Proteomic Identication of Salivary Biomarkers of Type-2 Diabetes

Paturi V. Rao, Ashok P. Reddy, Xinfang Lu, Surendra Dasari, Adiraju Krishnaprasad, Evan Biggs, Charles T. Roberts, Jr., and Srinivasa R. Nagalla*,
Departments of Endocrinology and Metabolism and Medicine, Nizams Institute of Medical Sciences University, Hyderabad 500 082, India, and DiabetOmics, LLC, Beaverton, Oregon 97006
Received May 23, 2008

The identication of biomarkers to noninvasively detect prediabetes/diabetes will facilitate interventions designed to prevent or delay progression to frank diabetes and its attendant complications. The purpose of this study was to characterize the human salivary proteome in type-2 diabetes to identify potential biomarkers of diabetes. Whole saliva from control and type-2 diabetic individuals was characterized by multidimensional liquid chromatography/tandem mass spectrometry (2D-LC-MS/MS). Label-free quantication was used to identify differentially abundant protein biomarkers. Selected potential biomarkers were then independently validated in saliva from control, diabetic, and prediabetic subjects by Western immunoblotting and ELISA. Characterization of the salivary proteome identied a total of 487 unique proteins. Approximately 33% of these have not been previously reported in human saliva. Of these, 65 demonstrated a greater than 2-fold difference in abundance between control and type-2 diabetes samples. A majority of the differentially abundant proteins belong to pathways regulating metabolism and immune response. Independent validation of a subset of potential biomarkers utilizing immunodetection conrmed their differential expression in type-2 diabetes, and analysis of prediabetic samples demonstrated a trend of relative increase in their abundance with progression from the prediabetic to the diabetic state. This comprehensive proteomic analysis of the human salivary proteome in type-2 diabetes provides the rst global view of potential mechanisms perturbed in diabetic saliva and their utility in detection and monitoring of diabetes. Further characterization of these markers in a larger cohort of subjects may provide the basis for new, noninvasive tests for diabetes screening, detection, and monitoring.
Keywords: Human saliva proteome diabetes prediabetes preclinical diagnosis

Introduction
The twin epidemics of obesity and diabetes threaten to overwhelm healthcare systems in the U.S. and worldwide. Over the last several years, type-2 diabetes mellitus has reached almost epidemic proportions, with about 18 million affected individuals in the U.S. alone. Improved detection techniques and biomarkers are urgently needed across the entire spectrum of diabetes initiation and progression. Since 7% of newly diagnosed type-2 diabetes patients in the U.S. have been diabetic for approximately 4-7 years before diagnosis,1 the ability to ascertain those individuals at risk for the development of clinically apparent diabetes is critical to effectively focus potentially limited clinical resources. In particular, it is desirable to screen and start treating glucose-intolerant individuals as early as possible, since, even before the onset of diabetes,
* To whom correspondence should be addressed. Srinivasa R. Nagalla, MD, DiabetOmics, LLC, 20,000 NW Walker Road, Beaverton, OR 97006. E-mail, nagallasr@diabetomics.com; tel,(503) 686-0138; fax, (503) 628-1360. Department of Endocrinology and Metabolism, Nizams Institute of Medical Sciences University. Departments of Endocrinology and Metabolism and Medicine, Nizams Institute of Medical Sciences University. DiabetOmics, LLC. 10.1021/pr8003776 CCC: $40.75 2009 American Chemical Society

vascular lesions gradually develop with deterioration of glucose tolerance. Additionally, -cell function is seriously compromised by the time that overt alterations in glucose homeostasis, such as impaired glucose tolerance (IGT) and impaired fasting glucose (IFG), are manifest; thus, timely intervention is important to maintain residual insulin secretory capacity. The determination of IGT and IFG is itself an issue due to the relatively invasive nature of these assessments, particularly that of IGT by an oral glucose tolerance test. An important additional diagnostic problem is monitoring of glucose homeostasis for conrming diabetes. Compliance with glucose monitoring is poor because of the pain and inconvenience of conventional blood collection using lancets. The present study sought to identify alternative, noninvasive biomarkers with the potential for both preclinical and clinical diabetes screening and assessment. We have previously reported proteomic analysis of differentially abundant urinary proteins in type-2 diabetes patients with diabetic nephropathy.2 Since analysis of saliva has the advantages of being safe, easy to access, and noninvasive, salivary diagnostics will be increasingly important in monitoring pre- and established diabetes and its complications. Previous studies have described the salivary proteome per se,3,4
Journal of Proteome Research 2009, 8, 239245 239
Published on Web 01/02/2009

research articles
Table 1. Clinical Characteristics of Controls and Subjects with Preclinical and Type-2 Diabetes
control IGT
a

Rao et al.
IFG + IGT

DM

Age (year) Duration of DM (year) Height (cm) Weight (kg) BMI (kg/ m2) Waist (cm) Hip (cm) Waist-to-hip ratio Blood pressure, systolic (mmHg) diastolic (mmHg) Plasma glucose, fasting (mg/dL) 2 h after oral glucose (mg/dL) Serum cholesterol (mg/dL) Triglycerides (mg/dL) HDL cholesterol (mg/dL)
a

46.2 ( 11.32 NA 163 ( 13.42 73.66 ( 17.74 27.57 ( 4.47 93.4 ( 13.2 99.4 ( 6.84 0.94 ( 0.08 135.4 ( 29.8 84 ( 9.67 86 ( 8.57 113.4 ( 21.03 201.8 ( 13.97 116.6 ( 36.69 44.6 ( 21.27

43.8 ( 13.18 NA 167.8 ( 7.53 72.36 ( 8.44 25.81 ( 3.64 92.2 ( 6.83 95.2 ( 6.53 0.97 ( 0.09 135.6 ( 27.25 89 ( 11.18 93.6 ( 6.69 154.8 ( 19.23 215.2 ( 42.76 147.8 ( 54.27 39.4 ( 6.27

48.8 ( 8.23 NA 166.4 ( 4.04 71.7 ( 6.51 25.9 ( 1.96 94.8 ( 6.61 96.8 ( 6.02 0.98 ( 0.02 139.4 ( 17.46 89.4 ( 9.61 106.4 ( 7.44 159.8 ( 16.72 153.6 ( 15.99 178.4 ( 94.11 32.4 ( 4.1

49.75 ( 7.89 4 ( 2.16 168.5 ( 6.95 71.18 ( 10.89 25.01 ( 2.76 88.75 ( 4.5 93.25 ( 6.02 0.95 ( 0.02 127 ( 16.95 79.25 ( 12.37 158 ( 69.66 248 ( 134.71 202.75 ( 45.32 232.25 ( 205.61 33.5 ( 7.72

Data are means ( SD. NA, not applicable; IFG, impaired fasting glucose; IGT, impaired glucose tolerance; DM, type-2 diabetes.

while other studies have described alterations in salivary dynamics5 or the differential abundance in saliva of single factors, such as matrix metalloproteinase-8 (MMP-8)6 and epidermal growth factor7 in type-2 diabetes. To date, however, there has not been a systematic assessment of potential salivary biomarkers of diabetes. In this report, we describe the human diabetic salivary proteome of 487 proteins identied with high condence, of which one-third have not been previously described. Of these, 65 proteins were characterized as being differentially abundant in saliva from patients with type-2 diabetes versus controls. A subset of these proteins was demonstrated to also be differentially abundant in patients with prediabetes (i.e., IGT and IGF + IFG). Further renement and validation of this candidate biomarker set in a larger cohort can provide the basis for the development of a rapid, convenient, and specic test for insulin secretion and function in at-risk patients.

Research Design and Methods

Subjects. From a prospective observational study, 40 subjects were chosen for salivary proteome analysis. In the study group, 10 subjects had (previously diagnosed) type-2 diabetes (DM) on specic therapy, and 10 each had impaired glucose tolerance (IGT) or both impaired fasting glucose (IFG) and IGT. The diagnosis of DM and prediabetes (IFG and IGT) was based on American Diabetes Association criteria. IFG was diagnosed if the fasting plasma glucose level was elevated (between 100 and 125 mg/dl after an overnight fast), and IGT if the 2-h plasma glucose level was elevated (between 140 and 199 mg/dl) after an oral glucose tolerance test. The control (CTRL) group consisted of 10 clinically healthy individuals between 36 and 62 years of age. Exclusion criteria for the control group included pregnancy, alcohol consumption, tobacco products (former or current), chronic medical illness, and history of any drug treatment within the previous 3 months. No subjects had evidence of gingival inammation or other oral disease. Subjects were asked not to eat, smoke, or drink (except water) for an overnight fast prior to collection of saliva samples. Their diets were similar with respect to protein content and uptake of fat and carbohydrates. The study population was comprised of a fairly homogeneous Asian Indian population, and socioeconomic status was similar for all groups (survey data). The clinical characteristics of healthy controls and study subjects are shown in Table 1; there were no statistically signicant
240 Journal of Proteome Research Vol. 8, No. 1, 2009

differences between the groups for any of the characteristics listed in the Table. Informed consent was obtained from the subjects following the institutional review board guidelines for human subjects (Nizams Institute of Medical Sciences, Hyderabad, India). Sample Collection and Processing. Unstimulated saliva (20 mL) from diabetic, prediabetic, and control groups was collected. Briey, at 8 a.m. (before breakfast), the subjects were asked to rinse their mouths thoroughly with water, then to tilt their heads forward and allow saliva to ow into a sterile container for 5 min. These specimens were immediately frozen and stored at -80 C until analysis. Saliva samples (10 mL) were centrifuged at 10 000g for 20 min at 4 C to discard cellular debris and nuclei. The supernatants were transferred to 4-mL Ultrafree 5K membrane concentrators (Millipore Corp., Billerica, MA) and spun at 7000g to reduce the volumes to 1 mL. Individual samples were used for preliminary analyses by one- and two-dimensional gel electrophoresis, and for accurate quantication of the biomarkers using immunoassays. Five saliva samples each from the control and DM groups were pooled separately and subjected to two-dimensional liquid chromatography (2-DLC) and LC-tandem mass spectrometry (LC-MS/MS) analysis as previously described8 and briey outlined below. 2-DLC Sample Processing. Following protein assay, 1-mg portions of samples were reduced, alkylated, and digested with trypsin, and the resulting peptides were separated with strong cation-exchange (SCX) chromatography. SCX chromatography was performed using a 100 2.1-mm polysulfethyl A column (The Nest Group, Southborough, MA). A total of 80 fractions were collected and desalted using a 96-well Vydac C18 silica spin plate (The Nest Group). The desalted fractions were consolidated into 43 fractions, dried, and dissolved in 20 L of 5% formic acid for LC-MS/MS analysis. LC-MS/MS Analysis. Portions of each fraction (9 L) were analyzed by LC-MS/MS using an Agilent 1100 series capillary LC system and an LTQ ion-trap mass spectrometer (Thermo Electron, San Jose, CA) with an Ion Max electrospray source (ThermoFinnigan, San Jose, CA). Samples were applied at 20 L/min to a trap cartridge, and then switched onto a 0.5 250-mm Zorbax SB-C18 column (Agilent Technologies, Palo Alto, CA) using mobile phase A containing 0.1% formic acid. Survey MS scans were alternated with three data-dependent MS/MS scans using the dynamic exclusion feature of the

Salivary Proteome in Diabetes

control software to increase the number of unique peptides analyzed. Mass spectra les were generated using Bioworks Browser software (version 3.1, ThermoFinnigan, San Jose, CA) with m/z range of 400-4000 Da, a minimum of 15 ions, and a low TIC threshold of 500. A total of 1 729 998 tandem mass spectra were generated from all LC-MS/MS analyses. Peptide and Protein Identication. Tandem mass spectra were searched against a composite protein database containing forward and reversed entries (decoy proteins) of Swiss-Prot (version 52.1) and TrEmbl (version 35.1) databases selected for human subspecies. Splice variants were generated from the human subset of the Swiss-Prot (version 52.1) database using the varsplice program from SwissKnife package (version 1.62). Forward and reverse entries of the generated splice variants were also added to the composite protein database. All searches were performed using the X! Tandem9,10 search engine congured to use 1.8 and 0.4 Da as parent and fragment ion mass tolerances, respectively, and trypsin enzyme specicity while deriving peptide candidates from the database. X! Tandem was congured to search with a xed carbamidomethyl modication on cysteine residues and several potential in vivo modications. Peptide identications from samples were assembled into proteins using probabilistic protein identication algorithms11 implemented in Scaffold software (version 1.6, Proteome Software, Portland, OR). Peptide and protein identications in all samples were compiled together to generate a comprehensive salivary proteome. Proteins with one or more unique peptide identications (p g 0.8) were considered as likely to be present in the sample. Protein entries were further curated to reduce redundancy by removing subset proteins and collapsing degenerate protein identications into a single entry. All immunoglobulin variants identied in the sample were also collapsed into a single entry. Annotations of identied hypothetical sequences were corrected, if possible, by checking their sequence homology with known proteins in the Swiss-Prot human database (version 52.1) using NCBI BLAST software. A hypothetical annotation was accepted into the nal protein list if the corresponding blast search did not turn up any homologous (g90% sequence homology) known human proteins. Protein identications with at least three unique peptide identications in at least one sample were considered to be present in saliva (Supplemental Table 1 in Supporting Information). Label-Free Quantication. The total number of tandem mass spectra matched to a protein (spectral counting) is a labelfree, sensitive, and semiquantitative measure for estimating its abundance in complex mixtures.12-14 The spectral count difference between two complex samples is used to quantify the relative expression of a protein.8,15-18 In this study, salivary proteins with at least three unique peptide identications in at least one sample were considered for label-free quantication. Shared spectral counts of nondegenerate proteins belonging to the same family with signicant sequence homology (>50%) were combined into single entry. Shared spectral counts of nondegenerate proteins that did not t the aforementioned criteria were assigned to one of the proteins using Occams razor approach. Curated proteins were subjected to independent pairwise comparisons to determine differentially abundant proteins between control and diabetes groups using either a 2 2 2 or Fishers exact test as described earlier.17 Normalization of spectral counts to account for experimental variability was built into the pairwise comparison model automated using a SAS program (version 9.1). A protein was

research articles
considered as signicantly differentially abundant between the samples if the comparison had a p-value of e0.05 in either the 2 or Fishers exact test. The fold change (FC) in the level of differentially abundant proteins was calculated using the equation described elsewhere.12 Western Immunoblotting and Enzyme-Linked Immunosorbent Assay (ELISA). For Western blot analyses, saliva from ve individual patient samples from each experimental group (CTRL, IGT, IGT + IFG, and DM) was pooled, and 15 g of protein was resolved on 10-20% Tris-Tricine gels and transferred to PVDF membranes. Membranes were blocked with 5% fat-free milk in TBST for 2 h at room temperature and incubated with primary antibody (alpha-1-antitrypsin, cystatin C, alpha-2-macroglobulin, and transthyretin from Dako North America, Inc., Carpinteria, CA; and salivary alpha amylase from Sigma, St. Louis, MO) for 1 h. All primary antibodies were diluted 1:4000, except alpha-1-antitrypsin (A1AT), which was diluted 1:5000. After three 10-min washes with TBST, membranes were incubated with anti-rabbit IgG-HRP secondary antibody (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) for 1 h, washed and visualized with the SuperSignal West Pico chemiluminescent substrate system (Pierce, Rockford, IL). Chemiluminescence was scanned on a LAS-3000 instrument using LAS-3000 Lite software, and scanned images were visualized and quantied using MultiGauge v.3.1 software (FujiFilm Life Science, Inc., Stamford, CT). ELISAs using a microtiter plate assay were performed individually on samples from 5 subjects in each group chosen randomly and matched across the groups. Primary antibodies, secondary antibodies, and reference proteins were obtained from Dako North America, Inc. A standard curve was generated by four-parameter curve-tting using SoftmaxPro V 1.11 software, (Molecular Devices Corp., Sunnyvale, CA). The concentrations of the individual samples were estimated from the average values of triplicates in comparison to the standard curve. Concentrations of individual biomarkers are expressed as means ( SEM. Statistical signicance was estimated by Kruskal-Wallis nonparametric ANOVA for 4-group comparison and the Wilcoxin two-sample test for pairwise comparisons.

Results
Human Diabetic Salivary Proteome. A total of 2172 proteins were identied at a single unique-peptide (p g 0.8) threshold. To reduce the false-positive rate, a stringent three-uniquepeptide threshold was adapted, which resulted in 586 identications, with a false-positive rate of 0.5%. The protein list was further curated by collapsing all the immunoglobulin entries into a single entry, and sample processing artifacts (such as trypsin and keratin) and decoy proteins were removed. The resulting 487 proteins of the salivary proteome and their corresponding ranking in control and diabetes subjects are shown in Supplemental Table 1 in Supporting Information. The salivary proteomes from this study and the current literature are cross-referenced and marked accordingly. A total of 315 (67%) proteins found in this study have been conrmed by other studies. Thus, 33% of the salivary proteins identied in this study are novel. The type-2 diabetes salivary proteome was functionally annotated using GO annotations from DAVID and BioHarvester informatics resources as shown in Figure 1. A majority of the salivary proteins have metabolic (42%) and immune response (11%) functions. Proteins with other cellular functions included such as cell organization and biogenesis (11%), cell comJournal of Proteome Research Vol. 8, No. 1, 2009 241

research articles
Discussion

Rao et al.

Diabetes is a major problem worldwide and is a leading cause of morbidity that is attributable to largely preventable metabolic complications. To date, however, no robust marker of diabetes or its vascular complications has been validated for general clinical use. With the emergence of diseasemodication drugs in diabetes, there is an increasing need for diagnostic markers to ensure that these therapies are targeted to the correct patient population. Saliva has multiple advantages as a diagnostic body uid due to its noninvasive, safe, simple, and cost-effective nature. With the use of a comprehensive and rigorous proteomic approach comprised of 2-DLC fractionation, LC-MS/MS identication, and spectral counting quantication, we characterized 487 proteins in human whole saliva and identied 65 that were signicantly different in relative abundance between controls and DM patients. Perhaps not surprisingly, a majority of the differently abundant proteins are predicted to have functions in metabolism, followed by the functional categories of development, cell organization and biogenesis, immune function, cell communication and proliferation, and apoptosis. Potential Role of Inammatory and Immunomodulators as Predictors of Prediabetes and Diabetes. Among the upregulated proteins identied in this study, a signicant number are associated with immune function and have been previously reported to be associated with diabetes in other body uids. The presence of inammatory factors among this biomarker set is consistent with the proposed role of a chronic subclinical inammatory state in the genesis of the metabolic syndrome and diabetes.19,20 The protease inhibitors CysC, leukocyte elastase inhibitor (LEI), and uteroglobin have known diabetes associations. CysC is a potent inhibitor of lysosomal cysteine proteinases. In serum, it is a known marker for glomerular ltration and has been shown to be signicantly elevated in cardiovascular disease and diabetes.21 LEI regulates the activity of neutrophil proteases, including polymorphonuclear elastase (PMN-E). In human plasma, PMN-E is a marker for hypertension and micro- and macrovascular disease in type-2 diabetes.22 Human uteroglobin, also called blastokinin, is secreted from Clara cells, and in human urine, is an indicator of renal tubular function in diabetes.23 Neutrophil collagenase, or MMP-8, has been linked to local tissue damage in saliva from patients with uncontrolled type-2 diabetes.6 However, serum MMP-8 levels were reported to be elevated in coronary heart disease independent of known risk factors, including diabetes.24 Thus, upregulated MMP-8 in human saliva in type-2 diabetes could be a nonspecic inammatory marker. TTR, or prealbumin, is a known transport protein for both thyroxine and retinol (vitamin A). Stockholm Diabetes Prevention Program investigators have reported both up- and down-regulation of serum TTR, depending on the particular patient cohort studied.25 Transport and Metabolic Indicators in Early Detection of Diabetes. A1AT, A2MG, and plasma retinol binding protein 4 (RBP4) were elevated in diabetic saliva and have been previously implicated in diabetes. A1AT, potentially through its demonstrated antiapoptotic activity, has also been shown to prevent or reverse diabetes, prevent the development of type-1 diabetes in mice, prolong islet allograft survival in rodents, and reduce beta-cell apoptosis in vitro.26 We have earlier reported the overexpression of A1AT in the urine of patients with diabetic nephropathy.2 RBP4 is a novel adipokine of the lipocalin family involved in the development of obesity

Figure 1. Functional annotation of human whole saliva in type-2 diabetes. The majority of the proteins are predicted to have metabolic and immune response functions.

munication and proliferation (6%), development (5%), and apoptosis (5%). Quantication of Diabetic Salivary Proteome Using Spectral Counts. The spectral counts of the salivary proteins were subjected to label-free quantication to nd differentially abundant proteins between the control and DM groups. Proteins with a relative differential abundance of g2.0-fold and which passed the label-free quantication with a p-value of e0.05 were considered as signicantly differentially abundant between the two groups. A total of 65 differentially abundant salivary proteins are shown in Table 2 grouped according to their functional annotations. Validation of Potential Biomarkers Using Immunodetection. To independently conrm the relative abundance of salivary proteins identied by 2D-LC-MS analysis, we performed Western blotting and ELISA. To further explore their potential differential presence in prediabetes, we tested samples from the IGT and IGT + IFG groups in addition to DM. As shown in Figure 2, the levels of A1AT, cystatin C (CysC), alpha-2macroglobulin (A2MG), and transthyretin (TTR) were elevated in DM, which correlated with the results of spectral counting. Salivary alpha amylase, which was not regulated in DM, was used as control protein. Scanning densitometry analysis showed the relative levels of A2MG as 1.4-, 1.9-, and 2.1-fold; A1AT as 1.4-, 2.0-, and 2.5-fold; and TTR as 1.26-, 1.38-, and 1.68-fold higher in the IGT, IGT + IFG, and DM groups, respectively, compared to control ()1.0). As also shown in Figure 2, A1AT, A2MG, and TTR showed a relative increase in abundance with disease progression. In contrast, the relative levels of CysC were 1.54-, 1.28-, and 1.29-fold higher in the IGT, IGT+IFG, and type-2 diabetes groups, respectively, compared to control; i.e., highest in the IGT group. Immunoassay measurements of A2MG on individual subjects correlated with the Western blot data and their differential abundance based upon label-free quantication (Figure 2, bottom panel). There were signicant differences in A2MG concentrations in the 4 groups when compared via KruskalWallis nonparametric ANOVA (p ) .0186 for the 4-group comparison). In pairwise comparisons, the group signicantly different from control was the DM group (p ) 0.0137 via Wilcoxin two-sample test). These data suggest that the differential abundance of specic salivary proteins in DM saliva is potentially presaged by their differential abundance in prediabetes.
242 Journal of Proteome Research Vol. 8, No. 1, 2009

Salivary Proteome in Diabetes

research articles
diabetes vs control

Table 2. Salivary Proteins Showing Differential Abundance in Subjects with Type-2 Diabetes and Controlsa

function

Swiss-Prot accession

protein name

fold change

p-value

Metabolism

Immune response

Development

Extracellular matrix protein

Signal transduction

Cell organization and biogenesis

P23280 P14618 P06737 Q549C7 P22894 P00491 O60235 P30838 Q13231-3 Q9UBR2 P00558 O60218 Q13787 P00915 P00918 Q86U62 P27824 Q6FHH3 Q4VAX6 Q9NP55 P13671 P01009 P01034 P30740 P01040 P04083 Q4VB24 Q09666 Q9NZT1 Q01469 Q06830 Q96RM1 P31151 Q5TCI8 P07355 P15924 P30043 P07998 A2RTY6 P19827 P36222 Q14624 P80303 Q9UKR3 O43240 Q7M4Q5 P39687 Q5VY30 P23528 P62258 P12429 Q04917 O15511 P60953-2 P01023 P28676 P61160 P26038 O95274 P67936-2

Cell motility

Carbonic anhydrase 6 Pyruvate kinase isozymes M1/M2 Glycogen phosphorylase, liver form Transthyretin Neutrophil collagenase Purine nucleoside phosphorylase Transmembrane protease, serine 11D Aldehyde dehydrogenase, dimeric NADP-preferring Isoform 2, 3 and 4 of Chitotriosidase-1 Cathepsin Z Phosphoglycerate kinase 1 Aldo-keto reductase family 1 member B10 Apolipoprotein B-100 Carbonic anhydrase 1 Carbonic anhydrase 2 Proteasome (prosome, macropain) subunit, beta type, 7 Calnexin Uteroglobin Serpin peptidase inhibitor, clade B Protein Plunc Complement component C6 Alpha-1-antitrypsin Cystatin-C Leukocyte elastase inhibitor Cystatin-A Annexin A1 Histone cluster 1, H1e Neuroblast differentiation-associated protein AHNAK Calmodulin-like protein 5 Fatty acid-binding protein, epidermal Peroxiredoxin-1, -2 and -6 Small proline-rich protein 2F Protein S100-A7 Lamin A/C Annexin A2 Desmoplakin Flavin reductase Ribonuclease pancreatic Interalpha (Globulin) inhibitor H2 Interalpha-trypsin inhibitor heavy chain H1 Chitinase-3-like protein 1 Interalpha-trypsin inhibitor heavy chain H4 Nucleobindin-2 Kallikrein-13 Kallikrein-10 Basic proline-rich peptide IB-8a Acidic leucine-rich nuclear phosphoprotein 32 family Retinol binding protein 4, plasma Colin-1 14-3-3 protein epsilon Annexin A3 14-3-3 protein eta Actin-related protein 2/3 complex subunit 5 Isoform 2 of P60953 Cell division control protein 42 homologue precursor Alpha-2-macroglobulin Grancalcin Actin-like protein 2 Moesin Ly6/PLAUR domain-containing protein 3 Isoform 2 of P67936 Tropomyosin alpha-4 chain

3.84 3.47 3.32 2.4 2.36 -2.08 -2.13 -2.19 -2.2 -2.85 -3.18 -3.32 -4.13 -4.36 -5.54 -6.11 -7.74 10.43 6.05 5.48 4.75 3.24 2.22 2.03 -2.42 -3.57 6.05 3.08 -2.17 -2.55 -2.59 -2.85 -2.94 -3.26 -4.25 -5.88 -6.11 3.78 3.16 2.8 2.65 2.59 2.05 -4.48 -4.99 5.4 3.32 2.15 2.11 -2.25 -2.68 -2.95 6.05 4.75 2.23 -7.09 3.36 2.04 -2.3 -3.75

<0.0001 0.0002 0.0105 0.0246 0.0039 0.0032 0.012 0.0034 0.0263 0.0361 <0.0001 0.0127 <0.0001 <0.0001 0.0002 0.0184 0.0005 <0.0001 0.0101 <0.0001 0.036 <0.0001 0.0007 0.011 0.0042 <0.0001 0.0101 0.0472 0.0151 <0.0001 <0.0001 0.0361 0.003 <0.0001 0.0014 <0.0001 0.0003 0.0015 0.0102 0.0042 0.0173 0.006 0.005 0.0265 0.0024 0.019 0.0105 0.0143 0.0464 0.01 0.008 0.0438 0.0101 0.036 <0.0001 0.0083 0.0476 0.0006 0.0236 0.0006

a Spectral counts of human salivary proteins with 3 or more unique peptide identications were subjected to label-free quantication. Proteins that were signicantly differentially abundant (p-value e0.05) by at least (2.0-fold are shown above. Proteins are grouped according to their function. Fold change between the groups was quantied using equation described by Old et al.12

and insulin resistance.27 Its abundance in human saliva in the present study was consistent with elevated RBP4 postulated as a plasma biomarker of early metabolic disturbances associ-

ated with DM and for related nephropathy and cardiovascular disease.28 A2MG variations in diabetes were reported with protease-antiprotease imbalance in children who were at
Journal of Proteome Research Vol. 8, No. 1, 2009 243

research articles

Rao et al.
in association with progression of prediabetes to the diabetic state underscores the importance of a systematic analysis of these candidate biomarkers in prediabetic saliva, as well as their variability in individual samples, by immunoassays. As recent studies have shown that early and multifactorial intervention in DM prevents cardiovascular complications and mortality,35 advances in understanding molecular aspects of preclinical diabetes will further facilitate accurate diagnosis and early intervention.

Acknowledgment. The authors thank their colleagues at Nizams Institute of Medical Sciences and Dr. Rakesh Mittal, Senior Deputy Director General, Indian Council of Medical Research, for their support. Supporting Information Available: Salivary_proteome_ diabetes_suppl_table_1.xls, list of 487 proteins of the salivary proteome and their corresponding ranking in control and diabetes subjects; Salivary_proteome_diabetes_suppl_gure_1.tif, gure of salivary proteome. This material is available free of charge via the Internet at http://pubs.acs.org. References
(1) Harris, M. I.; Klein, R.; Welborn, T. A.; Knuiman, M. W. Onset of NIDDM occurs at least 4-7 yr before clinical diagnosis. Diabetes Care 1992, 15 (7), 8159. (2) Rao, P. V.; Lu, X.; Standley, M.; Pattee, P.; Neelima, G.; Girisesh, G.; Dakshinamurthy, K. V.; Roberts, C. T., Jr.; Nagalla, S. R. Proteomic identication of urinary biomarkers of diabetic nephropathy. Diabetes Care 2007, 30 (3), 62937. (3) Denny, P.; Hagen, F. K.; Hardt, M.; Liao, L.; Yan, W.; Arellanno, M.; Bassilian, S.; Bedi, G. S.; Boontheung, P.; Cociorva, D.; Delahunty, C. M.; Denny, T.; Dunsmore, J.; Faull, K. F.; Gilligan, J.; Gonzalez-Begne, M.; Halgand, F.; Hall, S. C.; Han, X.; Henson, B.; Hewel, J.; Hu, S.; Jeffrey, S.; Jiang, J.; Loo, J. A.; Ogorzalek Loo, R. R.; Malamud, D.; Melvin, J. E.; Miroshnychenko, O.; Navazesh, M.; Niles, R.; Park, S. K.; Prakobphol, A.; Ramachandran, P.; Richert, M.; Robinson, S.; Sondej, M.; Souda, P.; Sullivan, M. A.; Takashima, J.; Than, S.; Wang, J.; Whitelegge, J. P.; Witkowska, H. E.; Wolinsky, L.; Xie, Y.; Xu, T.; Yu, W.; Ytterberg, J.; Wong, D. T.; Yates, J. R.; Fisher, S. J. The proteomes of human parotid and submandibular/ sublingual gland salivas collected as the ductal secretions. J. Proteome Res. 2008, 7 (5), 19942006. (4) Hu, S.; Loo, J. A.; Wong, D. T. Human saliva proteome analysis and disease biomarker discovery. Expert Rev Proteomics 2007, 4 (4), 5318. (5) Bernardi, M. J.; Reis, A.; Loguercio, A. D.; Kehrig, R.; Leite, M. F.; Nicolau, J. Study of the buffering capacity, pH and salivary ow rate in type 2 well-controlled and poorly controlled diabetic patients. Oral Health Prev. Dent. 2007, 5 (1), 738. (6) Collin, H. L.; Sorsa, T.; Meurman, J. H.; Niskanen, L.; Salo, T.; Ronka, H.; Konttinen, Y. T.; Koivisto, A. M.; Uusitupa, M. Salivary matrix metalloproteinase (MMP-8) levels and gelatinase (MMP9) activities in patients with type 2 diabetes mellitus. J. Periodontal. Res. 2000, 35 (5), 25965. (7) Oxford, G. E.; Tayari, L.; Barfoot, M. D.; Peck, A. B.; Tanaka, Y.; Humphreys-Beher, M. G. Salivary EGF levels reduced in diabetic patients. J. Diabetes Complications 2000, 14 (3), 1405. (8) Nagalla, S. R.; Canick, J. A.; Jacob, T.; Schneider, K. A.; Reddy, A. P.; Thomas, A.; Dasari, S.; Lu, X.; Lapidus, J. A.; Lambert-Messerlian, G. M.; Gravett, M. G.; Roberts, C. T., Jr.; Luthy, D.; Malone, F. D.; DAlton, M. E. Proteomic analysis of maternal serum in down syndrome: identication of novel protein biomarkers. J. Proteome Res. 2007, 6 (4), 124557. (9) Craig, R.; Beavis, R. C. TANDEM: matching proteins with tandem mass spectra. Bioinformatics 2004, 20 (9), 14667. (10) Fenyo, D.; Beavis, R. C. A method for assessing the statistical signicance of mass spectrometry-based protein identications using general scoring schemes. Anal. Chem. 2003, 75 (4), 76874. (11) Nesvizhskii, A. I.; Keller, A.; Kolker, E.; Aebersold, R. A statistical model for identifying proteins by tandem mass spectrometry. Anal. Chem. 2003, 75 (17), 464658. (12) Old, W. M.; Meyer-Arendt, K.; Aveline-Wolf, L.; Pierce, K. G.; Mendoza, A.; Sevinsky, J. R.; Resing, K. A.; Ahn, N. G. Comparison of label-free methods for quantifying human proteins by shotgun proteomics. Mol. Cell. Proteomics 2005, 4 (10), 1487502.

Figure 2. (Top panel) Western blot analysis of alpha-2 macroglobulin (A2MG), alpha-1-antitrypsin (A1AT), cystatin C (Cys C), transthyretin (TTR), and salivary alpha-amylase (AMYS). (Bottom panel) ELISA analysis of saliva A2MG showing mean concentrations in controls (CTRL) and study subjects with impaired glucose tolerance (IGT), IGT and impaired fasting glucose (IFG), and diabetes mellitus (DM). Error bars denote SEM p ) 0.0186 for the 4-group comparison via Kruskal-Wallis nonparametric ANOVA. p ) 0.0137 for pairwise comparison between control (CTRL) and diabetes mellitus (DM) via Wilcoxin two-sample test.

greater risk of developing vascular complications.29 A cardiac isoform of A2MG has been shown to be an early marker of cardiac hypertrophy and increased left-ventricular mass in myocardial-infracted diabetic patients.30 A2MG in human saliva was reported as a proinammatory factor.31 Carbonic anhydrase I (CA-I) was down-regulated in saliva, and was reported to be decreased in erythrocytes32 and increased in vitreous of DM patients. Apolipoprotein (apo) B-100 is one of the two main forms of apoB in chylomicrons and low-density lipoproteins. Diabetic subjects are known to have elevated apoB-100 as a result of increased production and reduced fractional catabolic rates.33 One of the nine development-related salivary proteins down-regulated in type-2 diabetes was lamin A/C. Mutations in the LMNA gene are characterized by loss of subcutaneous adipose tissue, insulin resistance, dyslipidemia, and type-2 diabetes.34 Potential Biomarkers for Prediabetes. An important nding of this study is that a subset of salivary biomarkers of established DM identied by proteomic proling were also differentially abundant in the saliva of patients with IGT alone and IGT + IFG as assessed by direct Western immunoblot and ELISA analyses. The relative increase of some of these markers
244 Journal of Proteome Research Vol. 8, No. 1, 2009

Salivary Proteome in Diabetes

(13) Liu, H.; Sadygov, R. G.; Yates, J. R. A model for random sampling and estimation of relative protein abundance in shotgun proteomics. Anal. Chem. 2004, 76 (14), 4193201. (14) Zybailov, B.; Coleman, M. K.; Florens, L.; Washburn, M. P. Correlation of relative abundance ratios derived from peptide ion chromatograms and spectrum counting for quantitative proteomic analysis using stable isotope labeling. Anal. Chem. 2005, 77 (19), 621824. (15) Zybailov, B.; Mosley, A. L.; Sardiu, M. E.; Coleman, M. K.; Florens, L.; Washburn, M. P. Statistical analysis of membrane proteome expression changes in Saccharomyces cerevisiae. J. Proteome Res. 2006, 5 (9), 233947. (16) Gravett, M. G.; Thomas, A.; Schneider, K. A.; Reddy, A. P.; Dasari, S.; Jacob, T.; Lu, X.; Rodland, M.; Pereira, L.; Sadowsky, D. W.; Roberts, C. T., Jr.; Novy, M. J.; Nagalla, S. R. Proteomic analysis of cervical-vaginal uid: identication of novel biomarkers for detection of intra-amniotic infection. J. Proteome Res. 2007, 6 (1), 89 96. (17) Pereira, L.; Reddy, A. P.; Jacob, T.; Thomas, A.; Schneider, K. A.; Dasari, S.; Lapidus, J. A.; Lu, X.; Rodland, M.; Roberts, C. T., Jr.; Gravett, M. G.; Nagalla, S. R. Identication of novel protein biomarkers of preterm birth in human cervical-vaginal uid. J. Proteome Res. 2007, 6 (4), 126976. (18) Pang, J. X.; Ginanni, N.; Dongre, A. R.; Hefta, S. A.; Opitek, G. J. Biomarker discovery in urine by proteomics. J. Proteome Res. 2002, 1 (2), 1619. (19) Fernandez-Real, J. M.; Pickup, J. C. Innate immunity, insulin resistance and type 2 diabetes. Trends Endocrinol. Metab. 2008, 19 (1), 106. (20) Pickup, J. C.; Crook, M. A. Is type II diabetes mellitus a disease of the innate immune system? Diabetologia 1998, 41 (10), 12418. (21) Larsson, A.; Helmersson, J.; Hansson, L. O.; Basu, S. Serum cystatin C is associated with other cardiovascular risk markers and cardiovascular disease in elderly men. Int. J. Cardiol. 2008, 125 (2), 2634. (22) Piwowar, A.; Knapik-Kordecka, M.; Warwas, M. Concentration of leukocyte elastase in plasma and polymorphonuclear neutrophil extracts in type 2 diabetes. Clin. Chem. Lab. Med. 2000, 38 (12), 125761. (23) Hong, C. Y.; Chia, K. S. Markers of diabetic nephropathy. J. Diabetes Complications 1998, 12 (1), 4360. (24) Qiang, H.; Zhou, Z. X.; Ma, A. Q.; Cheng, H.; Zhou, P. [Implications of serum matrix metalloproteinase-8 elevation in patients with acute coronary syndrome]. Nanfang Yike Daxue Xuebao 2007, 27 (6), 8313.

research articles
(25) Sundsten, T.; Ostenson, C. G.; Bergsten, P. Serum protein patterns in newly diagnosed type 2 diabetes mellitus-inuence of diabetic environment and family history of diabetes. Diabetes Metab. Res. Rev. 2008, 24 (2), 14854. (26) Zhang, B.; Lu, Y.; Campbell-Thompson, M.; Spencer, T.; Wasserfall, C.; Atkinson, M.; Song, S. Alpha1-antitrypsin protects beta-cells from apoptosis. Diabetes 2007, 56 (5), 131623. (27) Graham, T. E.; Yang, Q.; Bluher, M.; Hammarstedt, A.; Ciaraldi, T. P.; Henry, R. R.; Wason, C. J.; Oberbach, A.; Jansson, P. A.; Smith, U.; Kahn, B. B. Retinol-binding protein 4 and insulin resistance in lean, obese, and diabetic subjects. N. Engl. J. Med. 2006, 354 (24), 255263. (28) Cabre, A.; Lazaro, I.; Girona, J.; Manzanares, J.; Marimon, F.; Plana, N.; Heras, M.; Masana, L. Retinol-binding protein 4 as a plasma biomarker of renal dysfunction and cardiovascular disease in type 2 diabetes. J. Intern. Med. 2007, 262 (4), 496503. (29) Lisowska-Myjak, B.; Pachecka, J.; Kaczynska, B.; Miszkurka, G.; Kadziela, K. Serum protease inhibitor concentrations and total antitrypsin activity in diabetic and non-diabetic children during adolescence. Acta Diabetol. 2006, 43 (4), 8892. (30) Annapoorani, P.; Dhandapany, P. S.; Sadayappan, S.; Ramasamy, S.; Rathinavel, A.; Selvam, G. S. Cardiac isoform of alpha-2 macroglobulin-a new biomarker for myocardial infarcted diabetic patients. Atherosclerosis 2006, 186 (1), 1736. (31) Aurer, A.; Jorgic-Srdjak, K.; Plancak, D.; Stavljenic-Rukavina, A.; Aurer-Kozelj, J. Proinammatory factors in saliva as possible markers for periodontal disease. Coll. Antropol. 2005, 29 (2), 435 9. (32) Gambhir, K. K.; Ornasir, J.; Headings, V.; Bonar, A. Decreased total carbonic anhydrase esterase activity and decreased levels of carbonic anhydrase 1 isozyme in erythrocytes of type II diabetic patients. Biochem. Genet. 2007, 45 (5-6), 4319. (33) Hogue, J. C.; Lamarche, B.; Tremblay, A. J.; Bergeron, J.; Gagne, C.; Couture, P. Evidence of increased secretion of apolipoprotein B-48-containing lipoproteins in subjects with type 2 diabetes. J. Lipid Res. 2007, 48 (6), 133642. (34) Hegele, R. A. Phenomics, lamin A/C, and metabolic disease. J. Clin. Endocrinol. Metab. 2007, 92 (12), 45668. (35) Gaede, P.; Lund-Andersen, H.; Parving, H. H.; Pedersen, O. Effect of a multifactorial intervention on mortality in type 2 diabetes. N. Engl. J. Med. 2008, 358 (6), 58091.

PR8003776

Journal of Proteome Research Vol. 8, No. 1, 2009 245