iTRAQ-Multidimensional Liquid Chromatography and Tandem Mass

Dec 10, 2008 - Pathway analysis of differentially expressed proteins unraveled novel networks linking inflammation and development of epithelial dyspl...
2 downloads 0 Views 5MB Size
Download PDF
iTRAQ-Multidimensional Liquid Chromatography and Tandem Mass Spectrometry-Based Identification of Potential Biomarkers of Oral Epithelial Dysplasia and Novel Networks between Inflammation and Premalignancy Ranju Ralhan,*,†,§,⊥ Leroi V. DeSouza,†,‡,§ Ajay Matta,†,§,⊥ Satyendra Chandra Tripathi,⊥ Shaun Ghanny,‡,§,| Siddhartha DattaGupta,# Alok Thakar,∇ Shyam S. Chauhan,⊥ and K. W. Michael Siu*,†,‡,§ Departments of Chemistry and Biology, Centre for Research in Mass Spectrometry and Department of Mathematics and Statistics, York University, 4700 Keele Street, Toronto, Ontario, Canada M3J 1P3, and Departments of Biochemistry, Pathology, and Otorhinolaryngology, All India Institute of Medical Sciences, New Delhi 110029, India Received July 7, 2008

Chronic exposure of the oral mucosa to carcinogens in tobacco is linked to inflammation and development of oral premalignant lesions (OPLs) with high risk of progression to cancer; there is currently no clinical methodology to identify high-risk lesions. We hypothesized that identification of differentially expressed proteins in OPLs in relation to normal oral tissues using proteomic approach will reveal changes in multiple cellular pathways and aid in biomarker discovery. Isobaric mass tags (iTRAQ)-labeled oral dysplasias and normal tissues were compared against pooled normal control by online liquid chromatography and tandem mass spectrometry. Verification of biomarkers was carried out in an independent set of samples by immunohistochemistry, immunoblotting, and RT-PCR. We identified 459 nonredundant proteins in OPLs, including structural proteins, signaling components, enzymes, receptors, transcription factors, and chaperones. A panel of three best-performing biomarkers identified by iTRAQ analysis and verified by immunohistochemistrysstratifin (SFN), YWHAZ, and hnRNPKsachieved a sensitivity of 0.83, 0.91, specificity of 0.74, 0.95, and predictive value of 0.87 and 0.96, respectively, in discriminating dysplasias from normal tissues, thereby confirming their utility as potential OPL biomarkers. Pathway analysis revealed direct interactions between all the three biomarkers and their involvement in two major networks involved in inflammation, signaling, proliferation, regulation of gene expression, and cancer. In conclusion, our work on determining the OPL proteome unraveled novel networks linking inflammation and development of epithelial dysplasia and their key regulatory proteins may serve as novel chemopreventive/therapeutic targets for early intervention. Additionally, we identified and verified a panel of OPL biomarkers that hold promise for large-scale validation for ultimate clinical use. Keywords: Oral premalignant Lesions • Dysplasia • Head and neck/oral squamous cell carcinoma • iTRAQ labeling • Multidimentional Liquid Chromatography • Tandem Mass Spectrometry • Potential Biomarkers •

Introduction The link between inflammation and cancer are currently under intense investigation. By comparison, the association * To whom correspondence should be addressed at Centre for Research in Mass Spectrometry, Department of Chemistry, York University, 4700 Keele Street, Toronto, Ontario, Canada M3J 1P3. For K.W.M.S.: tel, (416)650-8021; fax, (416)736-5936; e-mail, [email protected]. For R.R.: tel, (416)736-2100 ext. 40048; fax, (416)736-5936; e-mail, [email protected]. † Department of Chemistry, York University. § Centre for Research in Mass Spectrometry, York University. ⊥ Department of Biochemistry, All India Institute of Medical Sciences. ‡ Department of Biology, York University. | Department of Mathematics and Statistics, York University. # Department of Pathology, All India Institute of Medical Sciences. ∇ Department of Otorhinolaryngology, All India Institute of Medical Sciences.

300 Journal of Proteome Research 2009, 8, 300–309 Published on Web 12/10/2008

between inflammation, due to chronic exposure of oral mucosa to tobacco carcinogens, and the development of oral premalignant lesions (OPLs) remains ill-defined. The biology of oral premalignancy is at present not well-understood. The question as to why only some OPLs progress to cancer while others do not remains an enigma. It is increasingly evident that the identification of high-risk OPLs before manifestation of cancer is of utmost importance for effective intervention. Leukoplakia, the most common heterogeneous oral precancerous lesion, with a global prevalence of 2-3% in the general population, is associated with increased risk of cancer.1-3 The high morbidity and mortality associated with oral cancer make leukoplakia a serious global public-health problem. Herein, leukoplakia with biopsy-proven squamous dysplasia is defined 10.1021/pr800501j CCC: $40.75

 2009 American Chemical Society

research articles

Oral Dysplasia Networks and Biomarker Discovery 4

as OPL. The presence of dysplastic areas in the oral epithelium is associated with up to 36% incidence of subsequent oral squamous cell carcinoma (OSCC), though nondysplastic lesions may occasionally progress to cancer.5-7 Clinical, histological, and molecular markers may contribute toward assessing the risk of a patient in developing cancer; however, there is no clinically proven predictor of malignant transformation for dysplastic and nondysplastic leukoplakias. The use of fluorescence visualization devices for detecting clinically occult OPLs and field alterations in tumor margins of OSCC patients have yielded contradictory findings.8,9 Chronic exposure of the oral mucosa to carcinogens/growth promoters in tobacco or alcohol leads to accumulation of epigenetic, genetic, and metabolic alterations, often resulting in development of premalignant lesions that acquire increasingly abnormal phenotypes.10 Deletions at critical chromosomal loci in OPLs are associated with an increased risk of cancer development.11 Recently, podoplanin in conjunction with histology was reported as a marker for oral cancer risk assessment in OPL patients, but this diagnostic modality needs validation in larger studies.12 Proteomic analyses of head-and-neck/oral squamous cell carcinomas (HNOSCCs) show that expression patterns of proteins may have some predictive power for clinical outcome and personalized risk assessment.13-15 However, no study has been devoted to identifying the OPL proteome. There is a need for documentation and understanding of global expression changes in OPLs to secure a holistic understanding of the perturbations in cellular processes in these lesions. Herein we use iTRAQ-labeling and LC-MS/MS analysis to discover, identify, and differentially quantify proteins in tissue homogenates of OPLs relative to normal tissues. Our study is important and timely as we seek to identify differentially expressed proteins in OPLs relative to histologically normal tissues. Recent studies based on differential labeling with mass-tagging reagents, such as the isotope-coded affinity tag, ICAT,16 or isobaric masstagging reagent, iTRAQ (Applied Biosystems, Foster City, CA), followed by multidimensional liquid chromatography (LC) and tandem mass spectrometry (MS/MS) resulted in discovery of potential cancer markers (PCMs) for endometrial cancer and head-and-neck cancer.17-21 The tissue microenvironment plays an important role in cancer development.22,23 Examination of protein expressions in tissue homogenates, which comprise different cell types, will take into account the contributions of dynamic changes taking place in the oral-tissue microenvironment and may provide new avenues for the early detection of high-risk OPLs. The multiplexing ability afforded by the iTRAQ reagent, available in four mass tags, permits simultaneous comparison of four samples, and provides us with a means to analyze individual OPLs and histological normal oral tissues, while simultaneously comparing them against a pool of histological normal oral tissues that serves as a common reference. Identification of proteins differentially expressed in OPLs, in comparison with normal oral tissues, helps in revealing changes in multiple cellular pathways and unraveling novel molecular networks that may link inflammation and cancer in the early stages of oral premalignancy development, prior to malignant transformation, and thus constitute the basis for development of OPL biomarkers.

Materials and Methods Samples and Reagents. OPL and normal tissues (controls) were retrieved from an in-house research oral-tissue bank, with approval from the Human Ethics Committee of the All India Institute of Medical Sciences (AIIMS), New Delhi, India. Biopsies/excised tissue specimens of leukoplakia were collected and banked with informed consent from patients undergoing treatment in the Otorhinolaryngology Outpatient Department of AIIMS. Normal tissues were collected with informed consent from patients attending the Dental Outpatient Department, AIIMS, for tooth extraction. After excision, tissues were flashfrozen in liquid nitrogen within 20-30 min of devitalizing, and stored at -80 °C; one piece of tissue was collected in 10% formalin for histopathological analysis. Relevant clinical and pathological data were recorded in a predesigned form as described21 and are summarized in Supplementary Table S1 in Supporting Information. The histologic diagnoses (dysplasia for OPLs and histological normal oral epithelium for controls) were rendered using microscopic examination of hematoxylin-and-eosin-stained frozen sections of each research tissue block. The tissue from the mirror-face was then homogenized in 0.5 mL of PBS with a cocktail of protease inhibitors (1 mM AEBSF, 10 µM leupeptin, 1 µg/mL aprotinin, and 1 µM pepstatin) and stored at -80 °C. Samples were clarified by centrifugation and protein concentrations were estimated before trypsin treatment.21 The experiments were performed in three sets of four samples; a pool of noncancerous oral-tissue homogenates was used as a control in each set of experiments. Equal amounts of total proteins from lysates of six histological normal samples were pooled to generate a common reference “control sample” against which all the OPL samples were compared. This pool was used as the reference sample in all the iTRAQ sets analyzed, eliminating the need for introducing a set of iTRAQ analyses for normalizing the different controls used in each individual OPL versus normal comparison.21 Trypsin digestion and labeling were performed according to the manufacturer’s protocol. Each analytical set comprised 4 × 100 µg of each sample labeled as follows: control (normal pool) was labeled with one iTRAQ tag; two OPL samples were labeled with two other iTRAQ tags; and an individual histological normal tissue sample was labeled with the fourth iTRAQ tag. Thus, a total of six OPLs and three histological normal samples were compared to the control sample in three iTRAQ sets. The order in which the samples were labeled within each of these three sets was randomized to minimize any potential systematic error and bias. The iTRAQ analyses of these samples were performed with one run of online reverse phase LC-MS/MS for preliminary examinations, and three replicate runs per set of online two-dimensional LCMS/MS analyses, confirming the reproducibility in the experiment.21 The samples were analyzed on a Q-STAR Pulsar hybrid quadrupole/time-of-flight tandem mass spectrometer (Applied Biosystems/MDS SCIEX, Foster City, CA) in informationdependent acquisition (IDA) mode with the scan cycles set up to perform a 1-s MS scan followed by five MS/MS scans of the five most abundant ions for 2 s each. The method was also set up to select the least-abundant ions in the MS scan that are nearest to a threshold of 10 counts on every fourth scan. Data acquisition was performed without any repetitions and with a dynamic exclusion of 30 s. Relative protein abundances were determined using the 114.1, 115.1, 116.1 and 117.1 Th reporter Journal of Proteome Research • Vol. 8, No. 1, 2009 301

research articles

Ralhan et al. a

Table 1. Average iTRAQ Ratios for OPLs and Histologically Normal Control Oral Tissue Samples accession no.

protein name

D1

D2

D3

D4

D5

D6

N1

N2

N3

N4

N5

N6

spt|P31947 spt|P63104 spt|P61978 spt|P06454 spt|P14625 trm|O15256 spt|Q16778 spt|P14625 spt|P06748 spt|P04080 spt|P29966 trm|Q6NSB4 spt|P01009 spt|P32119 spt|Q01469 gb|AAH25314.1 spt|P31949 trm|Q8N5F4 trm|Q12771 prf|0904262A spt|P14618 sptIP09651 sptIP04792 spt|P23528 emb|CAA25833.1 spt|P22392 spt|P13639 spt|P31151 spt|P30086 spt|P27482 spt|P06396 spt|P60709

Stratifin YWHAZ hnRNPK PTHA HSP90B1 Parathymosin Histone H2B.1 GRP 94 Nucleophosmin 1 Cystatin B MARCKS DLC1 Alpha 1 Anti-Trypsin Precursor Peroxiredoxin 2 FABP5 IGHG1 protein Calgizzarin IGL 2 p37AUF1, hnRNPD SOD2 PKM2 hnRNPA1 Hsp27 Cofilin Glyceraldehyde3phosphate Dehydrogenase NDP Kinase B Elongation Factor 2 S100 A7 PE Binding protein CALM 3 Gelsolin Precursor Beta Actin

0.74 1.02 9.86 0.64 2.86 NQ 2.97 2.86 1.2 0.78 0.67 1.63 0.45 0.89 0.78 1.13 0.48 ND 7.77 ND 1.04 1.3 1.17 1.13 1.37 0.93 1.26 NQ ND 1.25 1.04 1.22

2.15 3.07 1.14 0.61 1.65 NQ 1.25 1.65 0.96 1.14 0.5 0.29 0.57 0.72 1.43 0.58 0.45 ND 4.35 ND 1.37 1.7 1.88 1.27 1.55 1.7 1.47 NQ ND 0.97 1 1.55

1.94 1.19 1.13 1.19 1.46 NQ 1.73 1.19 1.56 0.36 0.57 ND 0.67 0.62 0.79 1.15 ND ND ND ND 1.26 1.16 0.78 1.45 0.75 1.35 ND 1.3 ND 1.15 0.95 0.86

1.53 1.88 1.71 1.53 3.38 3.97 0.87 3.38 1.66 0.39 0.61 ND 0.85 0.63 0.67 0.67 ND ND ND ND 1.66 1.25 1.27 2.06 1.5 1.17 ND 1.4 ND 1.15 1.03 1.35

1.77 0.81 1.25 3.91 0.8 2.35 1.96 0.8 ND 0.54 0.72 0.5 1.36 0.64 0.58 ND 0.84 4.09 ND 1.34 1.09 0.9 1.13 1.17 1.13 1.1 1.35 ND 1.4 4.55 1.04 0.92

1.63 1.5 1.5 3.54 0.99 2.89 1.52 0.99 ND 0.78 0.6 0.53 0.99 0.47 0.66 ND 1.18 1.77 ND 1.51 1.12 1.87 2.35 1.17 1.12 1.25 1.28 ND 1.57 1.08 1.15 1.11

0.88 1.18 1.04 0.69 1.16 NQ 1.2 1.16 0.77 0.96 0.56 0.43 0.92 0.99 0.89 1.11 0.51 ND 0.97 ND 1.12 0.81 0.96 0.73 1.39 0.93 1.03 NQ ND 0.89 0.9 1.15

0.74 0.84 1.24 0.91 1.19 NQ 0.82 0.61 0.97 0.29 0.57 ND 0.77 0.7 0.97 1.24 ND ND ND ND 0.94 1.33 1.17 0.93 0.62 1.06 ND 0.88 ND 1 1.05 0.78

0.75 0.99 1.1 1.57 0.61 0.81 1.83 1.46 ND 0.82 0.83 0.75 0.79 0.69 0.87 ND 1.19 0.95 ND 1.23 1.09 1.04 0.83 1.05 1.36 1.24 1.34 ND 1.05 0.89 1.02 1.11

0.84 1 1.23 0.88 0.98 ND 1.2 1 1.17 0.87 ND 0.68 0.81 0.6 0.83 ND 1.23 ND ND ND 1.01 1.23 0.89 1.04 1.17 1.09 1.22 0.81 ND 0.51 0.91 0.94

0.78 0.8 0.9 1.16 1.05 ND 1.38 0.81 1.05 0.77 0.54 ND 0.69 0.75 0.93 ND ND ND ND 0.81 0.87 0.88 1.23 0.9 0.71 ND 1.07 1.03 ND 1.02 1.05 0.71

1.03 0.97 ND 1.66 1.02 ND 1.5 0.85 ND 0.73 0.83 0.99 1.06 ND 0.93 ND 1.19 ND ND 1.13 0.97 1.07 0.95 ND 0.97 0.77 0.93 1.09 ND 1.06 1.02 0.99

a Ratios are from the comparison between OPLs (D1-D6) and the pooled normal sample, and that between histologically normal oral tissues (N1-N3) and the pooled normal sample; (N4- N6) are histologically normal oral tissues analyzed in an earlier iTRAQ analysis, using the same pooled normal control to demonstrate consistent iTRAQ ratios in different experiments analyzed over different time periods. (ND, not detected; NQ, not quantified).

ions in the MS/MS scans of the iTRAQ-labeled peptides (21). The ratios of the peak areas of the iTRAQ reporter ions reflect the relative abundances of the peptides and the relative concentrations of the proteins in the samples. Larger, sequenceinformation-rich fragment ions were also produced under these MS/MS conditions and gave the identity of the protein from which the peptide originated. Data Analysis. The software used for data acquisition was Analyst QS 1.1 (Applied Biosystems/MDS SCIEX) and for analysis ProteinPilot.21,24 The database searched was the Celera human database (human KBMS 20041109) with a total of 178 243 entries, provided by Applied Biosystems. Identified proteins were grouped by the software to minimize redundancy; all peptides used for the calculation of protein ratios were unique to the given protein or proteins within the group. The ProteinPilot cutoff score used was 1.3, which corresponds to a confidence limit of 95%. Biomarker Panel Analysis. The average iTRAQ ratio from the replicate injections was calculated for each protein. Proteins selected for further statistical analysis met the following criteria: (1) detection in g3 of 6 OPLs, and >50% of which showed expression changes g50% relative to the control sample, and/ or (2) known to be of interest based on their biological functions or associations with tumorigenesis. These criteria are similar to those employed in a number of our studies involving iTRAQ analysis.17,20,21 These selected proteins are listed in Table 1, along with two housekeeping proteins (to contrast the 302

Journal of Proteome Research • Vol. 8, No. 1, 2009

performances of the potential biomarkers). To identify a panel of best-performing proteins that can distinguish between OPL and normal tissues, each protein in Table 1 was individually assessed for its ability to discriminate between histological normal and OPL samples by evaluating its receiver operating characteristic (ROC) performance based on the iTRAQ ratio values in terms of sensitivity and specificity using the ROCR package within the R statistical computing environment.21,25 Proteins giving the highest AUC values were selected for biomarker panel analysis and used as input variables into a Nav¨e Bayes model, implemented in JAVA26 in the WEKA package.27 Nine trials of 3-fold cross-validation were used for each biomarker panel input into the Na¨ve Bayes model. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated for each trial and the averages are shown in Table 2A. The ROC curve for the panel of the three-best biomarkerssstratifin, YWHAZ (143-3 zeta), and heterogeneous nuclear ribonucleoprotein K (hnRNPK)sis depicted in Figure 1A. Verification of Candidate Biomarkers by Immunohistochemistry. The panel of the three-best biomarkers, stratifin, YWHAZ and hnRNPK, together with two additionally promising proteins, S100A7 and prothymosin alpha (PTHA), were evaluated for their performances using immunohistochemistry (IHC) on an independent set of 30 OPLs and 21 histological normal oral tissues. We included S100A7 because it had high individual AUC value and was identified as one of the best-performing PCMs

research articles

Oral Dysplasia Networks and Biomarker Discovery Table 2. Receiver-Operating Characteristic analyses of (A) the iTRAQ Ratios and (B) IHC Scores of Three Best-Performing Biomarkers, YWHAZ, Stratifin, and hnRNPK, Individually and As a Panel of the Three biomarker

sensitivity

PPV

NPV

AUC

YWHAZ Stratifin hnRNPK Panel of the three

(A) iTRAQ Analysis 0.33 1.0 0.81 1.0 0.17 1.0 0.83 0.74

specificity

1.0 1.0 1.0 0.87

0.43 0.75 0.38 0.69

0.78 0.82 0.78 0.85

YWHAZ Stratifin hnRNPK Panel of the three

(B) IHC Analysis 0.90 0.95 0.77 0.95 0.80 0.91 0.91 0.95

0.96 0.96 0.92 0.96

0.87 0.74 0.76 0.88

0.93 0.93 0.89 0.97

in our earlier iTRAQ analysis of HNOSCCs;21 it was important to determine whether overexpression of S100A7 occurred in early stages in the development of HNOSCC. PTHA was included because it also had high individual AUC value and had been reported to be important in other cancers.28-30 The sources of the antibodies and dilutions used for IHC are given in Supplementary Table S2 in Supporting Information. After histological confirmations of dysplasia in OPLs and normal oral mucosa in the control tissues, paraffin-embedded tissue sections were processed for immunohistochemistry.21 Briefly, after antigen retrieval, tissue sections were incubated with the primary antibodies (see Supplementary Table S2 in Supporting Information for details) for 16 h at 4 °C, followed by the respective biotin-conjugated secondary antibodies and detected using the streptavidin-biotin complex (DAKO LSAB plus kit, DAKO Cytomation, Glostrup, Denmark) and diaminobenzidine as the chromogen. In the negative controls, the primary antibody was replaced by isotype-specific nonimmune mouse IgG to ensure specificity. HNOSCC sections with known immunopositivity for respective proteins as reported earlier21 were used as positive control in each batch of sections analyzed (Supplementary Figure S1 in Supporting Information). Evaluation of Immunohistochemical Staining. Immunopositive staining was evaluated in five areas of the tissue sections as described.21 Sections were scored as positive if epithelial cells showed immunopositivity in the cytoplasm, plasma membrane, and/or nucleus when observed by two evaluators who were blinded to the clinical outcome. These sections were scored as follows: 0, < 10% cells; 1, 10-30% cells; 2, 30-50% cells; 3, 50-70% cells; and 4, >70% cells showed immunoreactivity. Sections were also scored semiquantitatively on the basis of intensity as follows: 0, none; 1, mild; 2, moderate; and 3, intense. Finally, a total score (ranging from 0 to 7) was obtained by adding the scores of percentage positivity and intensity for each of the 51 sections (30 OPLs and 21 histologically normal tissues). The IHC data were subjected to statistical analysis as described above for the iTRAQ ratios. Western Blot Analysis of 14-3-3 Proteins in OPLs and Normal Tissues. Whole-cell lysates were prepared from three OPLs and three normal oral tissues using lysis buffer containing 50 mM Tris-Cl (pH 7.5), 150 mM sodium chloride, 10 mM magnesium chloride, 1 mM EDTA (pH 8.0), 1% Nonidet P-40, 100 mM sodium fluoride, 1 mM phenylmethylene sulfonylfluoride, and 2 µL/mL protease inhibitor cocktail.21 Equal amounts of proteins (80 µg/lane) from OPLs and normal tissues were resolved on 12% SDS-polyacrylamide gels. The proteins were then electrotransferred onto polyvinylidene difluoride (PVDF)

membranes. After blocking with 5% nonfat milk in TBS (0.1 M, pH 7.4), blots were incubated at 4 °C overnight with the respective antibodies (details given in Supplementary Table S2 in Supporting Information). The abundance of R-tubulin was used as a control for protein loading. Membranes were incubated with the respective secondary antibodies, horseradish peroxidase-conjugated rabbit/goat/mouse anti-IgG at the appropriate dilution in 1% bovine serum albumin for 2 h at room temperature. After each step, blots were washed three times with Tween (0.2%)-Tris-buffer saline (TTBS). Protein bands were detected by the enhanced chemiluminescence method (ECL, Santa Cruz Biotechnology, Inc.) on XO-MAT film. Reverse Transcription-Polymerase Chain Reaction (RT-PCR) Analysis. To determine if overexpression of the five proteins in OPLs were due to increase in the transcript levels, RT-PCR analysis was performed using total RNA isolated from OPLs and normal oral mucosa (3 each), using gene-specific primers for YWHAZ, stratifin, hnRNPK, S100A7 and PTHA, and β-actin as the control [see Supplementary Table S3 in Supporting Information for gene-specific primer sequences (synthesized by Microsynth, Switzerland) and PCR conditions used]. Syntheses of cDNAs (cDNAs) were carried out by reverse transcription of 2.0 µg of total RNA using MMLV reverse transcriptase. PCR amplification was carried out in a total volume of 20 µL containing 3 µL of reverse transcribed cDNA, 10× PCR buffer, 10 mM dNTPs, 20 µM of each primer and 1 U of Taq polymerase. After 5 min of initial denaturation, 32 amplification cycles of 1 min at 94 °Cs1 min at specific annealing temperature and 1 min at 72 °Cswere carried out, followed by a 10min elongation at 72 °C. β-Actin was used as a control to optimize the amounts of cDNAs generated. PCR products were separated on 1.5% agarose gel, stained with ethidium bromide, and visualized with ChemiImager IS-4400 (Alpha Innotech Corp., CA).21 Network Analysis. The 30 proteins listed in Table 1 were used for pathway analysis. HUGO or Swiss-Prot accession numbers were imported into the Ingenuity Pathway Analysis (IPA) Software (Ingenuity Systems, Mountain View, CA). The IPA database consists of proprietary ontology representing 300 000 biologic objects ranging from genes, proteins, and molecular and cellular processes. More than 11 200 human genes are currently represented in the database. The proteins were categorized based on location, cellular components, and reported or suggested biochemical, biologic, and molecular functions. Identified proteins were mapped to networks that were generated based on evidence from existing literature and then ranked by score. A score of 3 or higher has a 99.9% confidence level of not being generated by random chance alone and was used as the cutoff for identifying protein networks.

Results Discovery and Identification of Proteins in OPLs Using iTRAQ-LC-MS/MS and Biomarker Panel Analysis. The LC-MS/ MS analyses collectively resulted in identification of 439 nonredundant proteins; 216 were identified as single hits with more than 95% confidence. These proteins were classified on the basis of their known molecular functions (Supplementary Figure S2A in Supporting Information) and their involvement in different biological processes (Supplementary Figure S2B in Supporting Information). Of all the proteins identified, only 17 passed our criteria for further statistical analysis (vide supra). Of this subset, 15 proteins were confidently identified with a Journal of Proteome Research • Vol. 8, No. 1, 2009 303

research articles

Ralhan et al.

Figure 1. Receiver-operating characteristic analyses of a panel of three best-performing biomarkers, YWHAZ, stratifin, and hnRNPK: (A) iTRAQ ratios, and (B) IHC scores. See the text for details.

minimum of two peptide matches in each case (See Table S4 in Supporting Information for peptide sequences and coverage). Two proteins, parathymosin and DLC1, were identified by single peptides (See Figure S3A,B for the CID spectra of the single-peptide identifications in Supporting Information). These 17 proteins are given in Table 1, along with two structural proteins, β-actin and gelsolin precursor, as controls. Table 1 also depicts the variations in the levels of overexpressed and underexpressed proteins in individual OPL and histological normal tissues versus the pooled normal control in the same experiment (N1-N3) as well as across different sets of experiments (N4-N6). Our results suggested reproducibility within the experiment (triplicate injection of one preparation of each sample set) as well as across different sets of experiments (as demonstrated by the normal tissue analysis in Table 1), confirming the reproducibility in both the technique and in the experiments. These differential expression levels were averages of the replicate injections: 56.4% of the ratios varied by less than 10% from their respective averages shown, and 82.0% varied by less than 20%. Thirteen proteins that did not meet the aforementioned initial criteria, IGL2, p37AUF1 (hnRNPD), SOD2, PKM2, hnRNPA1, HSP27, cofilin, glyceraldehyde3-phosphate dehydrogenase, NDP kinase B, elongation factor 2, CALM3, PEBP and S100A7, were also included in Table 1 for further analysis, as they are of biological relevance in cancer development. Of these, 11 proteins were confidently identified with a minimum of two peptide matches in each case (See Table S4 in Supporting Information for peptide sequences and coverage. p37AUF1 (hnRNPD) was identified by a single peptide with a confidence of 99% (See Figure S3C for the CID spectra of the single-peptide identification in Supporting Information). SOD2 was identified by more than one unique peptide; however, the best-matching peptide was identified with a confidence of only 93%. Although this peptide did not meet our stipulated criteria for acceptance, manual verification of the spectrum showed good sequence coverage for this peptide (Figure S3D in Supporting Information). Further, the cumulative score, which included the lower confidence peptide matches, was >2.0 and corresponded to a confidence of 99%. The best-performing proteins that can differentiate between OPLs and normal tissues were identified by determining the individual ROC curves of the proteins in Table 1. The three proteins with the highest AUC values, YWHAZ, stratifin and hnRNPK, are listed in Table 2A together with their individual and collective figures-of-merit, including sensitivity and speci304

Journal of Proteome Research • Vol. 8, No. 1, 2009

ficity. As a panel, these three biomarkers achieved a sensitivity of 0.83 and a specificity of 0.74 in discriminating OPLs from histological normal oral tissues (Table 2A and Figure 1A). Verification of Candidate Biomarkers by Immunohistochemistry, Western Blot and RT-PCR Analyses. The panel of three potential biomarkers, YWHAZ, stratifin and hnRNPK, and two other proteins with high AUC values, S100A7 (0.56) and PTHA (0.56), were chosen for verification in an independent set of OPLs (30 cases) and normal tissues (21 cases) by IHC. Representative levels of expression and subcellular localizations of all five proteins in oral dysplastic tissues in comparison with normal tissues are shown in Figure 2. (Supplementary Figure S1 in Supporting Information depicts the positive and negative controls used for each protein analyzed by IHC.) These data were further verified by Western blot analysis (Figure 3a) at the protein level, as well as RT-PCR analysis at the mRNA level (Figure 3b). The differential expression suggested by iTRAQ ratios tended to be moderate, whereas the results of Western and RT-PCR analyses tended to show more extreme differential expression. Thus, Western and RT-PCR analyses, verified the differential expression reported by the iTRAQ analysis in trend, but not in scale. This discrepancy of scale has also been noted in other studies and has been ascribed to compression of the dynamic range of iTRAQ ratios.31 Specifically, in that study, we determined that a 2-fold differential expression as determined by iTRAQ analysis was in reality closer to a 4-fold differential expression in an absolute quantification study that was performed on the same samples. Importantly, in IHC analysis, the biomarker panel of YWHAZ, stratifin, and hnRNPK achieved a sensitivity of 0.91 and a specificity of 0.95 (Table 2B and Figure 1B) in discriminating OPLs from histological normal oral tissues. Network Analysis. To gain insight into the plausible biological processes in which these proteins might be involved, we used the Ingenuity Pathway Analysis tools (Ingenuity Systems, Inc. software) and discovered two major networks in OPLs (the merged network is shown in Figure 4; see also Supplementary Figures S4 and S5 in Supporting Information). The first network (depicted in Supplementary Figure S4 in Supporting Information) comprised 15 proteins that are primarily involved in inflammation, molecular transport, cellular movement, and cancer. The second network comprised seven proteins involved in cellular signaling, proliferation, and gene expression (See Supplementary Figure S5 in Supporting Information). To the best of our knowledge, ours is the first study reporting

Oral Dysplasia Networks and Biomarker Discovery

research articles

Figure 2. Immunohistochemical verification of iTRAQ-discovered potential biomarkers, YWHAZ, stratifin, hnRNPK, S100A7 and PTHA in OPLs and histologically normal oral tissues. Verification of these five potential biomarkers was carried out using an independent set of 30 OPLs and 21 histological normal oral tissues. Representative photomicrographs are shown here. Positive staining is brown and is intense in OPLs. The upper panel shows the normal tissues and the lower panel the OPL tissue sections. (A) The OPL sample shows intense cytoplasmic and nuclear staining for YWHAZ, while the normal mucosa shows no detectable immunostaining. (B) The OPL tissue section exhibits cytoplasmic staining for stratifin in the epithelial cells, while the normal mucosa shows no detectable immunostaining. (C) The OPL tissue section shows nuclear staining for hnRNPK in the epithelial cells, while no detectable immunostaining is evident in the normal mucosa. (D) The OPL sample shows intense cytoplasmic staining for S100A7 in epithelial cells, while the normal mucosa displays no detectable immunoreactivity. (E) The OPL sample exhibits intense nuclear staining for PTHA in epithelial cells, while no detectable immunostaining is evident in the normal sample. All panels show ×100 magnifications.

differential expressions of p37AUF1 and histoneH2B.1 in OPLs. These proteins and their cellular functions are listed in Supplementary Table S5 in Supporting Information.

Discussion To our knowledge, this is the first study that exploits tissue proteomics with iTRAQ-tagging and LC-MS/MS analyses of small, clinical OPL samples for protein expression analysis. Differential expression was determined using a pool of nonpaired histological normal oral tissues from individuals with no evidence of OPLs or oral cancer. The rationale for our choice of nonpaired histological normal oral tissues, as opposed to using the adjacent morphologically normal tissues obtained from OPL patients, as the control was based on the possibility of field cancerization. Field cancerization causes histologically normal tissues adjacent to the cancer site to harbor molecular changes; using these tissues as controls may mask or attenuate alterations in protein expressions that signify early events in tumorigenesis.32-34 Table 1 also shows the analysis of three histological normal samples used in a previous exercise to demonstrate consistency and validity over a period of 6 months.21 The number of OPLs examined in this studyssix for the LC-MS/MS and 30 for subsequent verificationsswas relatively modest, but was necessitated by the small number and size of OPL samples available. Nevertheless, we successfully demonstrated the utility of iTRAQ-labeling of small OPL biopsy specimens and detection of a large number of expressed proteins that led to the discovery of a panel of candidate OPL biomarkers. Replicate analyses demonstrated that the expression ratios were reproducibles82% were within 20% of the averages shown in Table 1. Differential expressions analyses of the proteomes between OPLs and histological normal oral tissues revealed 30

proteins that merit further examination and verification. A panel of three potential biomarkers selected by ROC analyses and two other biologically relevant proteins that had high AUC values were successfully verified to be overexpressed using an independent and larger set of OPLs and histological normal tissues by IHC and Western analyses, thus, confirming findings of the iTRAQ analyses. In addition, RT-PCR analyses showed increased levels of transcripts for all five proteins, suggesting that the increased protein expressions were due to upregulation at the transcriptional level. Our approach enabled the identification of a large number of proteins in oral premalignant lesions, including mediators of inflammatory response, redox system, proteases, chaperones, transcriptional regulators, calcium binding proteins, metabolic enzymes, and proteins involved in cell proliferation and growth, intermediary metabolism, signal transduction, cell cycle regulation, cell death, cell motility and cell morphology. Pathway analyses unraveled important novel links between inflammation and cancer. Importantly, it showed direct interaction between all the three proteins sbdYWHAZ, stratifin and hnRNPKsthat constitute the panel of OPL biomarkers. The mechanism involved in upregulation of YWHAZ and stratifin in OPLs remains unknown. hnRNPK is an RNA-binding protein that regulates gene expression at both transcriptional and translational levels.35 Therefore, we speculate that hnRNPK may be involved in regulation of YWHAZ and stratifin in OPLs. Our analysis also suggests that hnRNPK directly regulates the expression of COX2, an enzyme implicated in the synthesis of prostaglandins, which are mediators of the inflammatory response. Our earlier in vitro and in vivo studies demonstrated that COX-2 activation and NF-κB overexpression are parallel events occurring in early precancerous stages of tobaccoassociated oral carcinogenesis and these events remain elevated Journal of Proteome Research • Vol. 8, No. 1, 2009 305

research articles

Figure 3. (a) Western blot analyses of YWHAZ, stratifin, hnRNPK, S100A7 and PTHA in representative OPLs (labeled as dysplasia) and histologically normal oral tissues. The OPLs and histologically normal oral tissues (n ) 3) were selected randomly from the same cohort of tissues as used for IHC analyses. Equal amounts of protein lysates from OPLs (D1-D3) and histologically normal oral tissues (N1-N3) were used. The panels show increased expression of (1) YWHAZ, (2) stratifin, (3) hnRNPK, (4) S100A7, and (5) PTHA in OPLs (D1-D3) as compared to the histologically normal oral tissues (N1-N3). R-Tubulin (Panel 6) was used as the loading control. (b) RT-PCR analyses of YWHAZ, stratifin, hnRNPK, S100A7, and PTHA in representative OPLs and histologically normal oral tissues selected randomly from the same cohort as used for IHC analyses: Panel 1 shows increased levels of YWHAZ transcripts in OPLs (D1-D3) as compared to the histologically normal oral tissues (N1-N3) that do not show detectable levels of YWHAZ transcripts. Panel 2 demonstrates increased levels of stratifin transcripts in OPLs (D1-D3) as compared to the histologically normal oral tissues (N1-N3) that show basal level (N1) and no detectable level (N2 and N3) of stratifin transcripts. Panel 3 shows increased levels of hnRNPK transcripts in OPLs (D1-D3) as compared to the histologically normal oral tissues (N1-N3), in which no detectable levels of hnRNPK transcripts are evident. Panel 4 exhibits increased levels of S100A7 transcripts in OPLs (D1-D3) as compared to no detectable levels in the histologically normal oral tissues (N1-N3). Panel 5 shows increased levels of PTHA transcripts in OPLs (D1-D3) as compared to the histologically normal oral tissues (N1-N3), in which no detectable levels of PTHA transcripts are evident. β-Actin (Panel 6) was used as a control for normalizing the quantity of RNA used. 306

Journal of Proteome Research • Vol. 8, No. 1, 2009

Ralhan et al. down the tumorigenic pathway. Tobacco carcinogens, including tobacco specific nitrosamines (TSNA), 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK), appear to activate these proteins involved in the inflammatory response of epithelial cells and initiation of the carcinogenic cascade.36 The other important links identified are regulation of NF-κB by SERPINA1 and phosphatidyl ethanolamine binding protein 1 (PEBP1). SERPINA1 has been shown to reduce the activation of NF-kB;37 decrease in SERPINA1 levels in OPLs may partly account for activated NF-κB in OPLs. Pathway analyses also showed, for the first time, that deregulation of calciumassociated proteins (cystatin B, FABP5 and S100A7) and mitochondrial dysfunction (SOD2) may play an important role in the development of OPLs. Importantly, YWHAZ was found to directly interact with nucleophosmin (NPM1), an important protein involved in increased cellular proliferation, and with HSP90B1, which has been reported to increase the activity of ERK. HSP90B1, YWHAZ, nucleophosmin1, parathymosin (PTMS), SOD2, and PEBP1 are all involved in the inhibition of apoptosis. Furthermore, HSP90B1, SOD2, and stratifin are associated with increased cell viability. Increased expressions of hnRNPK, PTHA, stratifin, SOD2 and nucleophosmin1, and reduced expression of DLC1, play a role in hyperproliferation of cells. Increased cell proliferation and inhibition of apoptosis are two hallmarks of cancer, and our data suggest that both events are occurring in early, premalignant stages. Alterations in cytoskeleton are important events in tumorigenesis; deregulation of YWHAZ and its interaction with MARCKS, beta-actin and tubulin identified herein suggest the implication of cytoskeletal reorganization in development of oral dysplasias. It is noteworthy that invasion is an important event in the progression of dysplasia as well as of cancer, and five proteins identified in our studysunderexpressions of DLC1, IGHG1 and FABP5, and overexpressions of S100A7 and PEBP1sare all involved in cell invasion. Significantly, this is the first report of p37AUF1 (hnRNPD) expression in OPLs, and pathway analyses suggest its potential interaction with stratifin and scaffold protein b (SAFB). PEBP1 is oncogenic, while FABP5 and SERPINA1 are metastatic. Taken together, the discovery of these alterations in OPLs suggests that these proteins may be associated with the potential of malignant transformation. Our findings herein certainly warrant additional validation; elucidation of functional significance in future studies will provide further insight into the biology of development and progression of OPLs. None of the aforementioned proteins can be classified individually as specific biomarkers for OPLs. However, a panel of the three best-performing biomarkers, YWHAZ, stratifin, and hnRNPK, confer satisfactory performance. Our current study clearly demonstrates overexpression of YWHAZ in OPLs, suggesting its involvement in early stages of oral carcinogenesis. Mechanistically, YWHAZ overexpression increases p53 protein degradation via hyperactivation of the phosphoinositide 3-kinase (PI3K)-Akt signaling pathway that phosphorylates MDM2. Replacement of p53 leads to luminal cell apoptosis. Indeed, YWHAZ overexpression and consequential downregulation of p53 have been proposed to be critical early events causing the morphologic and cytologic changes in early stage breast disease that promote the development of breast cancer.38 YWHAZ is known to be involved in diverse cellular processes, many of which are deregulated in HNOSCCs.39 Our ongoing studies on functional analysis of YWHAZ in oral cancer cells have also shown its involvement in the activation of PI3K-Akt signaling pathway and cytoskeletal reorganization (data not shown).

Oral Dysplasia Networks and Biomarker Discovery

research articles

Figure 4. Network analysis using Ingenuity Pathways Analysis (IPA) software. Differentially expressed proteins identified in OPLs in comparison with normal oral tissues were analyzed using the IPA software. Network analysis classified proteins into two networks on the basis of functions published in the literature; the merged network is shown. Solid lines show direct interactions/regulations, while dashed lines show indirect interactions/regulations of proteins at ends of the lines. Proteins shown in red are upregulated; those in green are downregulated in OPLs in comparison with normal tissues.

Stratifin, another member of the 14-3-3 protein family, was overexpressed in OPLs and emerged as one of the bestperforming potential biomarkers. Our present data suggest that the increase in stratifin expression is an early event in the development of cancer. Importantly, our recent proteomic analysis and others’ results showed overexpression of stratifin in HNOSCCs21,40 increased concomitant expression of stratifin and YWHAZ served as adverse prognosticators in HNOSCCs, underscoring the importance of these proteins in head-andneck tumorigenesis.41 Interestingly and significantly, stratifin was reported to be downregulated in HNSCCs by Roesch-Ely et al.,14 whereas we observed consistent overexpression of stratifin in iTRAQ and in IHC verification analyses. The HNSCCs in the study of Roesch-Ely et al.14 were from the German population with tobacco smoking and alcohol consumption being the major risk factors, while the clinical samples in this study, and in Lo et al.,40 were from Asian populations, where, in addition, chewing tobacco and/or betel quid, bidi smoking and HPV infection are important risk factors. In support of these observations, Bhawal et al.,42 reported that hypermethylation of stratifin promoter is not a frequent event in HNSCC. Moreover, IKK alpha, a catalytic subunit of the IKK complex, has been shown to protect the stratifin locus from hypermethylation; this function serves to maintain genomic stability in keratinocytes.43 Heterogeneous nuclear ribonucleoprotein K protein (hnRNPK), identified by iTRAQ analysis and verified by IHC in OPLs, is an interesting protein that has been strongly implicated as a key player of tumorigenesis.44 hnRNPK is overexpressed, aberrantly localized, and associated with poor prognosis in colorectal cancer,45 while its transcriptional upregulation was reported in OSCC.46 In view of a role of hnRNPK as a transformation-related protein, its overexpression in OPLs is

an important finding; in-depth studies are warranted to establish its link, if any, with transformation potential of OPLs. Prothymosin alpha (PTHA), overexpressed in a subset of OPLs, has been proposed to be a proliferation marker in patients with thyroid cancer.28 This protein was implicated in various other cancers, including gastric, lung, liver, colon, and breast cancers.29,30 S100A7, a small calcium-binding protein, is upregulated in abnormally differentiating keratinocytes, squamous carcinomas of different organs, and in a subset of breast tumors.47 S100A7 has been identified in oral premalignant epithelia by microarray analysis and proposed to be a marker for invasion.48 It is postulated to play a role in breasttumor progression in association with increased hypoxia and reactive oxygen species (ROS) by promoting angiogenesis.49 Increased hypoxia and ROS also occur in OPLs and HNOSCCs, and might explain the observed changes in S100A7 expression here. Reciprocal negative regulation between S100A7 and β-catenin signaling has been shown to play an important role in tumor progression of OSCC.50 In conclusion, proteomic analyses of OPLs revealed the integrated importance of alterations in multiple cellular processes and suggested novel links between inflammation and premalignancy, some of which may serve as potential chemopreventive/therapeutic targets. Validation of the panel of OPL biomarkers in larger studies will ascertain its clinical utility and long-term patient follow up will evaluate the potential of these biomarkers for predicting the risk of malignant transformation in OPLs. Abbreviations. AUC, area under the curve; ACTB, beta-actin; ICAT, isotope-coded affinity tags; iTRAQ, isobaric mass tags; LC, liquid chromatography; MS/MS, tandem mass spectrometry; OPLs, oral premalignant lesions; PBM, potential biomarker; HNOSCC, head-and-neck/oral squamous cell carcinoma; Journal of Proteome Research • Vol. 8, No. 1, 2009 307

research articles CALM 3, calmodulin; GRP94, endoplasmin precursor; hnRNPK, transformation related heterogeneous nuclear ribonucleoprotein K; MARCKS, myristoylated alanine-rich C kinase; DLC1, deleted in liver cancer 1; FABP5, fatty acid binding protein 5; HSP90B1, 90 kDa heat shock protein B or Grp94, 94-kDa glucose-regulated protein; hnRNPA1, heterogeneous nuclear ribonucleoprotein A1; IGHG1, immunoglobulin G heavy chain 1; PBS, phosphate-buffered saline; SCX, strong cation exchange; ID, internal diameter; IGL2, intergeniculate leaflet 2; p37AUF1, A+U-rich element binding factor 1 or hnRNPD, heterogeneous nuclear ribonucleoprotein D; Hsp27, 27 kDa heat shock protein; NDP kinase B, nucleoside diphosphate kinase B; IDA, information-dependent acquisition; PE binding protein, phosphatidyl ethanolamine binding protein (PEBP); TBS, tris-buffered saline; S100A 7, psoriasin; S100A 11, calgizarrin; SFN, stratifin (14-3-3 sigma); SERPINA1, serpin peptidase inhibitor clade A member 1 or alpha-1-antitrypsin; SD, standard deviation; ROC, receiver operating characteristics; PV, predictive values; PPV, positive predictive values; NPM1, nucleophosmin 1; NPV, negative predictive values; PTHA, prothymosin alpha; PK M2, pyruvate kinase isozyme M2; SOD2, copper-zinc superoxide dismutase; YWHAZ, 14-3-3 zeta; RSD, relative standard deviation; TSNA, Tobacco specific nitrosamines; NNK, 4-(methylnitrosamino)1-(3-pyridyl)-1-butanone.

Acknowledgment.

The authors gratefully acknowledge the support and cooperation of their pathology collaborator, Dr. Terence J. Colgan; and the technical assistance of Muntajib Alhaq and Maria Mendes (Mount Sinai Hospital) and Prof. Sudhir Bahadur (All India Institute of Medical Sciences). R.R. is a recipient of NCI-Novartis Translational Cancer Research Fellowship Award of UICC at Centre for Research in Mass Spectrometry, Department of Chemistry, York University. Funding: R.R., NCI-Novartis Translational Cancer Research Fellowship of International Union against Cancer (UICC), Ontario Institute for Cancer Research, Opportunity Fund, Canadian Institutes of Health Research (CIHR); A.M., Senior Research Fellowship of Council of Scientific and Industrial Research, India. K.W.M.S., Canadian Institutes of Health Research (CIHR), infrastructural support from the Ontario Research and Development Challenge Fund, Applied Biosystems, and MDS Analytical Technologies.

Supporting Information Available: Supplementary Tables S1-S5 and Figures S1-S5. This material is available free of charge via the Internet at http://pubs.acs.org.

Ralhan et al.

(7)

(8)

(9)

(10) (11)

(12)

(13)

(14)

(15)

(16)

(17)

(18)

(19)

(20)

References (1) Reichart, P. A. Oral mucosal lesions in a representative crosssectional study of aging Germans. Community Dent. Oral Epidemiol. 2000, 28, 390–8. (2) Petti, S. Pooled estimate of world leukoplakia prevalence: a systematic review. Oral. Oncol. 2003, 39, 770–80. (3) Napier, S. S. Natural history of potentially malignant oral lesions and conditions: an overview of the literature. J. Oral. Pathol. Med. 2008, 37, 1–10. (4) Warnakulasuriya, S.; Reibel, J.; Bouquot, J.; Dabelsteen, E. Oral epithelial dysplasia classification systems: predictive value, utility, weaknesses and scope for improvement. J. Oral Pathol. Med. 2008, 37, 127–33. (5) Warnakulasuriya, S.; Johnson, N. W.; van der Waal, I. Nomenclature and classification of potentially malignant disorders of the oral mucosa. J. Oral Pathol. Med. 2007, 36, 575–80. (6) Brennan, M.; Migliorati, C. A.; Lockhart, P. B.; Wray, D.; Al-Hashimi, I.; Axe´ll, T.; Bruce, A. J.; Carpenter, W.; Eisenberg, E.; Epstein, J. B.;

308

Journal of Proteome Research • Vol. 8, No. 1, 2009

(21)

(22) (23)

(24)

Holmstrup, P.; Jontell, M.; Nair, R. Management of oral epithelial dysplasia: a review. Oral Surg., Oral Med., Oral Pathol., Oral Radiol. Endod. 2007, 103, 1–12. Hsue, S. S.; Wang, W. C.; Chen, C. H.; Lin, C. C.; Chen, Y. K.; Lin, L. M. Malignant transformation in 1458 patients with potentially malignant oral mucosal disorders: a follow-up study based in a Taiwanese hospital. J. Oral Pathol. Med. 2007, 36, 25–9. Poh, C. F.; Ng, S. P.; Williams, P. M.; Zhang, L.; Laronde, D. M.; Lane, P.; Macaulay, C.; Rosin, M. P. Direct fluorescence visualization of clinically occult high-risk oral premalignant disease using a simple hand-held device. Head Neck 2007, 29, 71–6. Farah, S. C.; McCulloughb, M. J. A pilot case control study on the efficacy of acetic acid wash and chemiluminescent illumination (ViziLite) in the visualisation of oral mucosal white lesions. Oral Oncol. 2007, 43, 820–4. Hunter, K. D.; Parkinson, E. K.; Harrison, P. R. Profiling early head and neck cancer. Nat. Rev. Cancer 2005, 5, 127–35. Rosin, M. P.; Cheng, X.; Poh, C.; Lam, W. L.; Huang, Y.; Lovas, J.; Berean, K.; Epstein, J. B.; Priddy, R.; Le, N. D.; Zhang, L. Use of allelic loss to predict malignant risk for low-grade oral epithelial dysplasia. Clin. Cancer. Res. 2000, 6, 357–62. Kawaguchi, H.; El-Naggar, A. K.; Papadimitrakopoulou, V.; Ren, H.; Fan, Y. H.; Feng, L.; Lee, J. J.; Kim, E.; Hong, W. K.; Lippman, S. M.; Mao, L. Podoplanin: a novel marker for oral cancer risk in patients with oral premalignancy. J. Clin. Oncol. 2008, 26, 354– 60. Koike, H.; Uzawa, K.; Nakashima, D.; Shimada, K.; Kato, Y.; Higo, M.; Kouzu, Y.; Endo, Y.; Kasamatsu, A.; Tanzawa, H. Identification of differentially expressed proteins in oral squamous cell carcinoma using a global proteomic approach. Int. J. Oncol. 2005, 27, 59–67. Roesch-Ely, M.; Nees, M.; Karsai, S.; Ruess, A.; Bogumil, R.; Warnken, U.; Schno¨lzer, M.; Dietz, A.; Plinkert, P. K.; Hofele, C.; Bosch, F. X. Proteomic analysis reveals successive aberrations in protein expression from healthy mucosa to invasive head and neck cancer. Oncogene 2007, 26, 54–64. Patel, V.; Hood, B. L.; Molinolo, A. A.; Lee, N. H.; Conrads, T. P.; Braisted, J. C.; Krizman, D. B.; Veenstra, T. D.; Gutkind, J. S. Proteomic analysis of laser captured paraffin embedded tissues: A molecular portrait of head and neck cancer progression. Clin. Cancer. Res. 2008, 14, 1002–14. Gygi, S. P.; Rist, B.; Gerber, S. A.; Turecek, F.; Gelb, M. H.; Aebersold, R. Quantitative analysis of complex protein mixtures using isotope-coded affinity tags. Nat. Biotechnol. 1999, 17, 994– 9. DeSouza, L.; Diehl, G.; Rodrigues, M. J.; Guo, J.; Romaschin, A. D.; Colgan, T. J.; Siu, K. W. Search for cancer markers from endometrial tissues using differentially labeled tags iTRAQ and cICAT with multidimensional liquid chromatography and tandem mass spectrometry. J. Proteome Res. 2005, 4, 377–86. Li, H.; DeSouza, L. V.; Ghanny, S.; Li, W.; Romaschin, A. D.; Colgan, T. J.; Siu, K. W. Identification of candidate biomarker proteins released by human endometrial and cervical cancer cells using two-dimensional liquid chromatography/tandem mass spectrometry. J. Proteome Res. 2007, 6, 2615–22. Guo, J.; Colgan, T. J.; DeSouza, L. V.; Rodrigues, M. J.; Romaschin, A. D.; Siu, K. W. Direct analysis of laser capture microdissected endometrial carcinoma and epithelium by matrix-assisted laser desorption/ionization mass spectrometry. Rapid. Commun. Mass. Spectrom. 2005, 19, 2762–6. DeSouza, L. V.; Grigull, J.; Ghanny, S.; Dube´, V.; Romaschin, A. D.; Colgan, T. J.; Siu, K. W. Endometrial carcinoma biomarker discovery and verification using differentially tagged clinical samples with multidimensional liquid chromatography and tandem mass spectrometry. Mol. Cell. Proteomics 2007, 6, 1170–82. Ralhan, R.; Desouza, L. V.; Matta, A.; Chandra, Tripathi, S.; Ghanny, S.; DattaGupta, S.; Bahadur, S.; Siu, K. W. Discovery and verification of head-and-neck cancer biomarkers by differential protein expression analysis using iTRAQ-labeling and multidimensional liquid chromatography and tandem mass spectrometry. Mol. Cell. Proteomics 2008, 7, 1162–73. Ariztia, E. V.; Lee, C. J.; Gogoi, R.; Fishman, D. A. The tumor microenvironment: key to early detection. Crit. Rev. Clin. Lab. Sci. 2006, 43, 393–425. Koukourakis, M. I.; Giatromanolaki, A.; Harris, A. L.; Sivridis, E. Comparison of metabolic pathways between cancer cells and stromal cells in colorectal carcinomas: a metabolic survival role for tumor-associated stroma. Cancer Res. 2006, 66, 632–7. Shilov, I. V.; Seymour, S. L.; Patel, A. A.; Loboda, A.; Tang, W. H.; Keating, S. P.; Hunter, C. L.; Nuwaysir, L. M.; Schaeffer, D. A. The Paragon Algorithm: A next generation search engine that uses

research articles

Oral Dysplasia Networks and Biomarker Discovery

(25) (26) (27) (28)

(29)

(30) (31)

(32) (33) (34) (35) (36)

(37) (38) (39)

sequence temperature values and feature probabilities to identify peptides from tandem mass spectra. Mol. Cell. Proteomics 2007, 6, 1638–55. The R Development Core Team. The R Project for Statistical Computing. http://www.r-project.org. Sun Developer Network (SDN). Java.sun.com: The Source for Java Developers. http://java.sun.com. Witten, I. H.; Frank, E. Data Mining: Practical Machine Learning Tools and Techniques, 2nd ed.; Morgan Kaufmann: San Francisco, CA, 2005. Letsas, K. P.; Vartholomatos, G.; Tsepi, C.; Tsatsoulis, A.; FrangouLazaridis, M. Fine-needle aspiration biopsy-RT-PCR expression analysis of prothymosin alpha and parathymosin in thyroid: novel proliferation markers. Neoplasma 2007, 54, 57–62. Leys, C. M.; Nomura, S.; LaFleur, B. J.; Ferrone, S.; Kaminishi, M.; Montgomery, E.; Goldenring, J. R. Expression and prognostic significance of prothymosin-alpha and ERp57 in human gastric cancer. Surgery 2007, 141, 41–50. Sasaki, H.; Nonaka, M.; Fujii, Y.; Yamakawa, Y.; Fukai, I.; Kiriyama, M; Sasaki, M. Expression of the prothymosin-a gene as a prognostic factor in lung cancer. Surg. Today 2001, 31, 936–8. DeSouza, L. V.; Taylor, A. M.; Li, W.; Minkoff, M. S.; Romaschin, A. D.; Colgan, T. J.; Siu, K. W. M. Multiple reaction monitoring of mTRAQ labeled peptides enables absolute quantification of endogenous levels of a potential cancer marker in cancerous and normal endometrial tissues. J. Proteome Res. 2008, 7, 2535–34. Slaughter, D. P.; Southwick, H. W.; Smejkal, W. ‘Field cancerisation’ in oral stratified squamous epithelium. Cancer 1953, 6, 963–8. Braakhuis, B. J.; Leemans, C. R.; Brakenhoff, R. H. Expanding fields of genetically altered cells in head and neck squamous carcinogenesis. Semin. Cancer Biol. 2005, 15, 113–20. Garcia, S. B.; Park, H. S.; Novelli, M.; Wright, N. A. Field cancerization, clonality, and epithelial stem cells: the spread of mutated clones in epithelial sheets. J. Pathol. 1999, 187, 61–81. Valverde, R.; Edwards, L.; Regan, L. Structure and function of KH domains. FEBS. J. 2008, 275, 2712–26. Sawhney, M.; Rohatgi, N.; Kaur, J.; Shishodia, S.; Sethi, G.; Gupta, S. D.; Deo, S. V.; Shukla, N. K.; Aggarwal, B. B.; Ralhan, R. Expression of NF-kappaB parallels COX-2 expression in oral precancer and cancer: association with smokeless tobacco. Int. J. Cancer 2007, 120, 2545–2556. Shapiro, L.; Pott, G. B.; Ralston, A. H. Alpha-1-antitrypsin inhibits human immunodeficiency virus type 1. FASEB J. 2001, 15, 115– 122. Danes, C. G.; Wyszomierski, S. L.; Lu, J.; Neal, C. L.; Yang, W.; Yu, D. 14-3-3 zeta down-regulates p53 in mammary epithelial cells and confers luminal filling. Cancer Res. 2008, 68, 1760–7. Matta, A.; Bahadur, S.; Duggal, R.; Gupta, S. D.; Ralhan, R. Overexpression of 14-3-3zeta is an early event in oral cancer. BMC Cancer 2007, 7, 169.

(40) Lo, W. Y.; Tsai, M. H.; Tsai, Y.; Hua, C. H.; Tsai, F. J.; Huang, S. Y.; Tsai, C. H.; Lai, C. C. Identification of over-expressed proteins in oral squamous cell carcinoma (OSCC) patients by clinical proteomic analysis. Clin. Chim. Acta 2007, 376, 101–107. (41) Matta, A.; DeSouza, L. V.; Bahadur, S.; Gupta, S. D.; Ralhan, R.; Siu, K. W. Prognostic significance of head and neck cancer biomarkers previously discovered using iTRAQ-multidimensional liquid chromatography and tandem mass spectrometry. J. Proteome Res. 2008, 7, 2078–87. (42) Bhawal, U. K.; Tsukinoki, K.; Sasahira, T.; Sato, F.; Mori, Y.; Muto, N.; Sugiyama, M.; Kuniyasu, H. Methylation and intratumoural heterogeneity of 14-3-3 sigma in oral cancer. Oncol. Rep. 2007, 18, 817–824. (43) Zhu, F.; Xia, X.; Liu, B.; Shen, J.; Hu, Y.; Person, M.; Hu, Y. IKKalpha shields 14-3-3sigma, a G(2)/M cell cycle checkpoint gene, from hypermethylation, preventing its silencing. Mol. Cell 2007, 27, 214– 27. (44) Carpenter, B.; McKay, M.; Dundas, S. R.; Lawrie, L. C.; Telfer, C.; Murray, G. I. Heterogeneous nuclear ribonucleoprotein K is over expressed aberrantly localized and is associated with poor prognosis in colorectal cancer. Br. J. Cancer 2006, 95, 921–7. (45) Roychoudhury, P.; Chaudhuri, K. Evidence for heterogeneous nuclear ribonucleoprotein K overexpression in oral squamous cell carcinoma. Br. J. Cancer 2007, 97, 574–5. (46) Skopeliti, M.; Kratzer, U.; Altenberend, F.; Panayotou, G.; Kalbacher, H.; Stevanovic, S.; Voelter, W.; Tsitsilonis, O. E. Proteomic exploitation on prothymosin alpha-induced mononuclear cell activation. Proteomics 2007, 7, 1814–24. (47) Moubayed, N.; Weichenthal, M.; Harder, J.; Wandel, E.; Sticherling, M.; Gla¨ser, R. Psoriasin (S100A7) is significantly up-regulated in human epithelial skin tumours. J. Cancer. Res. Clin. Oncol. 2007, 133, 253–61. (48) Sparano, A.; Quesnelle, K. M.; Kumar, M. S. Genome-wide profiling of oral squamous cell carcinoma by array-based comparative genomic hybridization. Laryngoscope 2006, 116, 735–41. (49) Krop, I.; Ma¨rz, A.; Carlsson, H.; Li, X.; Bloushtain-Qimron, N.; Hu, M.; Gelman, R.; Sabel, M. S.; Schnitt, S.; Ramaswamy, S.; Kleer, C. G.; Enerba¨ck, C.; Polyak, K. A putative role for psoriasin in breast tumor progression. Cancer Res. 2005, 65, 11326–34. (50) Zhou, G.; Xie, T. X.; Zhao, M.; Jasser, S. A.; Younes, M. N.; Sano, D.; Lin, J.; Kupferman, M. E.; Santillan, A. A.; Patel, V.; Gutkind, J. S.; Ei-Naggar, A. K.; Emberley, E. D.; Watson, P. H.; Matsuzawa, S. I.; Reed, J. C.; Myers, J. N. Reciprocal negative regulation between S100A7/psoriasin and beta-catenin signaling plays an important role in tumor progression of squamous cell carcinoma of oral cavity. Oncogene 2008, 27, 3527–38.

PR800501J

Journal of Proteome Research • Vol. 8, No. 1, 2009 309