Plasma checkpoint protein 1 (Chk1) as a potential diagnostic biomarker for opisthorchiasis and cholangiocarcinoma

2.1 Ethics and subjects

This study was approved by the Human Ethics Committee No. 1 of Thammasat University (045/2562) and the Human Ethics Committee of Udonthani Cancer Hospital (UCH-CT 11/2563). The study was conducted in accordance with the code of ethics set by the Declaration of Helsinki and all its amendments. The subjects were divided into three groups: the control group (non-opisthorchiasis and non-cholangiocarcinoma, non-OVCCA), the O. viverrini infected group (OV), and the cholangiocarcinoma group (CCA). Non-OVCCA subjects were defined as patients who had normal alanine transaminase (ALT) plasma levels and were found to have no O. viverrini eggs after stool examination. OV subjects were defined as patients with stool exams positive for O. viverrini eggs and confirmed to not have any signs of CCA on ultrasonographic examination. CCA subjects were defined as patients with a definite diagnosis of CCA via histopathology.

2.2 Sample preparation

We collected 50 plasma samples from each group and used them for plasma proteomic analysis. The total protein concentration of plasma was measured using the DC™ Protein Assay Kit (Bio-Rad) according to the manufacturer’s protocol. To prepare a pooled plasma sample for each group, 10 μ g of plasma from each patient was combined into a sample for each group. The pooled plasma samples were then analyzed by liquid chromatography with tandem mass spectrometry (LC-MS/MS) used in shotgun proteomic studies. Other plasma samples from each group (28 for non-OVCCA, 32 for OV, and 32 for CCA groups) were used for the Chk1 evaluation by indirect enzyme-linked immunosorbent assay (ELISA).

2.3 Sample preparation for gel-free based proteomics

Gel-free digestion or an in-solution-based method was used in this study. Total protein pellets (5 μ g) were dried and dissolved in 10 mM ammonium bicarbonate. The reduction of sulfhydryl bonds was performed by mixing 10 mM Dithiothreitol (DTT) in 10 mM ammonium bicarbonate at 56 ∘ C for 1 h. Alkylation of protein sulfhydryl groups was then performed by adding 100 mM iodoacetamide (IAA) in 10 mM ammonium bicarbonate at room temperature for 1 h in the dark. To perform digestion, 250 ng sequencing grade trypsin was added to the solution and incubated at 37 ∘ C for 8 h. Tryptic peptides were finally dried at 45 ∘ C under vacuum and protonated with 0.1% formic acid before injection into LC-MS/MS.

2.4 Label-free proteomic quantification and data interpretation

LC-MS/MS analysis of the digested peptide mixtures was performed using the Waters SYNAPT HDMS™ system. The 1D-nanoLC was carried out using the Waters nanoACQUITY UPLC system. Four microliters (100 ng) of an equal amount of digested peptide were adjusted for the peptide amount and injected onto the reverse phase (RP) analytical column (20 cm × 75 μ m) packed with a 1.7 μ m bridged ethyl hybrid (BEH) C18 material (Waters). Peptides were eluted with a linear gradient from 2% to 40% acetonitrile and developed over 60 min at a flow rate of 350 nl/min. This was followed by a 15-minute washing period with 80% acetonitrile to clean the column before returning to 2% acetonitrile for the next sample. The effluent samples were electrosprayed into a mass spectrometer (Synapt HDMS) for tandem mass spectrometry (MS/MS) analysis of the peptides. Argon gas was used in the collision cell to obtain the MS/MS data. The raw LC-MS/MS spectra were obtained using DataBridge to convert all spectra to a TXT data format.

2.5 Bioinformatics

LC-MS/MS datasets for label-free quantification were also evaluated. Relative comparisons of peak intensities of each peptide ion in all datasets of LC-runs were analyzed using the DeCyder MS Differential Analysis software [25]. After the injection (label-free analysis), the expression value was normalized using the R package NormalyzerDE (version 1.1.13) [26], in which quantile normalization was applied to the relative expression data analysis, after adding 1 to all expression values to avoid errors upon log-transformation for visualization. Peak detection (signal/noise ratio 2) and peak matching (peptide ion peak alignment) were performed using the Batch and Pepmatch module in the DeCyder MS Differential Analysis software. All datasets of MS/MS peak lists from the Pepmatch module in the DeCyder MS Differential Analysis software were exported to a Mascot generic format (.mgf file format) and submitted for an in-house database search using Mascot software (Matrix Science, London, UK). The dataset of MS/MS peak lists was searched against Homo sapiens (NCBI database, 115,529 sequences) for protein identification by following certain parameters; trypsin as digesting enzyme, maximum of three missed cleavages, MS peptide tolerance was 0.6 Da, MS/MS peptide tolerance of 0.1 Da, carbamidomethylation modification of cysteine residue as fixed modification, and the oxidation of methionine and acetylation of the protein N-terminus as variable modifications. All experiments were carried out with at least three independent replicates ( n = 3), and all data were expressed as mean ± standard deviation. Analysis of variance (ANOVA) was performed using the DeCyderMS program in the PepMatch modules. The significance of differences was determined using Duncan’s multiple range test ( 0.05).

2.6 Identification of opisthorchiasis viverrini and cholangiocarcinoma related proteins

The GenInfo Identifier (GI) numbers, a unique number that identifies a specific sequence, of the plasma proteome-derived proteins of the non-OVCCA, OV, and CCA proteomes were used for a Venn diagram analysis using Jvenn (http://jvenn.toulouse.inra.fr/app/index. html) [27]. A Principal Component Analysis (PCA) analysis was also plotted. In this study, the OV-and CCA-specific proteins (presented protein) were identified by Venn diagram and included a comparison of the OV vs. non-OVCCA, CCA vs. non-OVCCA, and OV vs. CCA proteomes. The GI numbers, retrieved UniProt IDs of OV-and CCA-specific proteins were then obtained (http://www.uniprot.org/uploadlists) [28] and further analyzed for gene ontology using PANTHER (http://www.pantherdb.org) for their molecular functions [29]. Since O. viverrini and nitrosamine compounds were identified as risk factors of cholangiocarcinoma in Thailand, protein-protein and protein-chemical interactions were observed by STITCH (http: //stitch.embl.de) to predict O. viverrini infection and cholangiocarcinogenesis-related proteins and identify potential biomarkers for O. viverrini infection and CCA [30].

2.7 Evaluation of the checkpoint protein 1 (Chk1) protein in plasma by ELISA

From differential non-OVCCA, OV, and CCA proteomes, Chk1 was remarked and selected as a potential biomarker. Evaluation of the Chk1 as a biomarker was performed using an indirect ELISA. The 1:3200 diluted plasma samples were used, and each sample was performed in duplicate. One hundred microliters of the 28, 32, and 32 diluted plasma samples from the non-OVCCA, OV, and CCA groups, respectively, in coating buffer (100 mM NaHCO 3 , pH 8.5) were added in the 96-well plates (Maxisorp, NUNC) and incubated at 4 ∘ C overnight. The plates were washed with 300 μ l washing buffer (1x PBS pH 7.4: 140 mM NaCl, 2.6 mM KCl, 12 mM Na 2 HPO 4 , 1.2 mM KH 2 PO 4 , pH 7.4, and 0.05% Tween ® 20) three times using a microplate washer (APW-100, Allsheng). The plate was then blocked for non-specific binding using 200 μ l of 5% skim milk (HiMedia) and incubated for 1 h at room temperature. One hundred μ l of 1:1000 diluted anti-phosphorylated S345 Chk1 rabbit polyclonal antibody (Abcam) in antibody diluent (1% skim milk in 1x PBS pH 7.4) was added and incubated for 1 h at room temperature. The plate was washed again, and 100 μ l of 1:10000 diluted goat anti-rabbit IgG HRP conjugate in antibody diluent was added and incubated for 1 h at room temperature. A TMB substrate (0.1 mg/ml 3,3,5,5-tetramethylbenzidine [Sigma] diluted in 10 ml of 0.05 M citrate phosphate buffer (pH 5.0) and 2 μ l of 30%H 2 O 2 (Merck)] was added and incubated at room temperature in the dark for 20 min. The reactions were stopped using 2 N H 2 SO 4 . The absorbance values at 450 nm were measured and used for graph plots and statistical analyses.

2.8 Statistical and diagnostic analyses

Statistical analyses were performed using IBM SPSS Statistics for Windows, version 22.0. (IBM Corp, Armonk, NY, USA). To determine whether there were statistically significant differences between the ELISA values of each group, one-way analysis of variance (ANOVA) and subsequent Scheffe post-hoc test were performed. The data were tested for normal distributions using the Kolmogorov-Smirnov test. The dot plot and the area under the ROC curve (AUC) were analyzed using GraphPad Prism 8. The diagnostic parameters, including sensitivity, specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR), positive predictive value (PPV), negative predictive value (NPV), and accuracy were calculated using MEDCALC (https:// www.medcalc.org/calc/diagnostic_test.php).

OPEN IN VIEWER

3.1 Analyses of plasma proteome

A gel-free LC-MS/MS method was used in this study. A total of 1,822 plasma proteins were identified (Additional File 1). The numbers of retrieved GI numbers, and the number of retrieved UniProt IDs identified from non-OVCCA, OV, and CCA plasma proteomes were 1,758, 1,718, and 1,768 proteins respectively, and are summarized in Table 1. The highest number of identified plasma proteins was found in the CCA group (1,768 proteins), while the lowest number was found in the OV group (1,718 proteins). The heat map and clustering of triplicates of each group were analyzed (shown in Additional File 2) and the triplicates of each group were individually clustered (Additional File 2). The non-OVCCA plasma proteome was more closely related to the CCA than to OV plasma proteomes. The PCA was plotted, and the results were individually separated (Fig. 1) in each group, which agreed with the heat map (Additional File 2). To classify the overall protein task in the cellular machinery, the proteome dataset was subjected to a Gene Ontology analysis by PANTHER according to their biological function. Gene ontology showed similarity in the number and percentage of each molecular function category (Fig. 2).

OPEN IN VIEWER

A Venn diagram of the OV against non-OVCCA was performed (Fig. 3). The results showed that 1,663 proteins were shared between the non-OVCCA and OV. The 46 and 86 proteins were group-specific proteins of the OV and non-OVCCA proteomes, respectively (Fig. 3). To investigate the host response proteins against opisthorchiasis, the protein-protein interactions of 46 OV-specific proteins were predicted. The results showed a network of cell checkpoints and proliferation-related proteins which included the checkpoint protein 1 (Chk1), phosphatidylinositol-4-phosphate-3-kinase (PIK3C2B), proto-oncogene with serine/threonine kinase (PIM1), mitogen-activated protein kinase 1 (MAPK1), and vascular cell adhesion molecule 1 (VCAM1). Prediction of the protein and chemical interactions was not performed by STITCH. These results suggested that O. viverrini infection triggered DNA damage-related protein and cell proliferation.

OPEN IN VIEWER

OPEN IN VIEWER

OPEN IN VIEWER

From the Venn diagram, 1,714 proteins were shared between the CCA and non-OVCCA. The 45 proteins were specifically identified in the CCA proteome compared to the non-OVCCA proteome (Fig. 4). Using STITCH, the network of specific proteins identified in the CCA proteome was similar to the OV-specific proteins related to cell signaling of cell proliferation and programmed cell death (Fig. 4). These proteins, which included the mitogen-activated protein kinase 1 (MAPK1), WD, tetratricopeptide repeats 1 (WDTC1), Parkinson protein 7 (PARK7), BCL2-like 12 (BCL2L12), paired box 2 (PAX2), vascular cell adhesion molecule 1 (VCAM1), and checkpoint protein 1 (Chk1), were mainly linked to TP53, the apoptotic pathway protein. Cell proliferation-related proteins, such as phosphatidylinositol-4-phosphate-3-kinase (PIK3C2B) and protooncogene with serine/threonine kinase (PIM1) were also clustered with this network. Furthermore, indirect linkage of midasin homolog (MDN1) and programmed cell death 11 (PDCD11) to TP53 via CCAAT/ enhancer binding protein (CEBPZ) was also predicted. Two vacuolar proteins related to the retrieval of lysosomal enzyme receptors such as IGF2R and M6PR from endosomes to the Golgi were also identified in the CCA plasma proteome. Furthermore, a link between acyl-CoA synthetase medium-chain family member 3 (ASCM3) and aldehyde dehydrogenase 1 family (ALDH1A2) was also observed. Protein and chemical interactions were not recruited using STITCH. These results suggested that changes in the regulation of proteins related to cell signaling of cell proliferation and apoptosis may have accounted for this.

O. viverrini has been identified as a risk factor for cholangiocarcinoma in Thailand. Therefore, the absence and/or presence of proteins in the plasma will be a useful biomarker for differentiating opisthorchiasis from CCA or the transitional phase of O. viverrini infection to CCA. From the Venn diagram, 91 proteins were only found in the CCA plasma proteome, whereas 41 proteins were only found in the OV proteome (Fig. 5). The predicted protein-protein interactions of 91 CCA-specific proteins showed the protein network of apoptotic proteins such as BCL2-antagonist/killer 1 (BAK1), TSPY-like 5 (TSPYL5), and p53-induced death domain protein (PIDD). It has also been linked to several chromatin-related proteins and transcription regulation, which include absent, small, or homeotic-like (ASH2L), widely interspaced zinc finger motifs (WIZ), zinc finger protein 429 (ZNF429), helicase DNA binding protein 7 (CHD7), staufen RNA binding protein homolog 1 (STAU1), within bgcn homolog (WIBG), v-myb myeloblastosis viral oncogene homolog (avian)-like 2 (MYBL2), and RNA binding motif protein 25 (RBM25). Lipid metabolism and transfer-related proteins, such as adipocyte plasma membrane-associated protein (APMAP), StAR-related lipid transfer (START) domain containing 3 (STARDS), and sterol carrier protein 2 (SCP2), were also predicted for their relationship in the protein network. The link between two proteins, cytoskeletal proteins, which include kinesin family member 2C (KIF2C) and dystonin (DST), was also observed. Interestingly, the predicted protein network linked to AKT1, the upregulated proteins in CCA, were related to cell proliferation (Fig. 5). For CCA-absent proteins or OV-specific proteins, COPB2 linked to collapsin response mediator protein 1 (CRMP1) and mitochondrially encoded ATP synthase 6 (MT-ATP6) via predicted functional partners added by STITCH: carbamoyl-phosphate synthetase 2 (CAD) and ATP synthase, H + transporting, mitochondrial F1 complex, alpha subunit 1 (ATP5A1) were predicted (Fig. 5). By manual addition of nitrosamine, the results showed no interaction with plasma proteins. Overall, these results suggested that in the transition of opisthorchiasis to CCA, the regulation of gene expression by chromatin modifications, and transcription factors played a role in cholangiocarcinogenesis.

3.5 Evaluation of the checkpoint protein 1 (Chk1) protein in the plasma

OPEN IN VIEWER

Prediction of protein-protein interactions by STITCH suggested that the Chk1 was located in the center of the OV and CCA protein networks. These results suggested that the Chk1 played a role in pathogenesis during O. viverrini infection and cholangiocarcinogenesis. Therefore, the plasma Chk1 levels were investigated and evaluated. The average absorbance values of the non-OVCCA, OV, and CCA groups are summarized in Table 2. The ELISA values were normally distributed ( $P>$ 0.1, α = 0.05). The dot plot with the mean value is shown in Fig. 6. The mean ± SD absorbances of the non-OVCCA, OV, and CCA groups were 0.5263 ± 0.101, 0.6338 ± 0.1609, and 0.7296 ± 0.1534, respectively (Table 3). All groups were statistically significantly different from each other using Turkey’s multiple comparison test (Table 3). A one-way ANOVA followed by a Scheffe post-hoc analysis revealed that the mean ELISA values were significantly different between the different groups ( $P>$ 0.05) (Table 3). ELISA results showed a rising tendency of the Chk1 in non-OVCCA, OV, and CCA. To evaluate its diagnostic biomarker potential, the AUC was 0.6987 for patients infected with O. viverrini (Fig. 7) and 0.8571 for cholangiocarcinoma patients (Fig. 8). Using an OD 450 cut-off of 0.6668, 59.38% sensitivity and 85.71 % specificity for O. viverrini-infected patients were obtained, while using OD 450 cut-off values of 0.5995, 65.62% sensitivity, and 85.71% specificity for cholangiocarcinoma were obtained (Table 4). The positive likelihood ratio (PLR), negative likelihood ratio (NLR), positive predictive value (PPV), negative predictive value (NPV), and accuracy are summarized in Table 4.

OPEN IN VIEWER

OPEN IN VIEWER