iTRAQ-Based Proteomics Identification of Serum Biomarkers of Two Chronic Hepatitis B Subtypes Diagnosed by Traditional Chinese Medicine

Background. Chronic infection with hepatitis B virus (HBV) is a leading cause of cirrhosis and hepatocellular carcinoma. By traditional Chinese medicine (TCM) pattern classification, damp heat stasis in the middle-jiao (DHSM) and liver Qi stagnation and spleen deficiency (LSSD) are two most common subtypes of CHB. Results. In this study, we employed iTRAQ proteomics technology to identify potential serum protein biomarkers in 30 LSSD-CHB and 30 DHSM-CHB patients. Of the total 842 detected proteins, 273 and 345 were differentially expressed in LSSD-CHB and DHSM-CHB patients compared to healthy controls, respectively. LSSD-CHB and DHSM-CHB shared 142 upregulated and 84 downregulated proteins, of which several proteins have been reported to be candidate biomarkers, including immunoglobulin (Ig) related proteins, complement components, apolipoproteins, heat shock proteins, insulin-like growth factor binding protein, and alpha-2-macroglobulin. In addition, we identified that proteins might be potential biomarkers to distinguish LSSD-CHB from DHSM-CHB, such as A0A0A0MS51_HUMAN (gelsolin), PON3_HUMAN, Q96K68_HUMAN, and TRPM8_HUMAN that were differentially expressed exclusively in LSSD-CHB patients and A0A087WT59_HUMAN (transthyretin), ITIH1_HUMAN, TSP1_HUMAN, CO5_HUMAN, and ALBU_HUMAN that were differentially expressed specifically in DHSM-CHB patients. Conclusion. This is the first time to report serum proteins in CHB subtype patients. Our findings provide potential biomarkers can be used for LSSD-CHB and DHSM-CHB.


Introduction
Chronic hepatitis B virus (CHB) infection is a leading cause of cirrhosis and hepatocellular carcinoma (HCC) and, in addition to morbidity and mortality, creates significant economic and social burdens [1,2]. It is estimated that approximately 240 million people have CHB infection worldwide and CHB infection should be responsible for 650,000 cases of hepatocellular carcinoma [2,3]. Due to the pathogenicity of CHB, early detection of CHB infection is the goal of treatment to diagnose and prevent the progression [4]. To this end, several hepatitis B virus (HBV) markers have been identified, including antigens (hepatitis B surface antigen, HBsAg; hepatitis Be antigen, HBeAg; hepatitis B core antigen, HBcAg), antibodies (hepatitis surface antibody, anti-HBs; hepatitis Be antibody, anti-HBe; hepatitis B core antibody, anti-HBc), and immunoglobulin (Ig) G and immunoglobulin M; however, unequivocal diagnosis requires more biomarkers [5].

Ethics Statement.
This study was conducted in compliance with the Declaration of Helsinki, the ethics approval was granted by the research medical ethics committee of Chengdu University of Traditional Chinese Medicine, and signed informed consent was obtained from all participants.

Patients and Serum Collection.
A total of 104 CHB patients were enrolled from West China Hospital, Sichuan University, and filtered with strict clinical evaluation described below. For iTRAQ proteomics analysis, we obtained blood samples from 30 LSSD-CHB patients, 30 DHSM-CHB patients, and 20 healthy controls (HCTL). For western blot analysis, 9 LSSD-CHB patients, 9 DHSM-CHB patients, and 6 HCTL participants were enrolled. Serum was collected from blood sample (4 mL) following the manufacture's protocol. Briefly, blood sample was incubated at room temperature for 2 h in vacutainer blood handling tube (Becton Dickinson, New Jersey, USA) and centrifuged for 10 min at 3,000 rpm and 4 ∘ C. Serum sample, which is the result supernatant, was transferred into a clean polypropylene tube and stored at −80 ∘ C.

Clinical Evaluation.
The viral markers HBsAg, HBeAg, anti-HBs, anti-HBc, and anti-HBe were determined routinely in serum samples using standard procedures (AxSYM5; Abbott Laboratories, Rungis, France), as well as other molecular diagnostic markers like ALT (alanine transaminase), AST (aspartate aminotransferase), STB (serum total bilirubin), CB (conjugated bilirubin), UCB (unconjugated bilirubin), and HBV-DNA. Participants, who have hepatitis B history or HBsAg positive history for more than six months, were diagnosed as chronic HBV infection if they were positive to HBsAg and/or HBV-DNA. We used both western and Chinese medicine criteria to divide CHB patients into two groups. First, participants were satisfied with the following requirements: (1) serum HBsAg positive for over 6 months; (2) HBV-DNA positive; (3) continuous or repeated elevated serum ALT in last 12 months; (4) being 18∼60 years old; (5) no planed move during the test. Then, LSSD-CHB and DHSM-CHB patients were diagnosed using the clinical symptoms mentioned before. CHB patients were diagnosed as LSSD-CHB when they met the criteria: (1) Mi and Mii; (2) Mi, Sii, and Siii; (3) Mii and Si. DHSM-CHB patients were diagnosed as follows: (1) M1 and M2; (2) M1, S1, and S2; (3) M2 and two of the secondary symptoms. We also filtered the patients when they satisfied one of the following criteria: (1) being associated with other types of hepatitis viruses or human immunodeficiency virus (HIV); (2) cirrhosis, malignancy; (3) being diagnosed with fulminant hepatitis (including acute, subacute, and chronic severe hepatitis); (4) being associated with drug or toxic liver, autoimmune hepatitis, and genetic-metabolic liver disease; (5) heart, lung, kidney, endocrine, blood, and other serious diseases; (6) pregnant women and lactating women; (7) mental disorders, in line with Chinese Classification of Mental Disorders Diagnosis (CCMD-3) standard; (8) other individuals not suitable for the cohort study.

Protein Preparation.
Serum sample (200 L) from each patient was processed to reduce the complexity by using ProteoMiner6 Kits (Bio-Rad Laboratories, Hercules, CA, USA). Then, the sample was eluted using Lysis buffer at pH 8.5 (2M Thiourea, 7M Urea, 4% CHAPS, and 40 mM Tris-HCl), reduced using 10 mM DTT at 56 ∘ C for 1 h, and alkylated using 55 mM IAM in darkness for 1 h. After being precipitated within chilled acetone (4 × volume) at −20 ∘ C overnight, the protein sample was centrifuged at 30,000 ×g for 15 min at 4 ∘ C; the pellet was next dissolved in 500 L of 0.5 M triethylammonium bicarbonate (Applied Biosystems, Milan, Italy) and sonicated at 200 W in ice for 15 min. Finally, the samples were centrifuged again at 30,000 ×g for 15 min at 4 ∘ C, and the supernatant was quantified using Bradford Protein Assay Kit (CWBIO, Beijing, China) and stored at −80 ∘ C for subsequent analysis.
SFX fractions were resuspended in buffer A (2% ACN, 0.1% FA), followed by a centrifugation at 20,000 g for 10 min. Then, 10 L of the supernatant was loaded onto a 2 cm C18 trap column on a LC-20AD nanoHPLC (Shimadzu, Kyoto, Japan) by the autosampler and eluted onto a 10 cm analytical C18 column (inner diameter 75 m) packed in-house. At 300 nL/min the samples were loaded with buffer B (98% ACN, 0.1% FA) as the following procedural: 5% B for 1 min, a 44 min gradient from 2 to 35% B, a 2 min linear gradient to 80%, and 80% B for 4 min. Finally, the chromatographic conditions were restored in 1 min. Data acquisition was performed with an Q EXACTIVE (Thermo Fisher Scientific, San Jose, CA) coupled online to the HPLC, as described [19,20].

Database Search and Protein Quantification.
Database search and protein quantification were performed using Mascot (Matrix Science, London, UK; version 2.4.0). Briefly, raw data files acquired from the Orbitrap were converted into mascot generic format (MGF) files using msconvert tool of ProteoWizard (http://proteowizard.sourceforge.net/) [21]. To identify expressed proteins and quantify them, all 20 MGF files were merged and searched against UniProtHUMAN (2016 04, 152,544 sequences) database using Mascot with parameters: quantification: iTRAQ 8plex (Applied Biosystems iTRQA6 8-plex); enzyme: trypsin; fixed modification: carboxymethyl (C), iTRAQ8plex (N-term) and iTRAQ8plex (K); variable modifications: dioxidation (M), oxidation (M), and iTRAQ8plex (Y), mass values: monoisotopic; peptide mass tolerance: ±15 ppm; fragment mass tolerance: ±20 mmu; max missed cleavages: (1) The charge states of peptides were set to +2 and +3. Specifically, an automatic decoy database search was performed in Mascot by choosing the decoy checkbox in which a random sequence of database is generated and tested for raw spectra as well as the real database. To reduce the probability of false peptide identification, only peptides at the 95% confidence interval by a Mascot probability analysis greater than "identity" were counted as identified. And each confident protein identification involves at least one unique peptide.

Protein Different Expression and Functional Analysis.
To identify differentially expressed proteins in LSSD-CHB and DHSM-CHB compared to HCTL, we set a cut-off for fold change (>1.2) of protein abundance provided by Mascot and value (<0.05) calculated by edgeR [22]. Venn diagram of upand downregulated proteins was analyzed by InteractiVenn (http://www.interactivenn.net/) [23]. To annotate potential functions of proteins, UniProt IDs of candidate proteins were submitted to DAVID Bioinformatics Resources 6.7 (https://david.ncifcrf.gov/home.jsp) [24] and STRING v10 (http://string-db.org/) [25], Gene Ontology (GO), and KEGG pathway were selected, and we used false discovery rate (FDR) to control the results. Protein-protein interaction networks were analyzed by STRING.

Statistical
Analysis. Statistical analysis including the calculation of mean value, standard deviation (SD), and students' -test was performed by using GraphPad Prism (v

Diagnosis of the Patients.
To study serum protein biomarkers in LSSD and DHSM CHB patients, we obtained a total of 80 participants, including 30 LSSD-CHB, 30 DHSM-CHB patients, and 20 healthy volunteers. As shown in Table 1 and  these diagnosis biomarkers in the blood samples of LSSD-CHB and DHSM-CHB patients were similar. In addition, the levels of ALT and AST remained at a high level, compared to healthy individuals [26,27], which confirmed their CHB infection.

Identification of Differentially Expressed Proteins.
Differentially expressed proteins were defined as those showed greater than 1.2-fold change in relative abundance and a value < 0.05. Compared to HCTL we identified a total of 392 proteins differentially expressed (Table S2), of which 273 were identified in LSSD-CHB group and 345 in DHSM-CHB group. As shown in the volcano plots, we identified 172

Potential Biomarkers for CHB.
The identification of proteins differentially expressed in LSSD-CHB and DHSM-CHB groups relative to the HCTL group was of interest as these could provide leads for potentially useful diagnostic and prognostic biomarkers. First, we examined those 142 commonly upregulated and 84 commonly downregulated proteins. As shown in Table 2, the largest upregulated protein family was immunoglobulin related protein, showing 20 upregulated and 3 downregulated proteins identified. In clinical immunology, levels of immunoglobulins especially IgG can be used to characterize viral hepatitis in patients [28,29]. Four IgG subclasses (IgG1 to IgG4) differ in their heavy chain constant regions and have different effects on virus-cell fusion inhibition, virus neutralization, and overall course of infection, as have been reported for various viruses including HIV [30] and HBV [31]. Highly expressed proteins encoding heavy chains for immunoglobulins including IGHG1, IGHG3, IGHG4, and IGH@ have been reported     with upregulation in HBV [32,33] and HCC patients [34].
We also identified 57 upregulated and 62 downregulated proteins exclusively in DHSM-CHB patients (Table 4). Two IGL@ proteins (Q6GMX4 HUMAN and Q6PIQ7 HUMAN) were specifically upregulated in DHSM-CHB patients with 1.27-fold change. Transthyretin (TTR), upregulated 1.34-fold in DHSM-CHB, can be induced by hepatitis C virus and activate TGF-signaling pathway with furin [51]. Interestingly, we found three members of tubulin (TUBA4A, TUBB1, and TUBB8) were upregulated only in DHSM-CHB patients compared with HCTL. Although there are few reports about tubulin and HBV, it is well known that 42 kDa tubulin alpha-6 chain fragment in well-differentiated hepatocellular carcinoma tissues is from patients infected with HCV [52]. In addition, we found actinin, alpha 1 (ACTN1), which can directly interact with HCV [53], GAPDH, which can bind to the HBV posttranscriptional regulatory element [54], and polymeric immunoglobulin receptor (PIGR), the main transporter of IgA [55], were upregulated in DHSM-CHB but not in LSSD-CHB. Among DHSM-CHB specifically downregulated proteins we identified three members of keratin type I (KRT9, KRT10, and KRT14) and another three members of keratin type II (KRT1, KRT2, and KRT6B). Although there is no evidence showing relation between these six keratin proteins with CHB or other liver diseases, variant keratins are associated with progression of fibrosis during chronic hepatitis C infection [56]. Differentially expressed proteins exclusively detected in DHSM-CHB patients were predicted to be involved in the biological processes of "immune system process" (GO: 0002376), "response to stress" (GO: 0006950), "defense response" (GO: 0006952), "immune response" (GO: 0006955), and "single-organism metabolic process" (GO: 0044710).
Compared to HCTL group up-and downregulated proteins exclusively in LSSD-CHB and DHSM-CHB patients showed their potential ability of being biomarkers for these two subtypes of HBV induced CHB. Some of them have been reported in other studies; however, more experiments need to be performed to investigate their functions and validate their specificity and accuracy in clinical trials.

Validation of the Quantitative Proteomic Analysis.
To validate the results obtained by proteomics analysis, eight randomly selected proteins and internal control albumin with altered expression profile were monitored by western blotting in an independent group of samples. Figures 4(a) and 4(b) showed the western blots for eight proteins and internal control albumin. PSMA6 (20S proteasome alpha6), PSMA7 (20S proteasome alpha7/alpha8) were upregulated and PF4 V (platelet factor 4 variant) was downregulated in LSSD-CHB group compared to HCTL (Figure 4(c)). Except SERPING1 (plasma protease C1 inhibitor), AHSG (fetuin-A), ACTB (actin), CTSC (cathepsin C), and PLTP (phospholipid transfer protein) were upregulated in the serum of DHSM-CHB patients (Figure 4(d)). Although the difference between patients and healthy participants was not significant by western blotting analysis, their regulations in patients and healthy group were consistent with iTRAQ. The original images of western blots (see Figure S1) might contain some differences due to brightness and contrast settings.

Discussion
Quantitation of serum or plasma proteins using comparative proteomics has recently been suggested as a suitable approach for the detection of liver disease biomarkers [17,[57][58][59]. The iTRAQ technology has been proposed as a powerful alternative to common tools (e.g., ELISA) and a flurry of applications emerged in the literature.
In this study, iTRAQ LC-MS/MS proteomics was used to detect serum protein as biomarkers of LSSD-CHB and DHSM-CHB patients. We compared the proteomics profile of LSSD-CHB and DHSM-CHB patients with healthy individuals and indicated 142 upregulated and 84 downregulated proteins shared by these two CHB subtype diseases. Protein-protein interaction network ( Figure 5) showed several significant proteins might function in response to HBV, such as actins (ACTA2, ACTB, ACTBL2, ACTN3, and ACTN4), apolipoproteins (APOA2, APOA5, APOB, APOC3, and APOC4), heat shock proteins (HSP90AA1 and HSP90AB1), and proteasome subunit proteins (PSMA1 and PSMA4). It has been reported that HBV core proteins can interact with the C-terminal region of actin-binding protein [60] and HBV X protein (HBx) can block filamentous actin bundles by interaction with eEF1A1 (eukaryotic translation elongation factor 1 alpha 1) [61]. In addition, ACTA2 is a marker of hepatitis stellate cells and correlated significantly with necroinflammatory grades and fibrotic stages in CHB or CHC [13]. Apolipoproteins are supposed to enhance the infectivity of hepatitis virus during the infection [39,62] and            Control-1  Control-2  Control-3  LSSD-CHB-1  LSSD-CHB-2  LSSD-CHB-3  LSSD-CHB-4  LSSD-CHB-5  LSSD-CHB-6  LSSD-CHB-7  LSSD-CHB-8  LSSD-CHB-9 28 KDa are identified to interact with HBx as well [63]. Among the apolipoproteins APOA2 is a considerable biomarker because its expression is increased on both mRNA and protein levels in CHB patients [14,64]. HBx protein also interacts with heat shock proteins and enhances HBx-mediated apoptosis [65]. A HBV-specific peptide (TVATAMG) is associated with heat shock protein and has potential for engineering tumor vaccines against hepatocellular carcinoma and chronic HBV infection [66]. Heat shock proteins like HSP27, HSP90, and GRP78 are upregulated in HBV related hepatocellular carcinoma, associated with vascular invasion and intrahepatic metastasis and have potential to be prognosis markers [67,68]. Commonly downregulated complement proteins are important mediators of inflammation and contribute to the regulation of the immune response. C4, a predisposing factor to autoimmune chronic active hepatitis [69], is expressed lowly in chronic hepatitis C patient compared to that in controls [70]. Low serum levels of complement in viral hepatitis are associated with high titers of hepatitis-associated antigen [71]. It is said that complement proteins are related to hepatitis B vaccine and C4AQ0 (mutant C4) probably contribute to inefficient complement activation and failure of B cells to secret anti-HBs [72]. Our results confirmed the potential of previously reported proteins in diagnosis of patients infected by HBV.
LSSD-CHB and DHSM-CHB are two subtypes of CHB according to traditional Chinese medicine pattern classification. In this study we identified 47 and 119 differentially expressed proteins exclusively in LSSD-CHB and DHSM-CHB, respectively, which could be used as biomarkers for LSSD-CHB and DHSM-CHB patients. We showed top 5 highly expressed proteins with different expression in LSSD-CHB and DHSM-CHB patients compared to HCTL group in Figure 6. Using relative expression ratio calculated by MASCOT we found mean expression levels of some proteins were close in LSSD-CHB and DHSM-CHB but with different values, such as CFH (complement factor H), F2 (prothrombin), and FGA (fibrinogen alpha chain). As we know, prothrombin time is one of the markers of liver test; it is usually lower in HBV infected patients than in healthy people and a good marker for liver fibrosis [73,74]. CFH functions as a cofactor in the inactivation of C3b by factor I [75], which can  CFL1  C4BPB  SERPINA1  CDH1  H2AFV  CFHR1 AFM  PROS1  F5  HIST1H4J  CPB2  ACTN4  PRMT3  YWHAZ  EHD4  ACTN2  CFHR3  VTN  ATP2A3  APOA2  ACTA2  F13A1  TPM1  VWF  CP  CLTC  YWHAE  CTTN  HTRA1  LCAT  APOB  HSP90AB1  IGFBP5  TPM4 TLN1  RAB8A  AKAP9  PRDX6  ARF3  PLTP  ORM2  RAB11B  IGFBP4  LPL  RAP1B  MYL6  CALM1  MYLK  COL18A1  APOA5  IGFBP1  ARF4  PLEK  PARVB  TAGLN2  APOC4  MPO  FLNA  PRDX5  PCOLCE  RASGRP1  TSKU  SAR1A ST13  interact with IgG and is moderately depressed in the serum of patients with viral hepatitis [71]. FGA has a major function in hemostasis as one of the primary components of blood clots [76]. Fibrinogen-like protein 2 (FGL2) has been identified as a potential biomarker for severity of CHC infection [77]. Other proteins also have been reported to be associated with HBV infection. LGALS3BP (lectin galactoside-binding soluble 3 binding protein isoform 1) were downregulated in LSSD-CHB patients (fc = 0.71, value = 2.49 − 06) and DHSM-CHB patients (fc = 0.89, value = 0.061). Previous studies about LGALS3BP in CHB and HCC found its different expression on transcriptional level [78], while in current study we identified its protein was differentially expressed in CHB patients and had the potential to be a good marker for LSSD-CHB subtype. PON3 (serum paraoxonase/lactonase 3), which was upregulated exclusively in LSSD-CHB, might play a hepatoprotective role against histological alterations and hepatic cell apoptosis leading to liver disease [48].
DHSM-CHB specifically differentially expressed proteins like ITIH1 (inter-alpha-trypsin inhibitor heavy chain H1), THBS1 (thrombospondin-1), C5 (Complement C5), and ALB (albumin) have been also reported in hepatitis viral related diseases. The expression level of ITIH1 in HCTL group was similar to that in LSSD-CHB patients (fc = 0.99, value = 0.227) but was downregulated significantly in DHSM-CHB patients (fc = 0.81, value = 0.012). The low expression of ITIH1 indicated it can be used to differ DHSM-CHB from LSSD-CHB. In addition, it has been experimented to be downregulated in HCV infected patients [79] and hepatitis C associated hepatocellular carcinoma patients [80]. It is reported that HCV viral proteins act directly or indirectly on THBS1 in TGF-pathway [81]. By noninvasive imaging the gene expression of THBS1 was upregulated in liver cancer [82]. Interestingly, ALB has been reported as an important factor to score the risk of HCC in CHB patients [83]. In our study, ALB was downregulated in both CHB subtypes but significantly exclusively in DHSM-CHB. Our results confirmed its different expression in CHB patients and revealed that the criteria of ALB expression in CHB patients required more patients and experiments. Due to the fact that hepatitis B viral load in DHSM-CHB patients was significantly higher than that in LSSD-CHB patients, we assume HBV-DNA might be related to CHB patients with different syndromes and it requires further experiments. To our knowledge, this study appears to be the first iTRAQ based approach aimed at identifying leads for potential useful biomarkers of patients of CHB subtypes. The candidates identified in this study await rigorous clinical validation using large cohorts of patient samples and more experimental function analysis.