Saliva proteomic analysis reveals possible biomarkers of renal cell carcinoma

Abstract Early diagnosis is a key to improve the prognosis of renal cell carcinoma (RCC); however, reliable RCC biomarkers are lacking in clinical practice. In this study, we used isobaric tags for relative and absolute quantification-based mass spectrometry to identify salivary proteins as biomarkers for the diagnosis of RCC. The objective of this study is to discover biomarkers from saliva by utilizing high-throughput quantitative proteomics approaches. Saliva proteins from 124 RCC patients and healthy individuals were identified and quantified. RCC putative biomarkers were verified by real-time polymerase chain reaction or enzyme-linked immunosorbent assay in a prevalidation sample set. Seventy-one differentially expressed salivary proteins were identified. Serotransferrin, haptoglobin, KRT9, and S100A9, which in previous studies were found to be most closely related to cancers, were selected as putative RCC biomarkers. Haptoglobin and S100A9 were significantly elevated in RCC compared with healthy control samples, although the expression of serotransferrin and KRT9 did not differ between the groups. Furthermore, receiver operating characteristic curves with a cut-off value of 75.49 ng/mL for S100A9 revealed a sensitivity of 87.10% and a specificity of 91.94% for discriminating RCC patients from healthy individuals. Salivary haptoglobin differentiated RCC patients from healthy controls with a sensitivity of 85.48% and specificity of 80.65% (cut-off value 43.02 µg/mL). These results provide experimental evidence to support S100A9 and haptoglobin as potential novel, noninvasive biomarkers for the diagnosis of RCC.


Introduction
Renal cell carcinoma (RCC) is a significant public health issue worldwide. According to the National Central Cancer Registry of China 2015 annual report, the overall incidence of RCC was 3.35 per 10 5 population [1]. Currently, RCC detection relies on imaging technologies and traditional disease manifestations. However, as RCC often presents with few signs, symptoms, or laboratory abnormalities, the disease is frequently diagnosed at an advanced stage, when the prospect for a cure is limited. Therefore, the identification of efficient biomarkers for detecting RCC in clinical practice is the key to improving treatment and survival.
Using saliva as a diagnostic tool to monitor an individual's health or disease is an appealing objective for healthcare research. Saliva is one of the most complex and important body fluids, containing water, minerals, nucleic acid, electrolytes, and proteins and reflecting a large range of physiological information [2]. Indeed, the research has shown that saliva can be used for the diagnosis of clinical diseases due to its origin, composition, function, and interaction with substances that compose plasma [3]. With the advantages of simple, noninvasive, and cost-effective methodologies, saliva-based diagnostics can be used for laboratory investigations, including basic and clinical studies of oral to systemic diseases [4]. Proteomics, a relatively new platform for research into tumor hallmarks, facilitates highthroughput screening processes in clinical diagnosis. Thus, saliva proteomics has been used extensively to detect altered expression of various substances, with varying degrees of sensitivity and specificity in a range of cancers.
The aim of the present study is to investigate and identify potential biomarkers in human saliva to facilitate the diagnosis of RCC at early and intermediate stages.
We employed the isobaric tag for the relative and absolute quantification (iTRAQ) technique to compare salivary protein profiles between RCC patients and healthy subjects. Candidate protein biomarkers were selected from the iTRAQ results, and further clinical validation was performed by the real-time polymerase chain reaction (PCR) and the enzyme-linked immunosorbent assay (ELISA) using an independent sample cohort. The functions of putative biomarkers in various cancer types are only minimally understood, and the application of salivary analysis for RCC remains unclear.
2 Material and methods

Patients and sample collection
This trial was conducted and approved by the Committee on Human Research of the Second People's Hospital of ShenZhen. A total of 62 patients with RCC and 62 ageand sex-matched healthy individuals were enrolled. Patients diagnosed with another malignancy, undergoing chemotherapy or radiotherapy, or with a cardiovascular disorder, immunodeficiency, autoimmune disorder, hepatitis or human immunodeficiency virus (HIV) infection were excluded. The characteristics of the RCC patients with respect to age, sex, histopathological type, and tumor size and stage were collected for analysis.

Sample collection
Unstimulated whole saliva was collected from each consenting participant, essentially as described previously [5]. Briefly, participants were asked to refrain from eating, drinking, or brushing teeth for at least 8 h before collection at night. Whole saliva was collected in the next morning and centrifuged at 18,000 rpm for 10 min at 4°C. The supernatant was removed, treated with protease inhibitor cocktail (10 µL/mL) (Solarbio, Beijing, China) to minimize protein degradation, and stored at −80°C.

iTRAQ analysis
The total protein concentration of the soluble fraction of saliva was quantified using the bicinchoninic acid (BCA) assay (Thermo Fish Scientific). Proteins in each sample (100 μg) were precipitated with ice-cold acetone, and the precipitated proteins were digested with trypsin at 37°C overnight. iTRAQ reagent (Applied Biosystems, CA, USA) labeling was performed, yielding healthy control sample iTRAQ regent 113 and RCC sample iTRAQ regent 114. The samples were incubated at room temperature for 4 h and then desalted and dried. The iTRAQ-labeled peptides were fractionated by the high-performance liquid chromatography (HPLC) with the strong cation exchange (SCX) chromatography (Agilent Technology, CA, USA). The samples were subsequently loaded onto a C18 column (Waters Corporation, MA, USA). The components produced by the SCX chromatography were subjected to the mass spectrometry (MS) analysis.
The raw protein data were examined using the Mascot algorithm and subjected to a search with the human International Protein Index database (version 3.45) using ProteinPilot software (Applied Biosystems, SCIEX, MA, USA). To reduce false-positive data, a strict cut-off value for protein identification was applied with an unused ProtScore >1.3 and at least one peptide with 95% confidence limits [6,7]. A ratio of >1.5 or <0.7 for protein expression between the two groups was considered significant.
The gene ontology (GO) analysis was performed using the online program DAVID (https://david.ncifcrf. gov/). Differentially expressed proteins were classified according to their biological process, molecular function, and cellular component.

RNA isolation and quantitative reverse transcription PCR (RT-qPCR)
Total RNA was isolated from the salivary samples using the mirVana™ PARIS™ kit (Life Technologies). After treatment with DNase I, the RNA was reverse transcribed into cDNA with the Thermo Scientific Maxima First-Strand cDNA Synthesis kit for mRNA detection. Quantitative PCR was performed using the Applied Biosystems 7500 PCR instrument. Relative standard curves were constructed for S100A9, KRT9, haptoglobin, serotransferrin, and an endogenous reference (GAPDH) using serial dilutions of cDNA, and these standards were included in each run. Samples of unknown concentration were quantified relative to these standard curves. S100A9, KRT9, haptoglobin, and serotransferrin mRNA levels were normalized to that of GAPDH to account for cDNA loading differences. Each sample was analyzed in triplicate, and average values were used for the subsequent statistical analysis. Primer sequences were obtained from the primer bank (https://pga.mgh.harvard.edu/primerbank/), as follows: S100A9 sense 5′-GGT CAT AGA ACA CAT CAT GGA GG-3′ and anti-sense 5′-GGC CTG GCT TAT GGT GGT G-3′; KRT9 sense 5′-GGG GCC GAT TCA GCT CTT C-3′ and anti-sense 5′-CTA CTG GCA CTA AAA CCA CCC-3′; haptoglobin sense 5′-CAG CAC AGT CCC CGA AAA GAA-3′ and anti-sense 5′-CAG TCG CAT ACC AGG TGT CC-3′; serotransferrin sense 5′-GGT GGC AGA GTT CTA TGG GTC-3′ and anti-sense 5′-ACA GTA AAG TAA GCC TAT GGG GA-3′; and GAPDH sense 5′-ACA ACT TTG GTA TCG TGG AAG G-3′ and anti-sense 5′-GCC ATC ACG CCA CAG TTT C-3′.

ELISA
ELISA was employed to quantify salivary S100A9 and haptoglobin concentrations. The human CircuLex S100A9 ELISA kit (MBL, Nagova, Japan; detection limit of 6.55 pg/mL and dilution factor of 1:50) and the human haptoglobin ELISA kit (Abcam, Cambridge, USA; detection limit of 0.078 µg/mL and dilution factor of 1:2,000) were used. ELISA was also applied to assess saliva protein levels in the RCC (n = 62) and healthy control (n = 62) groups. ELISA was performed according to the instructions provided in each kit. Concentrations of proteins were analyzed based on a four-parameter logistic curve. The R-squared (R 2 ) values for our ELISA standard curves were typically in the range of 0.99 to 0.999.

Statistical analysis
Statistical analysis was performed using SPSS software version 19.0 and GraphPad Prism version 5.0. All variables under a normal distribution are presented as the mean ± standard deviation. Qualitative variables were analyzed with Student's t-test. A P value of less than 0.05 was considered significant. Receiver operating characteristic (ROC) curves were utilized to assess the discriminatory value of S100A9 and haptoglobin for differentiating between the two groups.

Analysis of the salivary proteome
To investigate differentially expressed proteins as potential biomarkers of RCC in saliva, we used an iTRAQ-based MS method to assess ten RCC patients and ten healthy control samples. A total of 1,467 proteins were quantified by MS. Fold changes in protein expression were calculated from the intensities of iTRAQ reporter ions between the groups. A total of 71 significantly differentially expressed proteins were identified in the saliva of RCC patients. Specifically, 36 proteins were downregulated (<0.7-fold) and 35 were upregulated (>1.5-fold) ( Table 2). Moreover, we categorized the differentially expressed proteins according to biological processes, molecular functions, and cellular components using the GO database. Regarding biological processes, 27% of the proteins are involved in cell motion or migration, whereas 36% are involved in defense responses and inflammatory defense. The most common molecular functions identified included protein binding (53%), enzyme regulation (41%), and structural constituent of the cytoskeleton (6%). In addition, 57% of the proteins are extracellularly located (Figure 1a-c). A network of protein-protein interactions is presented in Figure 1d.

Examination of S100A9 and haptoglobin as potential RCC biomarkers
Among the 71 differentially expressed salivary proteins, four well-documented proteins (S100A9, serotransferrin, haptoglobin, and KRT9) most closely associated with solid cancers in previous studies were selected as putative protein biomarkers. We first used the discovery sample set (n = 20) to confirm that the relative expression levels of these four genes were consistent with the iTRAQ results, with significant differences between cancer patients and healthy subjects based on RT-qPCR. As shown in Figure 2, S100A9 and haptoglobin were validated (≥2-fold change, P < 0.0001; n = 124) in RCC saliva samples, whereas serotransferrin and KRT9 expression did not exhibit differences between the groups. The expression patterns of the validated biomarkers were consistent with the iTRAQ results.

Prediction pattern of the validated biomarkers
To evaluate the clinical utility of S100A9 and haptoglobin in patients, we employed ELISA for further validation of individual saliva samples of RCC patients (n = 62) and healthy controls (n = 62). The saliva concentrations of S100A9 in the healthy controls and RCC patients were 50.79 ± 21.78 and 131.61 ± 56.61 ng/mL, respectively, and the concentration of haptoglobin in these two groups was 35.84 ± 9.86 and 56.25 ± 12.43 µg/mL, respectively. The saliva levels of S100A9 and haptoglobin in the RCC group were markedly increased compared with the control group. The expression trend of the validated biomarkers was consistent with the proteomics results ( Figure 3a). ROC curves were used to define the discriminatory power of the validated biomarkers between RCC patients and healthy individuals. As shown in Figure 3b and Table 3, with a cut-off value of 75.49 ng/mL, salivary S100A9 exhibited a sensitivity and specificity of 87.10% and 91.94%, respectively, in discriminating between RCC patients and normal individuals. If the cut-off value was set to 43.02 µg/mL, salivary haptoglobin differentiated RCC patients from healthy controls with a sensitivity of 85.48% and specificity of 80.65%.

Discussion
RCC accounts for 2-3% of all adult malignancies. According to the Globocan data released in 2015, the prevalence and mortality rates of RCC worldwide were 4.4 and 1.8 per 10 5 population, respectively [8]. RCC cases are frequently incidentally diagnosed during the imaging examination for nonspecific symptoms or other abdominal disease, which now accounts for half of all new RCC cases [9]. Some reports have identified that molecules such as TRAF-1 [10], Hsp27 [11], MMPs [12] or NMP-22 [13], which exhibit altered levels in serum or urine in RCC patients, may provide useful diagnostic information. However, no standardized serum or urine biomarkers are currently available for the accurate diagnosis of RCC in early and intermediate stages.
Therefore, a more efficient diagnosis technique or a biomarker is urgently needed. Serum molecules that are also present in saliva were first discovered in the 1970s [14]. Subsequently, other molecules, such as hormone-binding globulin and monoclonal gammopathy-related paraproteins, have been reported as found in human saliva, suggesting a mass transport mechanism between serum and saliva in a nonspecific and active manner. Although the exact roles of salivary biomarkers in specific systemic diseases remain unclear, changes in salivary biomacromolecule profiles during the development of cancer have been demonstrated in vivo [15].
Quantitative proteomics coupled with the bioinformatics analysis can provide novel insight into changes occurring at the molecular level in cancer. Therefore, the purpose of this work is to explore saliva-based noninvasive and cost-effective methods for the detection of RCC. Prior to their clinical use, we must first address whether identified biomarkers can be validated in different sample cohorts and also assess the robustness of their discriminatory power for RCC differentiation. In this study, we verified that S100A9 and haptoglobin proteins exhibit the discriminatory power for differentiating RCC patients from healthy controls, with sensitivities and specificities greater than 80% ( Table 3). These results strongly suggest that the abnormal expression of S100A9 and haptoglobin in saliva is closely related to RCC. S100A9 is a posttranslation-modified, calciumbinding protein with multiple ligands that is involved in the pathogenesis of inflammatory events and the   initial development of cancer cells [16][17][18][19]. According to the previous reports, S100A9 is associated with poor overall survival of non-small-cell lung cancer (NSCLC) patients [20,21]. TGF-β signaling has been shown to cooperate with S100A9 to influence cancer metastasis [22,23]; NF-κB binds to the S100A9 promoter and activates reactive oxygen species-dependent signaling pathways, protecting hepatocellular carcinoma cells from the apoptotic death [24]. S100A9 is also involved in the invasive phenotype at invasive margins of human colorectal carcinomas [25]. In breast cancer, S100A9 has been associated with high-grade disease and negative-ER and -PR status and high Ki67, p53, ERBB2, and EGFR expression [26,27]. Furthermore, S100A9 is abnormally expressed in many types of tumors, indicating that it has high value for cancer diagnosis. Our study also demonstrated that S100A9 levels were significantly increased in patients compared with healthy controls. Given that S100A9 is abnormally expressed in a variety of cancers, S100A9 in saliva may be a useful indicator for diagnosis of cancer, including RCC.
Haptoglobin is an acute-phase protein biosynthesized by the liver, and its expression is regulated by cytokines such as IL-1, IL-6 or TNF-α [28]. The level of haptoglobin is significantly increased in pathological conditions such as infection, acute inflammation, trauma, and tumors. Haptoglobin is involved in cell migration and extracellular matrix degradation, which may contribute to cancer angiogenesis or metastasis [29]. In addition, clinicopathological variables, TNM state and poor overall survival in non-small-cell lung carcinoma (NSCLC) patients have been associated with the levels of haptoglobin [30]. Recent research has further shown increased significance of codetection of haptoglobin and CYFRA21-1 for the diagnosis of lung squamous carcinoma [31]. Thus, it has been suggested that haptoglobin may be used as a new tumor marker with a high clinical application value for cancer. In our study, haptoglobin was evaluated in the saliva of RCC patients compared with healthy controls, with an area under the ROC curve of 0.909, demonstrating high diagnostic value for RCC. C-reactive protein (CRP) is a prognostic marker related to RCC diagnosis. We also analyzed the potential existence of a salivary biomarker with better discriminating power in detecting RCC than that of serum CRP. According to international criteria, the upper limit of serum CRP for a healthy individual is 10 mg/L. In the present study, 29 of 62 RCC patients showed CRP levels exceeding this limit, indicating that serum CRP had a sensitivity of 46.77% for the detection of resectable RCC ( Table 1). Compared with the healthy individuals, the RCC patients with serum CRP levels >10 mg/L showed significantly increased S100A9 and haptoglobin levels.
Our study demonstrates that patient-based saliva proteomics is a promising model for detecting biomarkers for RCC. In particular, our findings support the putative diagnostic value of S100A9 and haptoglobin in distinguishing RCC patients from normal healthy controls. However, several limitations of this study should be noted. First, the patient data were obtained from a single center. Approval for the use of a set of biomarkers for a given clinical decision relies on the results of large multicenter trials to adjust for possible biases; thus, the value of salivary proteins as biomarkers of RCC or other types of cancers (the potential salivary biomarkers of RCC was identified in this study, but the study can only assure their potential as cancer biomarkers, rather than biomarkers for the specific case of RCC) should be further evaluated in larger studies. In addition, as RCC is typically detected at an advanced stage, it is crucial to detect RCC as early as possible to improve the rate of survival. However, the clinical stages of the RCC patients in our study were not available in subcategories, and some of the cases were in intermediate and advanced stages (III + IV, 25.81%). Finally, the exact mechanism responsible for S100A9 and haptoglobin overexpression in cancer patients must be explored in further studies.

Conflict of interest:
The authors declare no conflict of interest.
Acknowledgments: The authors would like to express their gratitude for the support from the National Natural Science Foundation of China (No. 81273665) and Shenzhen City Science and Technology Development projects (JCYJ20130401113259980).