Serum proteomic predicts effectiveness and reveals potential biomarkers for complications in liver transplant patients

General information

General information on the donors and recipients evaluated during the perioperative period is summarized in Supplementary Table 1. There were 7 males and 3 females in the pre- and post-operative study group with an average age of 47.88 ± 8.49 years. For patients with AR (8 males and 2 females) and ITBL (7 males and 2 females) complications, the average age was 46.57 ± 8.75 and 52.00 ± 9.60 years, respectively. In healthy controls, there were 7 males and 3 females donors with an average age of 47.25 ± 8.73 years. There was no statistically significant difference between transplant patients and healthy controls with respect to the distribution of age and sex (p > 0.05).

In patients with ITBL, the average age of donors was higher than that of transplant excellent patients (55.75 ± 10.01 versus 42.54 ± 17.47 years, p = 0.2339), and anhepatic phase was significantly longer than transplant excellent patients (57.60 ± 7.28 versus 45.88 ± 8.43 minutes, p = 0.0377). The warm ischemia time was longer in patients with AR than in healthy transplant patients (14.57 ± 7.28 versus 9.63 ± 0.86 minutes, p = 0.0996). There were no major differences regarding the other factors.

Serum proteomic profiles of healthy controls, paired post-transplant patients and pre-transplant patients

We analyzed mass spectra of samples from healthy controls, paired post-transplant patients, and pre-transplant patients (perioperative group). Spectra are presented for the mass range of 1 to 10 kDa. The mass spectra differed between the healthy controls, paired post-transplant patients, and the pre-transplant patients (Figure 2A and 2B). Principal component analysis revealed that the perioperative group (Figure 2C and 2D) showed a slightly overlapping region, which suggested the possibility of exploring prognostic serum biomarkers to separate pre-transplant patients from control subjects. To evaluate the effect of transplantation, we analyzed the serum proteomic profiles in the perioperative group. ClinProTools analysis identified 79 different peaks, of which 5 differed significantly within the groups (fold change > 1.5; p < 0.001). Peaks 1-5 were downregulated in pre-transplant patients compared with healthy controls (Supplementary Table 2). The three most significantly differential peaks (Peak 1, m/z: 1949.82; Peak 2, m/z: 4100.54; Peak 3, m/z: 2666.86) observed in the mass spectra, showed similar values to those of post-transplant patients when compared with the healthy controls (Supplementary Table 2), which suggested that these three peptides could be potential biomarkers for transplant effectiveness.

Comparison of the spectra of these three peaks and their receiver operating characteristic (ROC) curves and area under the curve (AUC) values are displayed in Figure 2. After transplantation, these three peaks showed a tendency to return to levels as in healthy control (Figure 2E, 2G, and 2I). The AUC values of peaks 1, 2, and 3 were 1 (Peak 1, m/z: 1949.06), 0.93 (Peak 2, m/z: 4100.54), and 0.96 (Peak 3, m/z: 2666.86), respectively (Figure 2F, 2H, and 2J). Comparison of the spectra among healthy controls, paired post-transplant, and pre-transplant patients and their mean expression levels are shown in Figure 2K.

Serum proteomic profiles of healthy controls, transplant excellent patients and AR patients

We analyzed serum proteomic profiles in the AR group (healthy controls, transplant excellent patients and AR patients). Representative mass spectra of three samples and overall sum of the spectra differed in healthy controls, transplant excellent patients, and AR patients in the mass range from 1 to 10 kDa (Figure 3A and 3B). Figure 3C and 3D showed little overlapping region in principal component analysis for the AR group. ClinProTools analysis revealed 80 different peaks, of which two peaks (Peak 1, m/z: 1950.06; Peak 2, m/z: 2087.9) differed significantly among the AR group (fold change > 1.5; p < 0.001). Peaks 1 and 2 were significantly downregulated when comparing AR patients and transplant excellent patients to healthy controls (Supplementary Table 3).

Peptide mass spectra comparisons of the two peaks (Figure 3E and 3G) were consistent with the numerical results in Supplementary Table 3. The AUC values of Peak 1 (m/z: 1950.06) and Peak 2 (m/z: 2087.9) were 0.96 and 0.9, respectively (Figure 3F and 3H). The relative expression levels of these peptide peaks between healthy controls, excellent patients and AR patients are shown in Figure 3I. These results suggest that expression of two peptides in AR patients was higher than in the control subjects, indicating that they could be candidate diagnostic serum biomarkers for AR.

Serum proteomic profiles of healthy controls, transplant excellent patients and ITBL patients

We also analyzed serum proteomic profiling in the ITBL group (healthy controls, transplant excellent patients, and ITBL patients). Mass spectra differed in healthy controls, transplant excellent patients and ITBL patients (Figure 4A and 4B). Figure 4C and 4D show a slight overlapping region in principal component analysis in ITBL group.

ClinProTools analysis showed 72 different peaks, of which two (Peak 1, m/z: 2087.92; Peak 2, m/z: 1949.9) significantly differed between the three subject groups (fold change > 1.5; p < 0.001). Peaks 1 and 2 showed significant downregulation when comparing ITBL patients and transplant excellent patients to healthy controls (Supplementary Table 4).

Peptide mass spectra comparisons of the two peaks (Figure 4E and 4G) were consistent with the numerical results in Supplementary Table 4. The AUC values of Peak 1 (m/z: 2087.92) and Peak 2 (m/z: 1949.9) were 0.9625 and 1, respectively (Figure 4F and 4H). The relative expression levels of these peptide peaks among healthy controls, excellent patients and ITBL patients are shown in Figure 4I. Collectively, these results suggest that these peptides could be potential biomarkers for ITBL.

Identification of serum peptide biomarkers

All seven peptide peaks mentioned above from the perioperative group (Peak 1, m/z: 1949.82; Peak 2, m/z: 4100.54; Peak 3, m/z: 2666.86), AR group (Peak 1, m/z: 1950.06; Peak 2, m/z: 2087.9) and ITBL group (Peak 1, m/z: 2087.92, Peak 2, m/z: 1949.99) were identified using LC-ESI-MS/MS and the Uniprot database. The MS/MS spectrum of these peptides identified proteins including ACLY (m/z:1949.9), APOA1(m/z: 4100.81) and FGA (m/z: 2087.71 and 2666.86) (Table 1). The sequences of the peptides identified can be found in Supplementary Figure 1.

FunRich functional enrichment analysis and STRING interaction analysis

To further investigate the potential function and role of the three proteins, we performed FunRich analysis, and the results are shown in Figure 5A–5D. Firstly, the cellular component ontology described the subcellular structures and macromolecular complexes, and could therefore be used to annotate cellular locations of gene products. The identified proteins were scattered among various cellular components, including the lysosome (100%), extracellular space (66.7%), citrate lyase complex (33.3%), fibrinogen complex (33.3%), spherical high-density lipoprotein particle (33.3%), and platelet alpha granules (33.3%) (Figure 5A). Secondly, the molecular function analysis described the major functions of identified protein patterns, such as ATPase activity (33.3%), protein binding (33.3%), and transporter activity (33.3%) (Figure 5B). Thirdly, the three proteins were involved in a wide range of biological processes, including transport (33.3%), protein metabolism (33.3%), energy pathway (33.3%), and metabolism (33.3%) (Figure 5C). Finally, the biological pathway analysis demonstrated that these proteins were mainly involved in metabolism of lipids and lipoproteins (66.7%), acetyl-CoA biosynthesis (from citrate) (33.3%), ChREBP activates metabolic gene expression (33%), ABCA transporters in lipid homeostasis (33.3%), HDL-mediated lipid transport (33.3%), common pathway (33.3%), lipoprotein metabolism (33.3%), and formation of fibrin clots (clotting cascade) (33.3%) (Figure 5D). Overall, these proteins are very likely to play major roles in the metabolism of lipids and lipoproteins, pyruvate metabolism, and clotting cascade.

To explore the alteration of the protein interaction network in liver transplant patients from different groups, the STRING database (http://string-db.org) was used to generate a network from three seed proteins with their predicted functional partners. The results revealed a strong protein-protein interaction network among ACLY, FGA, and APOA1 (Figure 5E).

Full protein expression validation of identified peptides in patients’ serum

The potential biomarkers of transplant prognosis and complications diagnosis identified in our study, ACLY, FGA, and APOA1, were further validated with quantitative ELISA using the same cohort of samples (Figure 6A–6G). In the perioperative group, the mean serum concentration of ACLY was 1049.29 ± 150.42 pg/ml (range 863.21-1291.79 pg/ml) in healthy controls, 878.87 ± 96.26 pg/ml (range 713.21-1063.21 pg/ml) in paired post-transplant patients, and 533.50 ± 98.13 pg/ml (range 356.07-681.50 pg/ml) in pre-transplant patients (Figure 6A). The mean serum concentration of FGA was 1042.22 ± 126.63 ng/ml (range 872.50.21-1258.61 ng/ml) in healthy controls, 909.50 ± 115.95 ng/ml (range 734.55-1091.17 ng/ml) in paired post-transplant patients, and 564.39.50 ± 123.92 ng/ml (range 389.41-725.98 ng/ml) in pre-transplant patients (Figure 6B). The mean serum concentration of APOA1 was 16.70 ± 2.63 μg/ml (range 12.77-20.19 μg/ml) in healthy controls, 11.03 ± 2.85 μg/ml (range 7.08-15.13 μg/ml) in paired post-transplant patients, and 6.84 ± 2.38 μg/ml (range 4.14-11.70 μg/ml) in pre-transplant patients (Figure 6C). The expression levels of these three proteins were significantly lower in pre-transplant patients, with all p values < 0.0001, showing the same tendency as their peptides (Table 2, Figure 2).

In addition, the concentration of ACLY and FGA in healthy controls, transplant excellent patients, and AR and ITBL patients also showed a noticeably gradual decrease (Figure 6D–6G). Concentrations of ACLY in AR patients (629.50 ± 107.90) were significantly lower than those of transplant excellent patients (837.21 ± 124.72, p < 0.001), and those in transplant excellent patients were significantly lower than those in healthy controls (1049.29 ± 150.42, p < 0.001) (Figure 6D). Concentrations of FGA in AR patients (722.17 ± 148.64) were significantly lower than those of transplant excellent patients (880.83 ± 137.76, p < 0.05), and those in transplant excellent patients were significantly lower than those in healthy controls (1042.22 ± 126.63, p < 0.05) (Figure 6E). A similar trend was also observed in the peptide analysis of the ITBL group. Expression levels of ACLY and FGA in the ITBL patients were significantly lower than those in transplant excellent patients (p < 0.01, p < 0.001, respectively) and healthy controls (p < 0.001, both) (Figure 6F–6G).

All results indicated that ACLY, FGA, and APOA1 were all expressed at significantly lower levels in pre-transplant, AR, and ITBL patients compared to the other subjects, showing the same tendency as the proteomics. Finally, western blot analysis was performed to verify the relative serum expression changes for the identified proteins in different groups (Figure 6H–6J). Taken together, both ELISA and western blot results further verified the credibility of our proteomic analysis.

Correlation analysis between potential serum biomarkers and clinical information

To further demonstrate the clinical utility of our serum proteomic result, we analyzed the correlation between potential serum biomarkers and clinical information. Compared to the healthy control group, pre-transplant patients displayed pancytopenia (reduced white blood cell (WBC), red blood cell (RBC) and platelet (PLT)), lower neutrophil (NEU), lymphocyte (LYM) and monocyte (MONO), lower liver synthesis function markers (cholesterol (CHO), albumin (ALB) and fibrinogen (FIB)), but higher gamma-glutamyl transferase (GGT), total bilirubin (TBIL) and prothrombin time (PT). AR and ITBL patients also displayed pancytopenia, and lower LYM, lower liver synthesis function markers (ALB and FIB), but higher ITBL and PT, compared to the healthy controls. AR patients were more prone to having worse liver function with higher liver enzyme markers (ALT, AST, and GGT). The clinical laboratory values listed above (Supplementary Table 5) are consistent with clinical routine and previously published articles. In the perioperative group, ACLY was positively correlated with CHO and ALB, and FGA and APOA1 were positively correlated with CHO, ALB, and FIB (Table 2). Serum ACLY and FGA levels were not correlated with any laboratory value in the AR and ITBL groups, which is likely due to the smaller sample size.

Patients and samples

All serum samples and clinical information were collected after obtaining patients consent in accordance with the declaration of Helsinki for patients receiving standard orthotopic liver transplantation from January 2016 to January 2017. The study was approved by the ethics committee of the First Affiliated Hospital of Xi’an Jiaotong University. Each patient was followed-up after discharge at least every 2 weeks in the first year and every 2 months in the second year. 20 serum samples were obtained from 10 pairs of patients before and 14 days after transplantation. In addition, we screened AR patients (n = 10): patients newly diagnosed as AR with Banff criteria and ITBL patients (n = 9): patients newly diagnosed as ITBL using endoscopic retrograde cholangiography (ERC). All samples were obtained within 48 hours after the onset of AR or ITBL, and then the initial treatment was carried out. All the patients (n = 10) recovered well after transplantation, which we call excellent patients. Blood samples of patients without any complications were collected 14 days after surgery and followed up at least 2 years.. Blood samples from 10 healthy donors who were matched for age and gender without any evidence of diseases were also collected as controls.

The serum samples were collected in 10 ml vacuum tubes without anticoagulants and were kept at 4 °C for 1 h, then centrifuged at 3,000g for 15 min at 4 °C. The serum samples were distributed in 500 μl aliquots and stored at −80 °C until use.

MALDI-TOF MS analysis

Serum samples were separated using MB-WCX chromatography (Bruker Daltonics, Bremen, Germany) according to the manufacturer’s protocol. With the magnet lowered, 5 μl serum samples were diluted in 10 μl binding solution in a standard thin well PCR tube, added to 10 μl of MB-WCX beads and mixed by pipetting up and down. After incubated for 5 min at room temperature, the tube was placed into the magnetic separator and the beads were collected on the wall of the tube until the supernatant was clear (~1 min). The supernatant was then removed carefully and the magnetic beads were washed three times with washing buffer. Next, we eluted the peptide fraction from the magnetic beads with 5 μl of elution solution and 5 μl stabilization buffer. Finally, we spotted 1 μl eluted peptide and 1μl alpha-cyano-4-hydroxycinnamic acid (Bruker Daltonics) in 50% acetonitrile onto MALDI-TOF MS targets, and added 0.5% trifluoroacetic acid twice to the MALDI AnchorChip surface. All samples were spotted in triplicate to evaluate reproducibility.

ClinProTools analysis

All targets were analyzed using a calibrated Autoflex III MALDI-TOF MS (Bruker) immediately, with an optimized protocol of FlexControl software (Version 3.0, Bruker). Mass calibration was executed with a standard calibration mixture of peptides and proteins (mass range, 1,000-10,000 Da). All tests were performed in a blinded manner, including the serum analysis of different groups. All serum data were analyzed by Flex analysis software (Version 3.0; Bruker) and recognition of peptide patterns was performed using ClinProTools software (Version 2.2; Bruker), including spectra pretreatment, peak detection and peak calculation operation.

Peptide identification

The peptides were analyzed by liquid chromatography-mass spectrometry (LC-MS), which integrate Nano Acquity UPLC (Waters, USA) with LTQ Orbitrap XL mass spectrometer (Thermo Fisher Scientific, USA). 20 μl samples were trapped by C18 column (Symmetry^®C18, 3 μm, 0.10×20 mm, nanoAcquity™Column) at 600 nl/min for 3 min and then loaded on analytical column (Symmetry^®C18, 1.9 μm, 0.15×120 mm, nanoAcquity™ Column) at 400 nl/min for 60 min. Mobile phase A was a solution of 5% acetonitrile and 0.1% formic acid, and mobile phase B was a solution of 95% acetonitrile and 0.1% formic acid, and the column temperature was maintained at 35°C. The LTQ-Orbitrap XL mass spectrometer was operated in the data-dependent mode to switch automatically between MS and MS/MS acquisition. Survey full scan MS spectra with 2 microscans (m/z 300–1400) were acquired with the Q Exactive with a mass resolution of 70,000 at m/z 300, followed by 10 sequential LTQ-MS/MS scans. Dynamic exclusion was used with 2 repeat counts, with 18s repeat duration and 80s exclusion duration. For MS/MS, precursor ions were activated with 25% normalized collision energy at the default activation q of 0.25. The mass spectra were searched against the Uniprot database (https://www.uniprot.org/) using Mascot software (Version 1.5.2.8). To reduce false positives, a decoy database containing reverse protein sequences was added to this database. The search parameters were as follows: no enzyme, the variable modification was oxidation of methionine, peptide tolerance of 15 ppm, fragment mass tolerance of 20 mmu. Positive protein identification was confirmed with Peptide FDR ≤ 0.01.

Bioinformatics analysis

Functional enrichment and interaction network analysis tool (Funrich, http://www.funrich.org) was used to analysis cellular components, molecular functions, biological processes and biological pathway of the identified peptides. The interaction network of the differentially expressed proteins was built automatically by the STRING (Search Tool for the Retrieval of Interacting Genes/Proteins) system (http://string-db.org) with default settings.

ELISA

All serum samples were analyzed blindly in triplicate and the concentrations of ACLY (Shanghai Hengyuan, H-12925), FGA (Shanghai Hengyuan Biotech, H-12843) and APOA1 (Shanghai Hengyuan, H-11508) were measured according to the manufacturer’s instructions respectively.

Western blot

Western blot was used to quantify the protein expression of ACLY, FGA and APOA1. 10 μg of protein from serum samples were separated on 10% SDS-PAGE, transferred to PVDF membranes, and incubated with 1:1000 diluted primary antibodies against ACLY (GeneTex, GTX60666, Monoclonal), FGA (Proteintech Group, 20645-1-AP) and APOA1 (Proteintech Group, 14427-1-AP).

Statistical analysis

SPSS 25.0 statistical software and GraphPad Prism Version 6.0 (GraphPad Software, La Jolla, CA, USA) were used for statistical analysis. Data are shown as mean ± standard deviation and P<0.05 was considered statistically significant. Differences were evaluated using Student’s t test for continuous parametric data, Wilcoxon test for continuous nonparametric data, and Pearson’s chi-squared test for noncontinuous data. Pearson correlation analysis was used to determine the correlation between expression of serum biomarkers and clinical information.