhibition of Rho kinase, opening opportunities for treatment strategies in Imatinib-resistant CML. We are currently investigating this in primary CML cells.
We have examined alterations in the proteome and phosphoproteome resulting from expression of the leukemogenic tyrosine kinase BCR/ABL and TEL/PDGFR␤. The differences between the events stimulated by the two kinases strongly suggests that this is an approach that can aid in understanding specific pathophysiologic processes in leukemias associated with specific translocations. The approach provides starting points to further understand the role of BCR/ABL, TEL/ PDGFR␤, and other leukemogenic oncogenes in transformation using a common, clonal cell background, Ba/F3. One clear point is that BCR/ABL potentiates proteins involved in motility whereas no evidence exists in our data that TEL/PDGFR has a similar effect. Thus proteomics can define differences in the molecular mechanism of action of oncogenes.