Proteomics-Based Strategy to Identify Biomarkers and

May 16, 2006 - (MLL)1 gene on chromosome band 11q23 are present in most cases of acute ... III, 421 Curie Boulevard, Philadelphia PA 19104-6160...
1 downloads 0 Views 597KB Size
Proteomics-Based Strategy to Identify Biomarkers and Pharmacological Targets in Leukemias with t(4;11) Translocations Anastasia K. Yocum,† Christine M. Busch,† Carolyn A. Felix,‡ and Ian A. Blair*,† Center for Cancer Pharmacology, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania 19104-6160, and Division of Oncology, The Children’s Hospital of Philadelphia, Department of Pediatrics, University of Pennsylvania School of Medicine, Pennsylvania 19104-4318 Received May 16, 2006

Translocations and other aberrations involving the MLL (mixed lineage leukemia) gene result in aggressive forms of leukemias. Heterogeneity in partner genes, in chromosomal breakpoints, in MLL itself, and in the different partner genes results in heterogeneous fusion transcripts that can be alternatively spliced, which complicates deciphering a unifying mechanism of leukemogenesis. However, recent microarray studies completed with clinical leukemia specimens have uncovered several distinct mRNA signatures within MLL leukemia that differ from other types of leukemia. A global proteomics strategy using MV4-11 and RS4:11 cells in culture was employed to investigate possible protein signatures common to different MLL leukemias and to identify disease biomarkers and protein targets for pharmacological intervention. Initial proteomics screening experiments with two-dimensional differential in-gel electrophoresis revealed heat shock protein 90 alpha (HSP90R) as a potential target for pharmacological inhibition and nucleoside diphosphate kinase (nm23) as a biomarker for measuring treatment efficacy. Using a modified stable isotope labeling of amino acids in cell culture (SILAC) approach, coupled with two-dimensional liquid chromatography tandem mass spectrometry (2D-LCMS/MS), changes in abundance for over 500 proteins were measured. In addition, decreased expression of the novel biomarker nm23 was observed during HSP90 inhibition with 17-allylamino-17-demethoxygeldanamycin (17-AAG) in the MV4-11 cell line. The present study validates the use of a global proteomics strategy to uncover novel biomarkers and pharmacological targets for leukemias with MLL translocations. Additionally, several proteins were found to be expressed in concordance with microarray studies of mRNA expression in specimens from patients showing the value in comparing mRNA transcript and proteomic profiles. This work represents one of the most comprehensive proteomics screens of MLL leukemias that have been conducted to date. Keywords: MLL • chromosomal breakpoints • leukemia • proteomics • MS/MS • differential in-gel electrophoresis • heat shock protein 90 • SILAC • nucleoside diphosphate kinase

Introduction Chromosomal translocations of the mixed lineage leukemia (MLL)1 gene on chromosome band 11q23 are present in most cases of acute lymphocytic leukemia (ALL) and myelomonocytic and monoblastic variants of acute myeloid leukemia (AML) in infants and young children.1,2 They are also found in secondary leukemias following chemotherapeutic treatment with topoisomerase poisons.3 Chromosomal translocations of the MLL gene are also observed in a smaller proportion of de novo acute leukemias in adults.1 Despite general advances toward the cure of childhood leukemias, survival rates among patients with leukemias characterized by MLL translocations * To whom correspondence should be addressed. Center for Cancer Pharmacology, University of Pennsylvania School of Medicine, 854 BRB II/ III, 421 Curie Boulevard, Philadelphia PA 19104-6160. Tel: 215-573-9880. Fax: 215-573-9889. Email: [email protected]. † Center for Cancer Pharmacology. ‡ Division of Oncology. 10.1021/pr060235v CCC: $33.50

 2006 American Chemical Society

are poor, and the intensive therapies are toxic.4 These leukemias are generally clinically aggressive and difficult to treat; hence, there is a need for better leukemia biomarkers that address the mechanism of MLL leukemogenesis and new pharmacological targets for drug development. The complexity of the native MLL oncoprotein and the heterogeneity of MLL fusion proteins arising from the translocations make deciphering the mechanism of leukemogenesis and the functional role of MLL fusion proteins in leukemias with MLL translocations a challenge. Native MLL is a multidomain protein that undergoes proteolytic cleavage into N- and C-terminal fragments that reassociate in a multi-protein complex that regulates transcription.5 Functional domains of the MLL protein include N-terminal A-T hooks that mediate DNA binding, two speckled nuclear localization (SNL) domains, a DNA methyltransferase homology domain that mediates transcriptional repression, plant homeodomain (PHD) fingers that mediate protein-protein interactions, a transactivation doJournal of Proteome Research 2006, 5, 2743-2753

2743

Published on Web 08/18/2006

research articles main, and a SET domain at the C terminus with histone H3 lysine 4 methyltransferase activity.6-8 The SNL, PHD, and SET domains have regional amino acid similarity to Drosophila trx, which maintains Hox gene expression. In mammals, MLL maintains Hox gene expression early during skeletal, craniofacial, and neural development and hematopoiesis.1,9 MLL translocations involve many partner genes that encode different types of proteins, including nuclear transcription factors, proteins involved in transcriptional regulation, cytoplasmic proteins, and proteins expressed in various cellular compartments. Genomic breakpoint junction sequences or MLL chimeric transcripts involving more than 40 partner genes, including MLL itself, have been identified.1 Translocation of the MLL gene located on chromosome 11q23 to the AF4 gene located on chromosome 4q21 plays a pivotal role in leukemogenesis in infancy.10 The AF4 gene product is also a transcription factor.11 Heterogeneity in the MLL-AF4 translocation is the outcome of numerous events including multiple translocation breakpoint locations within the MLL and AF4 breakpoint cluster regions, the variable expression of the fusion transcripts from the newly derived chromosome 11, {der(11)}, and/or the derivative, der(4) chromosome, alternative splicing of the mRNAs, and post-translational modifications in the expressed fusion oncoproteins. Murine models have demonstrated that the der(11) protein products are leukemogenic.6 The der(11) protein products do not undergo proteolytic cleavage, and the MLL, PHD, transactivation, and SET domains are replaced by the C-terminus of the partner protein. It is conceivable that a protein is transcribed from the other derivative chromosome that results from the translocation. If this protein is actually expressed, any contribution it makes to the genesis or maintenance of leukemia remains to be determined. Recently, DNA microarray technology has been used to identify mRNA expression patterns or signatures in leukemias with MLL translocations. Armstrong et al. found highly divergent levels of expression of FLT3 tyrosine kinase receptor mRNA between cases of ALL and AML with and without MLL translocations, as well as differential expression of selected HOX genes and the MEIS-1 HOX cofactor gene.12 In addition, Yeoh et al. explored the utility of gene expression profiling in a larger cohort of pediatric patients with ALL for subgroup determination and predicting treatment failure.13 This latter study identified upregulated mRNAs that distinguished the major cytogenetic and phenotypic subsets of pediatric ALL and upregulated mRNAs that predicted relapse. Consistent with a role in the pathobiology of leukemias with MLL rearrangements, HOXA9 and MEIS-1 mRNAs were overexpressed exclusively in this subset.14 No single gene proved discriminatory between MLL and other leukemias. Most recently, Kohlmann et al. described a distinct mRNA transcript profile expression signature in all MLL-translocated leukemias.14 However, the different subtypes of MLL translocations did not cluster together but rather clustered binomially by lineage, either as ALL or AML. The objective of the present study was to apply global proteomics analyses of leukemias with MLL translocations to search for leukemia biomarkers and potential pharmacologic targets. Two cell lines harboring the t(4;11)(q21;q23) translocation fusing MLL to AF-4 were utilized in these studies because the t(4;11) is the most common MLL translocation. These cultured cells contain the same MLL-AF4 fusion protein but differ phenotypically. These cell lines were compared to CD34+ cells, an early hematopoietic progenitor control, in a proteomics-based experiment. It was hypothesized that a signature 2744

Journal of Proteome Research • Vol. 5, No. 10, 2006

Yocum et al.

could be teased out common to both phenotypes that is directly attributable to the MLL-AF4 fusion protein. This signature might be indicative of the mechanism of oncogenesis and contain possible pharmacological targets and biomarkers. Initial proteomics screening experiments were conducted with 2D-DIGE and MALDI-MS/MS using the MV4-11 and RS4: 11 cell lines. A modified stable isotope labeling by amino acids in cell culture (SILAC) method, coupled with two-dimensional liquid chromatography-tandem mass spectrometry (2D-LCMS/MS) was then used to quantify protein expression during inhibition of a pharmacological target identified in the initial screening.

Experimental Procedures Sample Preparation. MV4-11 and RS4:11 cells were obtained from the American Type Tissue Culture Collection (ATCC). CD34+ selected bone marrow cells were kindly provided from the Programs of Excellence in Gene Therapy, Hematopoeitic Cell Processing Core at Fred Hutchinson Cancer Research Center. The MV4-11 cell line was established from the blast cells of a 10-year-old male with biphenotypic B-AML. The RS4: 11 cell line was established from the bone marrow of a 32year-old female with ALL. Cells were collected, washed three times in phosphate buffered saline, and resuspended in detergent lysis buffer (6 M urea, 4% CHAPS, and 40 mM Tris) for whole cell extract experiments or in NE-PER (Pierce Biotechnology Inc., Rockford, IL), nuclear, and cytoplasmic extraction reagents for fractionation experiments. Protease Inhibitor Cocktail - EDTA free (Pierce Biotechnology Inc., Rockford, IL) was added to the lysis buffer and to NE-PER. All samples, cytoplasmic and nuclear fractions as well as whole cell lysates, were delipidated and desalted by methanol/ chloroform protein precipitation. The protein precipitate was then resuspended in the same detergent lysis buffer and incubated on ice for 1 h to ensure solubilization. The solution absorbance at 260 and 280 nm was then measured in triplicate, averaged, and the protein quantification determined by substitution of appropriate parameters in the following equation: concentration in mg/mL ) (1.55 × A280 nm) - 0.76 × A260 nm.15 Sample Labeling and 2D-DIGE. Protein sample preparation, labeling, and 2D-DIGE were performed as described previously.16 Whole protein extracts from each cell line, MV4-11 and RS4:11, were compared independently with proteins from control CD34+ cells (Figure 1). Cy-2 was used to label CD34+ proteins, Cy-3 was used to label MV4-11 or RS4:11 proteins, and Cy-5 was used to label the reference sample (MV4-11 + RS4:11 + CD34+ proteins) (Figure 1a). Experimental gels were run in triplicate and then replicated 12 days later. The pooled internal standard methodology of Alban et al.17 was used in this study. This standard was composed of equal amounts of protein from MV4-11, RS4:11, and CD34+ cells. A dye switch for each set of triplicate gels was performed as recommended by the manufacturer to normalize for inherent labeling differences. Additionally, an unlabeled 1 mg protein sample from MV4-11 and RS4:11 and respective subcellular fractions were collected for separation by conventional 2D-polyacrylamide gel electrophoresis (PAGE). The gels containing unlabeled protein served for spot matching to CyDye-labeled protein gels and for protein identification by MS. Imaging and 2D-DIGE Analysis. The Cy2, Cy3, and Cy5 stained spots on each gel were individually imaged using mutually exclusive excitation/emission wavelengths of 488/520

Biomarkers and Pharmacological Targets in Leukemia

research articles

Figure 1. (a) Schematic representation of experimental design for comparisons of differential protein expression in leukemia cell lines compared to control CD34+ cells using 2D-DIGE analysis. (b) Composite 2D-DIGE image of Cy-3 labeled MV4-11, Cy-2 labeled CD34+ whole cell protein extracts, and Cy-5 labeled reference sample (MV4-11 + RS4:11 + CD34+ whole cell protein extracts). Gels were imaged using a Typhoon 9410 Variable Mode Imager with the following fluorescence wavelength settings: Cy2, 488/520 nm; Cy 3, 532/580 nm; Cy5, 633/670 nm.

nm for Cy2, 532/580 nm for Cy3, and 633/670 nm for Cy5 using a Typhoon 9410 Variable Mode Imager (GE Healthcare Amersham Biosciences) (Figure 1b). Each of the 6 replicate gels contained 3 images for a total of 18 images for quantitative comparison. The Colloidal Blue stained gels were imaged using a wavelength of 633 nm. DeCyder (GE Healthcare - Amersham Biosciences) software was used for pairwise comparison of protein abundance changes with MV4-11 and RS4:11 compared to CD34+ cells. The Differential In-Gel Analysis (DIA) module of DeCyder was used for pairwise comparisons of MV4-11 and RS4:11 to CD34+ cells normalized to the internal standard pool. For each pairwise DIA comparison, e.g., MV4-11 vs CD34+ cells, the entire signal from each CyDye channel was normalized prior to the co-detection of protein-spot boundaries and the calculation of the volume ratio for each protein-spot pair. The Biological Variation Analysis (BVA) module of DeCyder was then used to simultaneously match all 20 protein-spot maps. Eighteen gel images are derived from the three replicates run twice and two images from the colloidal blue-stained picking gels. The BVA software compared protein abundance ratios from the cell line vs CD34+ cells. Average abundance changes were determined using paired Student’s t-test and p-values for the variance between ratios for each protein-pair across the two experiments, including three replicate gels for each experiment, and three samples on each gel, n ) 18. Fold abundance changes are reported, whereby a fold increase was calculated directly from the volume ratio, and a fold decrease equals 1/volume ratio. Significance levels were determined to be above two standard deviations of the mean volume ratios in the 99th percentile confidence using the DIA module. In-Gel Digestion, MS/MS, and Database Analysis. Proteins of interest, those with 1.5-fold or greater increased abundance for whole cell lysates, and 2.0-fold or greater increased or decreased abundance for cytoplasmic and nuclear fractions, were robotically excised into a 96-well plate format using an Ettan Spot Picker (GE Healthcare - Amersham Biosciences). Gel plugs were washed, digested with sequencing grade modified trypsin (Promega, Madison, WI), and prepared for MALDITOF/TOF/MS (4700 Proteomics Analyzer, Applied Biosystems, Framingham, MA) analysis as described previously.16 Trypsin autolytic peptides (m/z - 1045.55, 1940.94, 2211.10, and 2225.11) were used to internally calibrate each spectrum to a mass accuracy within 20 ppm. Data-dependent tandem MALDI-

TOF/TOF MS was performed on the top 12 abundant peptide ions (excluding ions from trypsin autolysis peptides) to generate amino acid sequence information for additional confirmation to the peptide mass fingerprint-derived identification. GPS Explorer (Applied Biosystems, Foster City, CA) software was used to perform a combined MS peptide fingerprint and MS/ MS peptide-sequencing search of the refseq nonredundant protein database. These searches allowed for carbamidomethylation of cysteine, partial oxidation of methionine residues, and one missed trypsin cleavage. Highest confidence identifications passed three criteria: first, a statistically significant search score >65 for the protein was required; second, the molecular weight and pI of the protein were consistent with the gel region from which the protein was excised; and third, that ions from the protein tryptic fragments accounted for the majority (>70%) of the ions present in the peptide mass fingerprint. Western Blotting. Proteins from each cell line (15 µg of each) were diluted 25-fold in LDS sample buffer (Invitrogen, Carlsbad, CA) and incubated at 60 °C for 10 min. Protein samples were separated on 4-20% Tris-Glycine gels (NuPAGE Novex gels, Invitrogen) and transferred to a nitrocellulose membrane (Invitrogen). SeeBlue Plus2 (Invitrogen) protein standard was used to estimate molecular weights. The following primary antibodies and dilutions were utilized: rabbit IgG HSP90 (which recognizes both R and β isoforms) (1:500), goat IgG PGK-1 (1: 500), rabbit IgG nm23 (1:1000), and goat IgG β-actin (1:200) were obtained from Santa Cruz Biotechnologies (Santa Cruz, CA). Rabbit IgG Laminin R (1:1000) and rabbit IgG Stathmin (1:500) were obtained from AbCam (Cambridge, MA) and Cell Signaling Technology (Danvers, MA), respectively. In vitro Inhibition Studies with 17-Allylamino-17-Demthoxygeldanamycin (17-AAG). Cells were seeded in fresh media (99 µL) at 5 × 104 cells per well in a 96-well plate for in vitro cellular assays. 17-AAG (Stressgen Bioreagents, Victoria, BC, Canada) was stored at -20 °C until used. It was dissolved in DMSO to give appropriate concentrations for a 1 µL dosage. Aliquots of treated cells were assayed for viability (CellTiter-Blue), cell proliferation (CellTiter 96Aqueous), and apoptosis (Apo-ONE) (Promega, Madison, WI) by colorimetry and fluorescence to determine approximate IC50 of 24 h treatment with 17-AAG according to the manufacturer’s instructions. Journal of Proteome Research • Vol. 5, No. 10, 2006 2745

research articles

Yocum et al.

Figure 2. (a) Image of 2 D-DIGE of MV4-11whole cellular protein extract stained with colloidal blue. (b) Image of 2 D-DIGE of RS4:11 whole cellular protein extract stained with colloidal blue. (c) Spots identified were found to be overexpressed in both RS4:11 and MV4-11 whole cellular protein extracts compared to healthy CD 34+ bone marrow.

SILAC Labeling and 2D-LC-MS/MS Analysis. MV4-11 cells were grown as indicated by ATCC, except with the substitution of two isotopic-labeled amino acids, L-Leucine (13C6, 98% 15N2, 98%) and L-Lysine:2HCl (13C6, 98% 15N2, 98%) (Cambridge Isotope Laboratories, Andover, MA). They were grown in dialyzed fetal bovine serum (HyClone, Logan, UT) for seven passages, harvested, and cryogenically preserved until further use as an internally labeled proteome standard. Unlabeled MV4-11 cells were treated with increasing doses of 17-AAG for 24 h. Cells were lysed in modified denaturing NP-40 buffer containing 150 mM sodium chloride, 1% v/v NP-40, 50 mM Tris pH 8.0, and 0.2% sodium dodecyl sulfate at 4 °C for 4 h. The soluble protein fraction was cleared by centrifugation at 17 000 × g for 10 min at 4 °C, and the supernatant was stored in an Eppendorf tube. Stable isotope labeled cell proteins (500 µg) and unlabeled 17-AAG treated cell proteins (500 µg) were pooled and prepared for digestion and off-line strong cation exchange (SCX) fractionation.18 The protein precipitate was resuspended in 0.2% (w/v) RapiGestSF (Waters Co. Milford, MA) in 50 mM NH4HCO3 and then placed in a boiling water bath for 5 min. Off-line SCX fractionation followed by on-line reversed-phase LC-MS/MS analysis was performed as described previously.18 Database Searching and Statistical Analysis. Raw MS/MS data were submitted to Bioworks Browser (Thermo Finnigan, San Jose, CA) and batch searched through TurboSEQUEST against an indexed human RefSeq database (version updated 6/05). The database was indexed using strict trypsin cleavage rules with a maximum of two missed cleavage sites and differential modifications of methionine oxidation, carboxyamidomethlyation on cysteine, 7 Da mass addition on leucine, and an 8 Da mass addition on lysine. The SEQUEST output files were analyzed and validated using Trans-Proteomic Pipeline (TPP) software by the Institute of Systems Biology 2746

Journal of Proteome Research • Vol. 5, No. 10, 2006

(ISB). ASAPratio, a module of the TPP software, was used to calculate the relative abundances of peptides and proteins as well as the corresponding confidence intervals.

Results Identification of Differentially Expressed Proteins in Cell Lines with MLL-AF4 Translocations. In comparing whole cellular proteins from MV4-11 and CD34+ cells by 2D-DIGE, 71 spot features were calculated to be g1.5-fold overexpressed in MV4-11 compared with CD34+ cells, including 39 that were >2.0-fold overexpressed on all 18 gel images with a p < 0.01 (Figure 2a). From these 71 spot features, 29 unique proteins were discerned out of the 41 confident identifications by MS with the confidence limits shown in Supplementary Table 1. Likewise, whole cellular proteins from RS4:11 were compared with whole cellular proteins from control CD34+ cells. 93 spot features were calculated to be g1.5-fold overexpressed in RS4: 11 compared with CD34+ cells, including 80 spot features that were g2.0-fold overexpressed on all 18 gel images with a p < 0.01 (Figure 2b). When the 93 spots were analyzed by MS, 74 spot features resulted in clear identifications with 41 unique proteins identified with the confidence limits shown in Supplementary Table 2. There were 11 proteins with increased expression in both leukemia cell lines when compared with CD34+ cells (Figure 2c). HSP90R was also identified as being overexpressed in both leukemia cell lines when compared with control CD34+ cells. It was confidently identified by 8 HSP90Rderived peptides, 6 of which were sequenced by MS/MS with a mass accuracy 50. Four trypsin autolysis products ions are labeled vertically, m/z 1045.55, 1940.94, 2211.10, and 2225.11, were used as internal mass calibration. (b) MALDI-MS/MS spectra indicating the amino acid sequence of peptide m/z 1348.6532 from HSP90R Figure 3a spectrum. An almost complete series of y and b product ions was observed.

Figure 4. GO chart by functional category of all proteins identified by 2D-DIGE and found to be overexpressed in MV4-11 and RS4:11 cells when compared with control CD34+ bone marrow cells.

Identification of Differentially Expressed Proteins in Subcellular Fractions Using 2D-DIGE. Expression of nuclear proteins from MV4-11 cells was compared with expression of nuclear proteins from CD34+ cells. There were 253 spot features, which were increased or decreased g2.0-fold, of which 141 spots were identified confidently as 84 unique proteins. In the nuclear protein comparison of RS4:11 and CD34+ cells, 260 spot features were found to be g2.0-fold increased or decreased, of which 161 spot features were identified confidently as 72 unique proteins. In the cytoplasmic fraction, 308 spots

were g2.0-fold increased or decreased in the MV4-11 cells when compared with theCD34+ cells; 101 proteins were identified by MS/MS and 64 were found to be unique. A similar comparison of the cytoplasmic proteins from RS4:11 and CD34+ cells revealed that 220 spot features were increased or decreased g2.0-fold from which 72 unique proteins were identified by MS/MS. There were 19 proteins differentially expressed in the nuclear fractions of both cell lines, and 11 proteins differentially expressed in the cytoplasmic fraction were common to both Journal of Proteome Research • Vol. 5, No. 10, 2006 2747

research articles

Yocum et al.

Table 1. Proteins Expressed in Subcellular Fractions of Both RS4:11 and MV4-11 Cells Compared with CD34+ Bone Marrow Cells (a) cytoplasmic fractions

GI accession numberb

avg. ratio RS

avg. ratio MV

6-phosphogluconolactonase annexin A2 COP9 signalosome subunit 4 eukaryotic translation initiation factor 4 growth factor receptor-bound protein 2 HSP 70kDa protein 5 HSP 70kDA protein 8 laminin receptor 1 pyruvate kinase 3 phosphoglycerate mutase 1 stress-induced phosphoprotein 1

gi|6912586|ref|NP_036220.1| gi|4757756|ref|NP_004030.1| gi|38373690|ref|NP_057213.2| gi|4503535|ref|NP_001959.1| gi|4504111|ref|NP_002077.1| gi|16507237|ref|NP_005338.1| gi|5729877|ref|INP_006588.1| gi|386857|gb|AAA36165.1| gi|33286418|ref|NP_002645.3| gi|4505753|ref|NP_002620.1| gi|5803181|ref|NP_00681 0.11

10.78 2.18 7.27 3.26 4.41 4.02 7.22 -2.5 -2.6 11.64 -2.77

-2.8 2.44 4.27 6.38 -3.06 13.4 9.46 -2.4 -2.3 2.02 3.49

(b) nuclear fractions

GI accession numberb

avg. ratio RS

avg. ratio MV

actin, gamma 1 propeptide beta 5-tubulin capping protein cleavage simulation factor, subunit 2 DEAD box polypeptide 5 DEAD box polypeptide 17 isoform p82 eukaryotic translation elongation factor 2 eukaryotic translation elongation factor 5A far upstream element-binding protein glutamate dehydrogenase 1 glutathione transferase HSP 70kDa protein 1A heterogeneous nuclear ribonucleoprotein A1 heterogeneous nuclear ribonucleoprotein Ka heterogeneous nuclear ribonucleoprotein Mb KH-type splicing regulatory protein nucleoside-diphosphate kinase 1b splicing factor proline/glutamine rich stathmin 1

gi|4501887|ref|NP_001605.1| gi|29788785|ref|NP_821133.1| gi|63252913|ref|NP_001738.2| gi|41327732|ref|NP_958439.1| gi|4758138|ref|NP_004387.1| gi|38201710|ref|NP_006377.2| gi|4503483|ref|NP_001952.1| gi|11136628|ref|NP_066944.1| gi|17402900|ref|NP_003893.2| gi|4885281|ref|NP_005262.1| gi|4504183|ref|NP_000843.1| gi|5123454|ref|NP_005336.2| gi|14043070|ref|NP_112420.1| gi|14165439|ref|NP_002131.2| gi|5803036|ref|NP_006796.1| gi|4504865|ref|NP_003676.1| gi|4557797|ref|NP_000260.1| gi|4826998|ref|NP_005057.1| gi|44890052|ref|NP_981946.1|

5.23 -2.82 -3.96 5.67 -2.59 -2.67 -2.94 2.74 -2.15 2.56 23.78 4.45 -2.02 -2.23 -3.03 -3.09 2.41 6.74 -2.6

2.82 -2.55 -3.22 -2.23 -2.64 -2.75 -2.92 -4.27 -2.15 2.49 -3.53 3.95 -2.07 -2.2 -2.89 -2.95 -3.5 -2.61 -4.55

(c)both cytoplasmic & nuclear fractions

GI accession numberb

avg. ratio RS

avg. ratio MV

enolase 1 glucose regulated protein HSP 70kDa protein 9B heterogeneous nuclear ribonucleoprotein H1 TCP-1

gi|4503571|ref|NP_001419.1| gi|20986531|ref|NP_620407.1| gi|24234688|ref|NP_004125.3| gi|5031753|ref|NP_005511.1| gi|57863257|ref|NP_110379.2|

-2.32 2.52 4.09 -2.01 5.86

-3.43 5.74 7.05 -2.4 3.81

a Immediately apparent is the localization of several proteins found in both compartments (c) complicating quantification. b Individual molecular weights and pIs can be readily obtained from the GI accession numbers.

cell lines (Tables 1a and 1b, respectively). Causing some concern, however, were 5 proteins differentially expressed in both the cytoplasmic and nuclear fractions of both cell lines (Table 1c), highlighting the complexity of this protein quantitation methodology during subcellular fractionation. Expression of HSP90 and nm23 in Cell Lines with MLLAF4 Translocations. To confirm the whole cellular protein 2DDIGE findings, Western blot analysis was completed on PGK1, HSP90, nm23, laminin receptor, and stathmin I (Supplementary Figure 1). β-actin was used as a loading control. PGK1, HSP90, and nm23 were overexpressed in the leukemia cells when compared with the CD34+ cells, confirming the 2D-DIGE results. Laminin receptor and stathmin I proteins were also found by Western blot analysis to be overexpressed in leukemia cells when compared with CD34+ cells. This contrasted with the subcellular fraction 2D-DIGE results showing both laminin receptor and stathmin I to be underexpressed when compared with CD34+ cells. Inhibition of HSP90 in MV4-11 Cells by 17-AAG. Treatment of the MV4-11 cells for 24 h resulted in increased fluorescence emission with increased dosages of 17-AAG indicating increased caspase 3/7 activity, which is a marker of apoptosis. (Supplementary Figure 2a). A similar cytotoxic trend was seen 2748

Journal of Proteome Research • Vol. 5, No. 10, 2006

with the fluorometric indicator dye resazurin for estimating the cell viability (Supplementary Figure 2b) and with Owen’s reagent for estimating cell proliferation (Supplementary Figure 2c). These data are consistent with a previous estimate of a 40 nM as the IC50 at 24 h for cell proliferation, cell viability, and apoptosis in the MV4-11 cell line.19 SILAC Labeling Efficiency in MV4-11 Cells. Two independent 2D LC/MS/MS experiments were completed and independently analyzed to validate labeling efficiency. First, the nonlabeled sample was analyzed. No peptides were identified as containing an isotopically labeled amino acid. Next, the labeled sample was analyzed. Only 24 out of 6 594 peptides belonging to 8 common housekeeping proteins were identified that were not completely labeled with 13C and 15N at lysine and leucine residues. Each of the other 6 570 peptides contained