Introduction

The current status of hepatic carcinoma

Hepatic carcinoma is one of the most common types of malignant tumors in the digestive system, and its biological characteristics determine its high rate of metastasis and recurrence after radical resection, leading to poor prognosis for patients [1]. The 5-year recurrence rate of hepatic carcinoma patients undergoing radical surgery is 40–60%, and the 5-year survival rate is only 50–60% [2]. About 77% of hepatic carcinomas in China are associated with hepatitis B virus infection and have the characteristics of basic liver disease, hidden onset, atypical symptoms, and a poor prognosis. Most patients are in the mid- to late stages at the time of initial diagnosis and lose the opportunity for surgery or other local treatment. Even if surgery or other local treatments are possible, recurrence and metastasis are common [3]. Currently, the recommended treatment plans for mid- to late stages of hepatic carcinoma are liver transplantation, preoperative neoadjuvant therapy (induction or transformation) to reduce the tumor size and phase before resection, or transcatheter arterial chemoembolization (TACE), TACE + system therapy, hepatic arterial infusion chemotherapy (HAIC), and other interventional treatments, radiotherapy, system therapy, etc. [4]. Interventional therapy, local perfusion chemotherapy, systemic chemotherapy, and molecular targeted therapy can benefit some hepatic carcinoma patients, and their roles in the comprehensive diagnosis and treatment system are gradually emerging. However, up to now, there are very few types of targeted molecules available for clinical selection in hepatic carcinoma, and their effectiveness is low [5]. The use of serum alpha-fetoprotein (AFP) to diagnose primary hepatic carcinoma is simple and practical and is currently a commonly used diagnostic method in clinical practice. However, patients with hepatic carcinoma and hepatitis can detect serum AFP, but not all hepatic carcinoma patients show high levels of serum AFP. Early diagnosis biomarkers are urgently needed, and auxiliary treatment methods are also needed to reduce the recurrence rate after radical treatment for improved patient outcomes.

The importance of phosphorylation in hepatic carcinoma

Post-translational modifications (PTMs) are covalent modifications during or after protein synthesis and are also one of the reasons for the diversity of human proteoforms, including phosphorylation, methylation, acetylation, glycosylation, nitrosylation, ubiquitination, sulfonylation, nitration, etc. [6]. If a protein is modified at multiple sites, these modifications might also affect each other, resulting in synergistic or antagonism effects. Protein phosphorylation is catalyzed with multiple kinases to achieve complex and precise biological regulation and plays important roles in cell proliferation, survival, apoptosis, drug resistance, metabolism, transcription, and differentiation [7]. Abnormal phosphorylation in a protein usually leads to cell proliferation disorders, which cause tumor occurrence. Therefore, phosphoproteomics has become a hot field in cancer research and tumor drug research and development, and advances in mass spectrometry technology enable the simultaneous study of multiple PTMs [8]. About 30% of the encoded proteins are identified as being phosphorylated in their biological cycle. Moreover, phosphoprotein can also be dephosphorylated, which is catalyzed with phosphatase. Protein phosphorylation and dephosphorylation are dynamic reversible processes, which regulate the basic biological function of cells [9]. The occurrence rates of phosphorylation at residues serine (Ser; S), threonine (Thr; T), and tyrosine (Tyr; Y) are about 90%, 10%, and 0.05%, respectively. Although the phosphorylation abundance of residue Y is much lower than that of residues S and T, phosphorylation at residue Y is very important because specific tyrosine kinase inhibitors have shown some surprising effects in various cancers [10]. A large amount of evidence shows that phosphorylation phenotype is related to pathological characteristics such as histopathology grading, depth of tissue infiltration, etc. In recent years, an increasing number of phosphoproteins, phosphatases, and kinases in various cancers have been identified, which demonstrates that reversible protein phosphorylation regulates cancer-related biological processes [11]. One phosphoproteomics study found that many phosphatases and related phosphorylation sites are key switches in cancer biological processes [12]. Another phosphoproteomics study compared the chimeric antigen receptor T-cell therapy (CART) cell signaling to identify high and low levels of CD22 antigen in cancer cells and found multiple phosphorylation sites in LAT protein, the molecular mechanism of CD22-CART low antigen resistance, and a new CART that can help overcome resistance [13]. The rapid development of phosphoproteomics promotes the core value of mass spectrometry-based proteomics and PTMomics research in precision medicine.

Quantitative phosphoproteomics as a tool to reveal phosphorylation profiling in cancers

Quantitative phosphoproteomics refers to mass spectrometry-based quantitative proteomics that studies phosphoproteins [14]. In this technique, solid-phase metal affinity chromatography, TiO2, or antibody phosphorylation groups can be used to enrich phosphopeptides. In the process of phosphopeptide enrichment, monophosphorylated peptide segments and polyphosphorylated peptide segments are separated step by step, followed by mass spectrometry analysis [15]. Mass spectrometry-based quantitative phosphoproteomics can qualitatively and quantitatively analyze phosphoproteomes to provide detailed information on phosphorylation sites and their abundance. The 6-plex tandem mass tag (TMT) labeling reagents coupled with TiO2 enrichment of phosphopeptides and liquid chromatography-tandem mass spectrometry (LC–MS/MS) were well used to identify and quantify each phosphoprotein and phosphosite between cancer tissues and controls [16]. Moreover, integrative analysis of phosphoproteomics data and transcriptomics data from The Cancer Genome Atlas (TCGA) database offered more valuable biomarkers for PPPM in hepatic carcinoma [17].

Working hypothesis

We hypothesize that the phosphoproteome and phosphorylation-mediated signaling pathway networks are significantly different in human early-stage primary hepatic carcinoma tissues compared to control liver tissues. The differentially phosphorylated proteins (DPPs), phosphosites, phosphorylation-mediated signaling pathway networks, and key kinase profiling in combination with other omics data and clinical data offer great promise for insight into the accurate molecular mechanisms of early-stage primary hepatic carcinoma, the discovery of effective therapeutic targets/drugs, and the determination of reliable phosphorylation-related biomarkers for patient stratification, predictive diagnosis, prognostic assessment, and personalized medical services of early-stage primary hepatic carcinoma in the framework of predictive, preventive, and personalized medicine (PPPM; 3PM).

Study design

This study used TMT-labeling reagents coupled with TiO2 enrichment of phosphopeptides and LC–MS/MS to identify DPPs in early-stage primary hepatic carcinoma tissues compared to tumor-adjacent control tissues. Those identified phosphosites within DPPs will provide a new diagnostic or effective therapeutic target to detect cancer early or stop tumor growth. Phosphorylation, as an important switch in the regulation of signaling pathways in the body, is closely related to the occurrence, development, and metastasis mechanisms of tumors. Furthermore, phosphorylation-mediated molecular pathways and network changes provide a better understanding of molecular mechanisms in primary hepatic carcinoma and will help to develop new anti-tumor therapies to improve the therapeutic effects. The phosphorylation status of different phosphosites has a significant impact on their function and would show heterogeneity. Moreover, the integration of phosphoproteomics data and transcriptomics data from TCGA database provides in-depth information on the clinical values of phosphorylation in primary hepatic carcinoma.

Materials and methods

Tissue specimen

Human hepatic carcinoma and tumor-adjacent “normal” control tissues from patients were obtained from the Department of General Surgery, The Third Xiangya Hospital, Central South University. This study was reviewed and approved by the Third Xiangya Hospital Medical Ethics Committee of Central South University and the Medical Ethics Committee of Shandong First Medical University, China. After surgery, the tumor tissues and tumor-adjacent “normal” control tissues were obtained within 2 h, quickly frozen in liquid nitrogen (− 196 °C), and then stored at − 80 °C for subsequent phosphoproteomics analysis. Moreover, for primary hepatic carcinoma, it is easy to distinguish cancer tissue from tumor-adjacent normal control tissue with morphological judgment, including macroscopic judgment and microscopic distinction. Tumor-adjacent normal tissue is the normal liver tissue around the edge of the tumor. Because primary hepatic carcinoma generally has a margin envelope and the color and texture of cancer tissue and normal liver tissue are very different, it is not difficult to distinguish, which is also confirmed with a microscope. For this study, the distance was greater than 1 cm from the tumor-adjacent normal tissue site to the primary hepatic carcinoma tissue margin envelope to guarantee the reliability of the use of tumor-adjacent normal tissue as control tissue. The collected tissues were washed with phosphate-buffered saline (PBS) to remove residual blood and pollutants and eliminate irrelevant interfering tissues (blood vessels, fat, connective tissue, etc.). After it was rinsed thoroughly, tissue scissors were used to separate the tissue into small pieces (~ 1 mm3). It was frozen in liquid nitrogen for 5 min and stored at − 80℃. The clinical characteristics of each sample are listed (Table 1).

Table 1 Clinical characteristics of hepatic carcinoma tissue samples, and the paired tumor-adjacent normal control tissue samples from the same patients
Full size table

Protein extraction and digestion

The sodium dodecyl sulfate (SDS)-dithiothreitol (DTT)-Tris–HCl (SDT) buffer (4% SDS, 1 mM DTT, 100 mM Tris–HCl, and pH 7.6) was used for sample lysis and protein extraction. The amount of protein was quantified with the bicinchoninic acid (BCA) protein assay kit (Bio-Rad, USA). Proteins were extracted from cancer and tumor-adjacent normal tissues of 6 hepatic carcinoma patients (Table 1). The proteins extracted from 6 cancer tissues were mixed in equal amounts as cancer samples, and then the mixed cancer protein sample was divided into 3 parts for TMT quantitative phosphoproteomics. The proteins extracted from 6 tumor-adjacent normal tissues were mixed in equal amounts as control samples, and then the mixed control protein sample was divided into 3 parts for TMT quantitative phosphoproteomics. The 3 parts of cancer protein samples and 3 parts of control protein samples were analyzed with 6-plex TMT-labeled quantitative phosphoproteomics to determine DPPs and phosphosites. Protein digestion with trypsin was performed according to the filter-aided sample preparation (FASP) procedure. The tryptic peptides of each sample were desalted on C18 cartridges (Empore™ SPE Cartridges C18 (standard density), bed I.D. 7 mm, volume 3 ml, Sigma), concentrated by vacuum centrifugation and reconstituted in 40 µl of 0.1% (v/v) formic acid.

FASP digestion procedure

An amount (200 μg) of proteins for each sample was resolved into 30 μl SDT buffer that contained 4% SDS, 100 mM DTT, and 150 mM Tris–HCl at pH 8.0. The detergent, DTT, and other low molecular weight components were removed with urea lysis buffer (UA buffer) (8 M urea, 150 mM Tris–HCl at pH 8.0) by repeated ultrafiltration (Microcon units, 10 kD). A volume (100 μl) of 100 mM iodoacetamide in UA buffer was then added to block the reduced cysteine residues, and the samples were incubated (30 min; darkness). The filters were washed with 100 μl UA buffer three times and then with 100 μl of 25 mM NH4HCO3 buffer twice. Finally, the protein suspensions were digested with 4 μg trypsin (Promega) in 40 μl of 25 mM NH4HCO3 buffer (overnight; 37 °C), and the tryptic peptides were collected as a filtrate. The tryptic peptides of each sample were desalted with C18 cartridges (Empore™ SPE Cartridges C18 (standard density), bed I.D. 7 mm, volume 3 ml, Sigma), concentrated by vacuum centrifugation, and reconstituted in 40 µl of 0.1% (v/v) formic acid. The tryptic peptide content was estimated by ultraviolet (UV) light spectral density at 280 nm with an extinction coefficient of 1.1 in a 0.1% (g/l) solution.

TMT labeling of tryptic peptides

An amount (100 μg) of tryptic peptide mixture of each sample (cancer: 3 parts; control: 3 parts) was labeled with 6-plex TMT reagent (Thermo Fisher Scientific) according to the manufacturer’s instructions, and the TMT-labeled tryptic peptides were equally mixed (1:1:1:1:1:1).

Fractionation of TMT-labeled peptides

The mixed TMT-labeled tryptic peptides were fractionated by strong cation exchange (SCX) chromatography on the AKTA purifier system (GE Healthcare). Briefly, the dried tryptic peptide mixture was reconstituted and acidified with buffer A (10 mM KH2PO4 in 25% of acetonitrile, pH 3.0) and loaded onto a PolySULFOETHYL 4.6 × 100 mm column (5 µm, 200 Å, PolyLC Inc, Maryland, USA). The peptides were eluted at a flow rate of 1 ml/min with a gradient of 0–8% buffer B (500 mM KCl, 10 mM KH2PO4 in 25% of acetonitrile, pH 3.0) for 22 min, 8–52% buffer B during 22–47 min, 52–100% buffer B during 47–50 min, 100% buffer B during 50–58 min, and then buffer B was reset to 0% after 58 min. The elution was monitored by absorbance at 214 nm, and fractions were collected every 1 min. The collected fractions were desalted on C18 cartridges (Empore™ SPE Cartridges C18 (standard density), bed I.D. 7 mm, volume 3 ml, Sigma) and concentrated by vacuum centrifugation.

Enrichment of phosphopeptides

The fractionated peptide samples were reconstituted in 1.4 ml of precooled immunoaffinity purification (IAP) buffer. TiO2 beads were added to the mixture, shaken for 40 min, centrifuged, and discarded into the supernatant. The beads were put into the tips with a stopper, then washed with washing buffer 1 for 3 times, and then with washing buffer 2 for 3 times. The phosphopeptides were eluted with elution buffer, concentrated under vacuum, and then dissolved in 20 μl 0.1% formic acid for MS analysis.

LC–MS/MS analysis

LC–MS/MS analysis was performed on a Q Exactive HF mass spectrometer (Thermo Scientific) that was coupled to Easy nLC (Proxeon Biosystems, now Thermo Fisher Scientific) for 120 min. The enriched phosphopeptides were loaded onto a reversed-phase trap column (Thermo Scientific Acclaim PepMap100, 100 μm × 2 cm, nanoViper C18) connected to the C18 reversed-phase analytical column (Thermo Scientific Easy Column, 10-cm long, 75 μm inner diameter, 3 μm resin) in buffer A (0.1% formic acid) and separated with a linear gradient of buffer B (84% acetonitrile and 0.1% formic acid) at a flow rate of 300 nl/min controlled by IntelliFlow technology. The mass spectrometer was operated in positive ion mode. MS data were acquired with a data-dependent acquisition (DDA) method to dynamically choose the top 10 most abundant precursor ions from the survey scan (m/z 300–1800) for higher-energy collisional dissociation (HCD) ion fragmentation. The automatic gain control (AGC) target was set to 3E6, and the maximum inject time was 10 ms. The dynamic exclusion duration was 40.0 s. Survey scans were acquired at a resolution of 70,000 at m/z 200, the resolution for HCD spectra was set to 17,500 at m/z 200, and the isolation width was 2 m/z. The normalized collision energy was 30 eV, and the underfill ratio, which specifies the minimum percentage of the target value likely to be reached at maximum fill time, was defined as 0.1%. The instrument was run with the peptide recognition mode enabled.

Identification and quantitation of phosphoproteins

MS/MS spectra were used to search the protein database using the MASCOT engine (Matrix Science, London, UK; version 2.2) embedded into Proteome Discoverer 2.4. The searching parameters were set as enzyme = trypsin, max missed cleavages = 2, fixed modifications = carbamidomethyl (C) and TMT 6/10/16 plex (N-term), variable modifications = oxidation (M) and phosphor (ST)/phosphor (Y), peptide mass tolerance =  ± 20 ppm, fragment mass tolerance =  ± 0.1 Da, database pattern = Decoy, peptide false-discovery rate (FDR) =  ≤ 0.01, and protein FDR =  ≤ 0.01. The experimental bias was controlled by normalizing all peptide ratios by the median protein ratio, and the median protein ratio should be 1 after the normalization. The MS/MS data were used to determine the amino acid sequence of a phosphoprotein and its phosphosites. The TMT reporter ion intensities were used to quantify the phosphorylation level at each phosphosite and determine each DPP.

Bioinformatics analysis of DPPs in hepatic carcinoma

The reproducibility analysis of three repeated experiments in the cancer group or in the control group determined that the cutoff values (fold-change > 1.5 or < 0.67 plus p < 0.05) were used to identify each DPP. The identified DPPs were subsequently mapped to signaling pathways in the Kyoto Encyclopedia of Genes and Genomes (KEGG) database with the ClusterProfiler R package (http://bioconductor.org/packages/release/bioc/html/clusterProfiler.html) with both p < 0.05 and adjusted p < 0.05. Gene annotation of DPPs was performed with Cytoscape ClueGO (two-sided hypergeometric test, correlation coefficient > 0.99, and adjusted p value < 0.05 corrected with Benjamini–Hochberg), including biological processes (BPs), cellular components (CCs), and molecular functions (MFs). To evaluate their interactive associations, all DPPs were mapped in a protein–protein interaction (PPI) network in the STRING database with a combined score of > 0.9 (https://cn.string-db.org/).

Transcriptomics data of DPPs in hepatic carcinoma

The corresponding mRNA expression and clinical data of the identified DPPs in hepatic carcinoma tissue samples (n = 373) and non-paired tumor-adjacent normal control tissues (n = 50) (Total = 423 (373 + 50) hepatic carcinoma patients) were downloaded from TCGA website (https://portal.gdc.cancer.gov/). The R package limma (http://www.bioconductor.org/packages/release/bioc/html/limma.html) was used to identify differentially expressed genes (DEGs) between hepatic carcinoma (n = 373) and control (n = 50) tissue groups with moderated t-tests (p value < 0.05). The Kaplan–Meier method within the survival R package (https://www.rdocumentation.org/packages/survival/versions/3.2-3) was used to select overall survival (OS)-related DEGs (p value < 0.05) in 417 out of 423 hepatic carcinoma patients with full clinical information and survival data in TCGA database. For each DEG, its mRNA expression values among 417 hepatic carcinoma patients were ranked, and then 417 patients were grouped into high-expression value group (n = 208) and low-expression value group (n = 209) according to the mean value of the 417 mRNA expression values of each DEG for survival curve analysis. The association analysis between OS-related DEGs and drug sensitivity was performed with the Corrplot R package plus Spearman method (p < 0.05) based on the drug sensitivity data of 263 drugs among 60 tumor cell lines and transcriptomic data (n = 23,805 genes) among these 60 tumor cell lines from CellMiner (NCI-60 Analysis Tools) (https://discover.nci.nih.gov/cellminer/) in hepatic carcinoma.

Statistical analysis

Enrichment analysis was applied based on Fisher’s exact test, considering the whole quantified protein dataset as the background dataset. The Benjamini–Hochberg correction for multiple testing was further applied to adjust derived p values. The Kaplan–Meier method was used to generate survival curves for the subgroups in each dataset, and the Log-rank (Mantel–Cox) test was used to determine the statistical significance of a difference. In all cases, a p value under a threshold of 0.05 was considered statistically significant.

Results

DPPs in early-stage primary hepatic carcinoma tissues identified with quantitative phosphoproteomics

Human early-stage hepatic carcinoma tissues (n = 6) and paired tumor-adjacent control tissues (n = 6) were obtained from six patients from the Department of General Surgery, and clinical characteristics of each sample were collected (Table 1). All samples were at stage I (early-stage) without metastasis and premedication.

A total of 1326 phosphopeptides derived from 858 DPPs in human early-stage hepatic carcinoma tissues (n = 6) were identified compared to the paired tumor-adjacent control tissues (n = 6), including 538 phosphopeptides with decreased phosphorylation level, and 788 phosphopeptides with increased phosphorylation level (Supplementary Table 1). For those identified phosphopeptides, 1227 phosphopeptides showed only one phosphosite (T, S, and Y), and the rest phosphopeptides were identified with more than one phosphosite; for example, phosphorylation occurred at residues S927, S928, S931, and S934 in EP400 (GFDALQES*S*LDS*GMS*GRK), and phosphorylation occurred at residues T60, T64, T66, and Y69 in TRDV2 (T*QGNT*MT*FIY*REK). This study found that 11 phosphopeptides had three phosphosites, including GLOD5 (DT*T*MFY*SK), NEMF (T*ALNS*FMHS*K), SRSF2 (S*RS*RS*PPPVSK), KIAA1217 (KYPDSHLPT*LGS*KT*PPASPHR), DDX46 (S*RS*S*SPGNK), WDR75 (EFY*LSVYFFK), ADGRF5 (WS*S*QHS*K), ADGB (EIVSQT*T*AT*QEK), GDPD5 (Y*T*QVS*RQELR), DLC1 (PKT*T*AIQGIS*EKEK), and KIFC1 (LS*LS*RS*DER), and 85 phosphopeptides had two phosphosites (Supplementary Table 1). These phosphoproteins might be involved in human early-stage hepatic carcinoma pathophysiology.

Significant signaling pathways enriched with DPPs in human early-stage primary hepatic carcinoma

A total of 858 DPPs in human early-stage primary hepatic carcinoma were analyzed with KEGG pathway enrichment method, which identified 33 statistically significant signaling pathways (Fig. 1; Supplementary Table 2), including spliceosome (Fig. 2), glycolysis/gluconeogenesis (Fig. 3), B-cell receptor signaling pathway (Fig. 4), fatty acid degradation (Fig. 5), HIF-1 signaling pathway (Fig. 6), insulin signaling pathway, focal adhesion, alcoholic liver disease, regulation of actin cytoskeleton, proteoglycans in cancer, adipocytokine signaling pathway, glucagon signaling pathway, ErbB signaling pathway, insulin resistance, glioma, Fc epsilon RI signaling pathway, epithelial growth factor receptor (EGFR) tyrosine kinase inhibitor resistance, tight junction, drug metabolism (cytochrome P450), PPAR signaling pathway, metabolism of xenobiotics by cytochrome P450, carbon metabolism, starch and sucrose metabolism, neurotrophin signaling pathway, thyroid hormone signaling pathway, platelet activation, viral life cycle (HIV-1), biosynthesis of amino acids, leukocyte transendothelial migration, bacterial invasion of epithelial cells, arrhythmogenic right ventricular cardiomyopathy, proximal tubule bicarbonate reclamation, and hypertrophic cardiomyopathy. The flow-chart images of all these DPP-involved signaling pathways in human early-stage hepatic carcinoma were collected (Supplementary Fig. 1).

Fig. 1
figure 1

Phosphorylation involves significant signaling pathway alterations in human hepatic carcinoma. A total of 33 statistically significant signaling pathways (p < 0.05 and FDR < 0.05) were identified. The darker dot means the more significant enrichment. The size of the dot represents the number of DPPs in the pathways

Full size image
Fig. 2
figure 2

Phosphorylation-mediated spliceosome pathway in primary hepatic carcinoma. The red box means the increased DPPs in the pathway, and the green box means the decreased DPPs in the pathway

Full size image
Fig. 3
figure 3

Phosphorylation-mediated glycolysis/gluconeogenesis in primary hepatic carcinoma. The red box means the increased DPPs in the pathway, and the green box means the decreased DPPs in the pathway

Full size image
Fig. 4
figure 4

Phosphorylation-mediated B-cell receptor signaling pathway in primary hepatic carcinoma. The red box means the increased DPPs in the pathway, and the green box means the decreased DPPs in the pathway

Full size image
Fig. 5
figure 5

Phosphorylation-mediated fatty acid degradation in primary hepatic carcinoma. The red box means the increased DPPs in the pathway, and the green box means the decreased DPPs in the pathway

Full size image
Fig. 6
figure 6

Phosphorylation-mediated HIF-1 signaling pathway in primary hepatic carcinoma. The red box means the increased DPPs in the pathway, and the green box means the decreased DPPs in the pathway

Full size image

Functional characteristics of DPPs in human early-stage primary hepatic carcinoma

GO enrichment analysis of 858 DPPs hierarchically classified these DPPs into different clusters, including 57 BPs (Supplementary Table 3; Fig. 7A), 40 CCs (Supplementary Table 4; Fig. 7B), and 37 MFs (Supplementary Table 5; Fig. 7C). For BPs, the main phosphorylation-mediated BPs were metabolic process, cellular process, localization, biological regulation, viral process, biological adhesion, developmental process, positive regulation of the biological process, negative regulation of the biological process, regulation of biological process, and response to stimulus. For CCs, phosphorylation-mediated main CCs were extracellular space, intracellular anatomical structure, nucleoplasm, cytoplasm, cytosol, endomembrane system, nuclear body, cell junction, membrane-enclosed lumen, envelope, protein-containing complex, cell projection, organelle, perinuclear region of cytoplasm, supramolecular complex, nuclear protein-containing complex, catalytic complex, and ribonucleoprotein complex. For MFs, phosphorylation-mediated main MFs were chromatin binding, catalytic activity, carbohydrate derivative binding, molecular function regulator, GTPase activator activity, heterocyclic compound binding, structural molecule activity, binding, protein binding, enzyme activator activity, lipid binding, oxidoreductase activity, small molecule binding, enzyme regulator activity, positive regulation of catalytic activity, ion binding, positive regulation of molecular function, protein-containing complex binding, regulation of catalytic activity, regulation of molecular function, and organic cyclic compound binding.

Fig. 7
figure 7

The functional characteristics of DPPs in primary hepatic carcinoma. A The 57 significant biological processes (BPs) enriched with DPPs (p < 0.05 and FDR < 0.05). B The 40 significant cellular components (CCs) enriched with DPPs (p < 0.05 and FDR < 0.05). C The 37 significant molecular functions (MFs) enriched with DPPs (p < 0.05 and FDR < 0.05). The lower p value and more significant enrichment were shown with the larger node size. The same color indicated the same function group. Among the groups, a representative of the most significant term is lag-highlighted

Full size image

PPI networks involved by DPPs in human early-stage primary hepatic carcinoma

Those 858 DPPs were uploaded to STRING software for PPI analysis. The combined scores of nodes ranged from 0.900 to 0.999 (Supplementary Table 6; Fig. 8). Some of them showed both high-combined score (> 0.99) and high correlation coefficient of co-expression (> 0.99), such as RPL27A and RPS28, RPS28 and RPL27A, CCT8 and TCP1, TCP1 and CCT8, CCT2 and CCT8, CCT2 and TCP1, CCT8 and CCT2, RPL18A and RPS28, RPS28 and RPL18A, TCP1 and CCT2, RPL12 and RPL27A, RPL12 and RPL34, RPL27A and RPL12, RPL27A and RPL34, RPL34 and RPS3, RPL34 and RPL27A, RPL34 and RPL12, RPS28 and RPS6, RPS3 and RPL34, RPS6 and RPS28, RPL18A and RPL34, RPL18A and RPL29, RPL27A and RPS6, RPL29 and RPL18A, RPL34 and RPL18A, RPS6 and RPL27A, RPL18A and RPS6, RPL27A and RPS3, RPL34 and RPS6, RPS3 and RPL27A, RPS6 and RPL18A, RPS6 and RPL34, RPL12 and RPL18A, RPL12 and RPS6, RPL18A and RPL12, RPS3 and RPS6, RPS6 and RPS3, RPS6 and RPL12, MCM2 and MCM3, MCM3 and MCM4, MCM3 and MCM2, MCM4 and MCM3, RPL12 and RPS3, RPL18A and RPL27A, RPL18A and RPS3, RPL27A and RPL18A, RPS3 and RPL18A, RPS3 and RPL12, MCM2 and MCM4, and MCM4 and MCM2.

Fig. 8
figure 8

Protein–protein interaction (PPI) networks were constructed with multiple high combined scores and co-expressed DPPs

Full size image

DEGs, survival analysis, and drug sensitivity of DPPs based on transcriptomics data in hepatic carcinoma

The mRNA expression and clinical data of 858 DPPs in hepatic carcinoma tissue samples (n = 373) and non-paired tumor-adjacent normal control tissues (n = 50) were obtained from 423 hepatic carcinoma patients in TCGA database (https://portal.gdc.cancer.gov/) (Supplementary Table 7). A total of 105 DEGs were identified between hepatic carcinoma tissues (n = 373) and non-paired normal control tissues (n = 50) (Supplementary Table 8; Fig. 9).

Fig. 9
figure 9

Differentially expressed genes (DEGs; n = 105) identified with TCGA transcriptomics data in hepatic carcinoma tissues (n = 373) vs. tumor-adjacent normal control tissues (n = 50)

Full size image

Survival analysis of DPPs identified 125 OS-related DPPs in hepatic carcinoma patients between high- (n = 208) and low-mRNA-expression (n = 209) groups (Supplementary Table 9; Supplementary Fig. 2), and 44 out of 125 OS-related DPPs had a low p value (< 0.01), including OR2M4, NDRG2, EIF5B, SLC2A1, PLEKHA4, MIA3, GJA5, ANXA5, IQSEC1, SORBS2, CHMP4A, PGM2L1, CCDC137, ALDOA, GAPDH, HMGA1, RAB3IL1, UPB1, PSD4, LMOD1, AGFG1, MEF2C, SRSF2, TMCC1, FHOD3, SLC23A2, PIK3R1, PHKA2, SRSF11, MGMT, ENO1, CCT2, HNRNPM, SRRT, RBM17, TCP1, IRS1, SPP1, EPB41L2, TCOF1, SRSF3, YBX1, PRPF38B, and BMS1. All significant OS-related DPPs in human early-stage primary hepatic carcinoma were collected (Supplementary Fig. 2).

Combining analysis of 105 DEGs, 858 DPPs, and 125 OS-related DPPs identified 14 common molecules among “DPPs,” “OS-DPPs,” and “DEGs” (Fig. 10), including ANLN, CCT2, FTCD, HJURP, HNRNPM, MCM2, NDRG2, PECR, PNPLA7, SLC23A2, SSRP1, STMN1, TCOF1, and TOP2A.

Fig. 10
figure 10

Overlapping Venny plot and Kaplan–Meier (KM) survival curve. A Fourteen overlapped molecules among three sets of data (DPPs, DEGs, and OS-DDPs), including FTCD, NDRG2, CCT2, PECR, SLC23A2, PNPLA7, ANLN, HNRNPM, HJURP, MCM2, STMN1, TCOF1, TOP2A, and SSRP1. B–F KM survival curve of HNRNPM, FTCD, MCM2, PECR, and SLC23A2 in 417 hepatic carcinoma patients who were divided into high- (n = 208) and low-expression (n = 209) groups according to the mean value of each gene

Full size image

Drug sensitivity analysis of 125 OS-related DPPs, with NCI-60 analysis tools based on the drug sensitivity data of 263 drugs and 23,805 gene expression data among 60 tumor cell lines, found that 26 OS-related DPPs had significant associations with drug sensitivity (Fig. 11; Supplementary Table 10), including LMOD1, CAV2, UBE2E2, RAPH1, ANXA5, HDLBP, CUEDC1, APBB1IP, VCL, SRSF10, SLC23A2, EPB41L2, ESR1, PLEKHA4, SAFB2, SMARCAD1, VCAN, PSD4, RDH16, NOP56, MEF2C, BAIAP2L2, NAGS, SRSF2, FHOD3, and STMN1. A high correlation coefficient existed between some of these OS-related DPPs and drugs, including LMOD1 and karenitecin, CAV2 and tamoxifen, UBE2E2 and vemurafenib, BAIAP2L2 and dabrafenib, SRSF2 and chelerythrine, FHOD3 and staurosporine, STMN1 and chelerythrine, RDH16 and fulvestrant, ESR1 and SR16157, ESR1 and fulvestrant, UBE2E2 and bafetinib, MEF2C and bafetinib, and NAGS and rebimastat.

Fig. 11
figure 11

The OS-related DPPs in hepatic carcinomas showed significant associations with drug sensitivity. This analysis was performed with NCI-60 analysis tools based on the drug sensitivity data of 263 drugs and 23,805 gene expression data of 60 tumor cell lines

Full size image

Discussion

The onset of hepatic carcinoma is hidden, and the early diagnosis rate is low. About 70% of patients are already in the middle to late stage at the initial diagnosis [18]. Before 2017, only sorafenib and chemotherapy were available for drug treatment. In recent years, with the emergence of various new targeted drugs and immunotherapy drugs, drug therapy for advanced hepatic carcinoma has also had continuous breakthroughs; however, there are still many difficulties and challenges [19]. Due to the large number of patients with late-stage disease, the choice of medication treatment is still relatively limited. A considerable number of patients are not suitable or unable to tolerate chemotherapy, targeted therapy, and immune checkpoint inhibitors due to various reasons, such as poor overall condition, poor liver function, and heavy tumor burden [1]. Therefore, improving f the diagnostic rate of early-stage hepatic carcinoma patients is the key approach to improving the prognosis of hepatic carcinoma. Protein phosphorylation and dephosphorylation are reversible dynamic processes that rely on multiple kinases and phosphatases to achieve complex and precise biological regulation and play important roles in cell proliferation, survival, apoptosis, metabolism, transcription, and differentiation. Abnormal protein phosphorylation is usually associated with cell proliferation disorders, which lead to tumor occurrence. Therefore, phosphoproteomics has become a hot field in cancer research and tumor drug research and development [20]. One high-throughput phosphoproteomics analysis of hepatic carcinoma samples from different etiologies and clinical stages revealed the extensive molecular heterogeneity of hepatic carcinoma and the key regulatory pathways in the occurrence and development of different subtypes of hepatic carcinoma [21]. This present study focused on the phosphoproteome of human early-stage hepatic carcinoma, which identified 1326 phosphopeptides derived from 858 DPPs. Among these DPPs, many were closely related to hepatic carcinoma and were involved in a variety of basic biological processes and molecular functions. With the integrative analysis of DPPs, DEGs, and clinical data, this present study provides great significance in studying molecular mechanisms of the occurrence and development of early-stage primary hepatic carcinoma, discovering potential drug targets, and identifying effective biomarkers in the framework of PPPM.

The identified DPPs were closely related to malignant behaviors of hepatic carcinoma

This phosphoproteomics study on human early-stage primary hepatic carcinoma identified 1326 phosphopeptides derived from 858 DPPs. Some DPPs were reported to be closely associated with malignant tumor behavior in previous studies. For example, in terms of phosphorylation of Rho GTPase-activating protein 7-DLC1 (ratio T/N = 0.62), the phosphorylation switch of DLC1 was critically dependent on the dynamic interactions of DLC1 with TNS3 in hepatic carcinoma, and this switching process is indispensable for both the initiation and continuation of cancer cell migration [22]. In terms of phosphorylation of mitogen-activated protein kinase 8-MAPK8 (ratio T/N = 0.57), one study found that the increased phosphorylation level of MAPK8 was significantly related to cell viability and apoptosis, and MAPK8 might be one of the targeted genes of megestrol acetate (MA) and arsenic trioxide (ATO). Also, co-treatment with MA/ATO markedly improved the inhibition of cell viability and enhanced apoptosis, which enhanced antitumor efficacy in the treatment of hepatic carcinoma [23]. In terms of phosphorylation of tensin-1-TNS1 (ratio T/N = 0.64), it is a family of focal adhesion proteins and affects cell polarization, migration, and invasion. Tensin-1 has a phosphotyrosine-binding domain that binds β-integrin and a phosphatase and tensin homology domain that binds protein phosphatase-1α. One study analyzed the phosphosites of human S-tag-tensin-1 in hepatic carcinoma cells with mass spectrometry and found that the increased phosphorylation of S-tag-tensin-1 provided a hub to connect signaling pathways involving tyrosine kinases that could increase binding to endogenous pTyr proteins [24]. In terms of phosphorylation of nucleolar transcription factor 1-UBTF (ratio T/N = 1.88), it is an essential transcription factor responsible for ribosomal DNA transcription activation and might be involved in the cytotoxic mechanism of platinum drugs. One study analyzed protein phosphorylation patterns in different cisplatin- and oxaliplatin-DNA lesions with LC–MS/MS and found that the phosphorylation level of UBTF was decreased in hepatic carcinoma cell lines. Changes to site-specific phosphorylation levels of DPPs could provide valuable resources for mechanistic studies of anti-tumor agents [25]. In terms of phosphorylation of 5′-AMP-activated protein kinase catalytic subunit alpha-1-PRKAA1 (ratio T/N = 0.55), PRKAA1-related signaling pathway exacerbated chronic inflammation, energy metabolism, cellular aging, and hypoxia in the tumor microenvironment. During hepatic carcinoma formation, the decreased phosphorylation of PRKAA1 contributed to the upregulation of mTOR and HIF1α, which promoted the development of fatty liver into hepatic carcinoma [26]. In terms of phosphorylation of tight junction protein ZO-1-TJP1 (ratio T/N = 1.71), one study found that tyrosine phosphorylation of TJP1 leads to downregulation of TJP1 and contributes to redifferentiation of the glands and the occurrence and metastasis of hepatic carcinoma [27]. In terms of phosphorylation of cytochrome P450 2E1-CYP2E1 (ratio T/N = 0.40), it is an important metabolic enzyme in the liver and is involved in the occurrence and development of liver diseases and hepatocellular carcinoma. Under the induction of substrates, the overexpression of CYP2E1 produced a large amount of reactive oxygen species (ROS) to damage biofilm function through lipid peroxidation in the liver. The phosphorylation and degradation pathway of CYP2E1 could alter the expression and activity of its coding enzymes and participate in the occurrence and development of liver disease and hepatocellular carcinoma [28]. This present study was helpful in constructing the references of quantitative phosphoproteomics in hepatic carcinoma. Comparative analysis of phosphoproteomes between normal and pathological individuals can identify DPPs and further identify disease-specific protein phosphorylation to serve as molecular targets for new drug design or molecular markers for early-stage diagnosis.

Phosphorylation plays an important role in multiple signaling pathways in hepatic carcinoma

Pathway network analysis of 858 DPPs derived from early-stage primary hepatic carcinoma identified 33 statistically significant signaling pathways, and many of these phosphorylation-mediated signaling pathways were closely associated with the malignant behaviors of hepatic carcinoma, including gene splicing-related pathways, cancer metabolism-related pathways, immune-related pathways, and the HIF-1 pathway.

Phosphorylation-mediated gene splicing-related pathways in hepatic carcinoma

RNA alternative splicing (RNA-AS) is a common gene expression regulation mechanism, and the abnormal regulation of RNA-AS is closely related to tumor occurrence and development. RNA-AS can result in gene mutations at the specific site, mutations or abnormal expressions of splicing factors, as well as abnormalities of tumor cells [29]. RNA-AS can cause non-expression of tumor suppressor genes or induce specific oncogenic protein formation. At the same time, such splicing sites can also become a target for tumor treatment. Studies have already found that many small-molecule compounds can target and inhibit the RNA-AS process and have effective anticancer effects in preclinical studies [30]. Further understanding of the impact of RNA-AS abnormalities on tumor biological functions and targeting RNA-AS is needed for further research. The previous study found that DDX17 affected metastasis and progression of hepatic carcinoma cells by participating in RNA-AS of tumor-related genes, such as splicing factors [31]. Our study found that phosphorylation played an important role in the spliceosome signaling pathway in hepatic carcinoma. Some proteins in this pathway were found to be phosphorylated, including HSPA1A, TRA2B, PRPF38B, RBM17, PRPF40A, TRA2A, HNRNPM, SRSF6, HNRNPC, SF3B1, NCBP1, ACIN1, RBMX, SRSF2, SNRNP70, SRSF9, SF3B2, THOC1, SRSF10, DDX23, HNRNPA3, SRSF7, DHX8, SRSF4, DDX46, and SRSF3. The executor of precursor mRNA splicing in the nucleus is the spliceosome, which can recognize splicing signals before mature mRNA is nucleated and translated, remove non-coding introns, and splice exons that can encode proteins together [32]. Phosphorylation and dephosphorylation are necessary in the process of spliceosome circulation, which can regulate protein nucleocytoplasmic shuttle and protein–protein interaction [33]. A large number of splicing factors that play a major and regulatory role include arginine/serine (RS)-enriched domains. The low phosphorylation level of SR-related proteins can inhibit RNA-AS under mitotic or heat shock conditions. On the contrary, when its RS domain is phosphorylated, it can strongly bind to the exon splicing enhancer region and promote splicing [34]. One study found that a protein phosphatase (PPM1G) could interact with the phosphorylation of AS-related protein SRSF3, and change the AS patterns of genes, including MYC, MAX, MED1, EP300, and ELF1, in hepatic carcinoma cells [35]. In this study, some phosphoproteins were initially identified to affect RNA-AS, but more studies are needed to clarify how those phosphoproteins act as both inhibitors and promoters to regulate RNA-AS and where they bind to the rigid RNA sites.

Phosphorylation-mediated cancer metabolism-related pathways in hepatic carcinoma

Glycolysis is a ubiquitous energy metabolism process in the organism, which is a pathway where glucose generates pyruvate through a series of enzyme catalysis under the condition of insufficient oxygen and then reduces to lactic acid, also known as anaerobic oxidation [36]. In the presence of oxygen, oxidative phosphorylation is the main approach for body cells to obtain energy, and glycolysis is inhibited. However, in malignant tumors, even if sufficient oxygen is supplied, glycolysis is still very active [37]. The glycolytic capacity of tumor cells is 20–30 times that of normal cells, and the glucose uptake is about 10 times that of surrounding normal cells [37]. Therefore, the energy metabolism characteristics of malignant tumor cells are that glycolysis in an aerobic state is their main energy acquisition mode and glucose is their only energy substrate. Warburg discovered as early as the 1920s this phenomenon: tumor cells consume a large amount of glucose and produce lactate under aerobic conditions, known as the “Warburg effect” [38]. Phosphorylation plays an important role in glycolysis, and many intermediate products of glycometabolism are phosphorylated; for example, a typical representative is that glycogen metabolism is regulated by phosphorylation and dephosphorylation, which is fully reflected in the synergistic regulation of the synthesis and decomposition of glycogen [39]. When the glycogen synthase is phosphorylated, it changes from the active state to the inactive state to inhibit glycogen synthesis; at the same time, when the glycogen phosphorylase is also phosphorylated, it changes from the inactive state to the active state to promote glycogen decomposition [40]. Our study found that phosphorylation plays an important role in the glycolysis/gluconeogenesis signaling pathway in hepatic carcinoma, and some proteins in this pathway were found to be phosphorylated, including ADH4, TPI1, ADH5, ADH1B, PGM1, GAPDH, ADH1C, PGAM1, ALDOC, ACSS2, PGM2, ENO1, PCK1, and ALDOA. One quantitative phosphoproteomics study of a rat model with non-alcoholic steatohepatitis livers revealed the activation of the glycolysis pathway [41]. A preclinical study found that targeted therapy strategies based on the glucose metabolism characteristics of tumor cells were effective in anticancer therapy [42]. Currently, internationally targeted anti-tumor metabolic therapies based on glycolysis characteristics mainly include controlling glucose supply, reducing environmental glucose concentration with selective toxicity to low oxygen content tumor cells, inhibiting glycolytic enzymes, inhibiting HIF-1 production, and inhibiting the mTOR pathway [43]. In view of the limited efficacy of glucose metabolism reprogramming to single targets for inhibitors, the use of omics data to develop multi-targets and anti-tumor effects has great potential.

Lipid metabolism plays an important role in the structure, dynamic changes, steady-state regulation, and signal transduction processes of all organisms and links thousands of lipid molecules together through a complex macroscopic metabolic network [44]. High-throughput phosphoproteomics revealed that the overall regulatory effect of phosphorylation on lipid metabolism processes not only regulated lipid metabolism processes by affecting enzyme activity but also occurred at the functional site of fatty acid synthase (FAS) to change the length of the fatty acid chain [45]. Studies found that abnormal cell proliferation was a common characteristic of all malignant tumor cells, and the formation of cell membranes and signaling molecules during the process of abnormal cell proliferation requires more involvement in fat metabolism, including fatty acid metabolism and lipid droplet metabolism [46]. Protein phosphorylation could regulate the mechanisms of specific lipid synthesis and metabolism in tumor cells. For example, phosphorylated GSK3β resulted in GSK3β degradation, while GSK3β was an important upstream kinase of the key gene SREBP1 in fat metabolism [47]. Our study found that phosphorylation played an important role in the fatty acid degradation signaling pathway in hepatic carcinoma. Some proteins in this pathway were found to be phosphorylated, including ADH4, ADH5, ADH1B, ADH1C, ACADVL, HADHA, ACSL1, and EHHADH. Therefore, abnormal lipid metabolic pathways and lipid-related enzymes are potential targets for anticancer drug therapy in hepatic carcinoma.

Phosphorylation-mediated immune-related pathways in hepatic carcinoma

In recent years, significant progress has been made in the non-surgical treatment of hepatic carcinoma, and combined immunotherapy can significantly improve the survival time of a small proportion of patients with advanced hepatic carcinoma [48]. The combination of chemotherapy and PD-1 inhibitors enhances the immune response by inducing the release of tumor-specific antigens, which will make it possible to achieve the transformation of partially unresectable advanced hepatic carcinoma into resectable hepatic carcinoma [49]. New therapies combined with immunotherapy are important ways to achieve hepatic carcinoma transformation therapy and improve the survival of patients with advanced hepatic carcinoma [49]. Immunotherapy for hepatic carcinoma has gradually become a hot research topic and has achieved positive results in clinical practice. Some patients have a longer survival cycle and a lower recurrence rate after receiving immunotherapy. Immunotherapy is gaining global recognition as a new standard of care for hepatocellular carcinoma [50]. The targeted cytotoxic T immune checkpoint suppressor lymphocyte-associated protein 4 (CTLA-4) and anti-programmed cell death protein 1 (PD-1) drug preparations (ICIs) were used for hepatic carcinoma immunotherapy, which changed the traditional sorafenib treatment mechanism, and as an adjuvant treatment, it reduced the recurrence rate to a certain extent and improved the survival rate of hepatic carcinoma [51]. Phosphorylation plays an important role in immune-related pathways, which could regulate the immune system and reverse the immunosuppressive environment in hepatic carcinoma. Specifically, different phosphoproteins can transmit various signals through several intracellular signaling molecules within the cell, regulating their growth, differentiation, and death [52]. For example, one study revealed that macrophages utilized serine and threonine kinases (STK4)-mediated inflammatory responses by binding and phosphorylating IRAK1, which reduced the occurrence and development of chronic inflammation that leads to hepatic carcinoma [53]. Our study found that phosphorylation played an important role in the B-cell receptor signaling pathway in hepatic carcinoma. Some proteins in this pathway were found to be phosphorylated, including PTPN6, VAV1, DAPP1, PRKCB, AKT2, SOS1, LYN, BTK, INPP5D, PIK3R1, PIK3R3, BCL10, and PLCG2. In addition to the combination of immunotherapy and antiangiogenic therapy or targeted drugs, PTMs, including phosphorylation, also play a crucial role in the occurrence of hepatic carcinoma through various mechanisms to alter gene expression. Therefore, it is also considered to combine phosphorylation regulators for hepatic carcinoma therapy.

Phosphorylation-mediated HIF-1 pathway in hepatic carcinoma

Hypoxia is a major characteristic of solid tumors and a key physiological feature that distinguishes tumor tissue from normal tissue. However, the current identity of traceable molecular targets for hypoxic cancer cell survival remains a major challenge in cancer research [54]. The new hypoxia-related signaling pathway is a key mediator in the adaptive response process induced by hypoxia and a potential target for tumor therapy. Hypoxia can cause genomic instability in tumor cells and result in poor efficacy of conventional tumor treatments such as radiotherapy and chemotherapy. Therefore, how to control and alleviate hypoxia in tumor tissue is a focused research field in tumor treatment [55]. Phosphorylation plays an important role in hypoxia-related pathways; for example, PHD2 is known as a “low oxygen sensor,” which is highly dependent on the amount of oxygen. When it is phosphorylated, PHD2 becomes more active to promote cancer cell death in the hypoxic region of the tumor. Phosphatase PP2A/B55 can regulate the phosphorylation of PHD2, and one study designed molecules that blocked the function of PP2A/B55 phosphatase to produce anti-cancer effects [56]. The expression of HIF-1 would increase under hypoxic conditions in various tumors, and it affected the energy metabolism, angiogenesis, invasiveness, proliferation, resistance to radiotherapy and chemotherapy, and apoptosis of tumor cells by participating in the transcriptional regulation of multiple target genes, causing a series of reactions in cells and tissues to adapt to hypoxic environments [57]. HIF-1-targeted therapy might become an important means of cancer treatment. Our study found that phosphorylation played an important role in the HIF-1 signaling pathway in hepatic carcinoma. Some proteins in this pathway were found to be phosphorylated, including GAPDH, PRKCB, AKT2, CAMK2B, CAMK2D, ALDOC, ENO1, PIK3R1, PIK3R3, ALDOA, RPS6, SLC2A1, PLCG2, and FLT1. Tumor hypoxia is associated with many cancer “markers,” including impaired immune response, metabolic reprogramming, increased cancer stem cells, vascularization, invasion and metastasis, increased genomic instability, increased apoptosis, and reduced cell proliferation [58]. The sensitivity of tumor cells to hypoxia signaling pathway inhibitors varies greatly, and the molecular mechanisms that determine the sensitivity of tumor cells to inhibitors are still unclear, which limits the application of targeted hypoxia molecules in tumor therapy [59]. Therefore, a study on the mechanisms of sensitive genes and related phosphorylation that regulate hypoxia signaling pathways is of great significance for tumor treatment.

Cross-talking among different signaling pathways to form a complex phosphorylation-mediated regulatory network for primary hepatic carcinoma

A comprehensive analysis of all these signaling pathways found that cross-talking existed among different signaling pathways, thus forming a complex phosphorylation-mediated regulatory network for early-stage primary hepatic carcinoma. For example, glycolysis is the main energy supply method for malignant tumor cells, and tumor cells have a 20–30-fold higher ability to metabolize sugars than normal cells. Malignant tumor glycolysis produces a large amount of lactic acid to put the body in a high metabolic state [60]. At the same time, lactic acid enables tumor cells to survive under the microenvironment’s pH value, upregulating the expression of HIF-1, which might be related to tumor invasion and metastasis, leading to chemotherapy resistance [61]. In turn, high expression of HIF-1 promotes tumor cell glycolysis through feedback. Hypoxia can also initiate overexpression of glucose transporter proteins, alterations in oncogenes or tumor suppressor genes, PI3K/AKt/mTOR pathways, and overexpression of glycolytic enzymes, which jointly promote glucose uptake and hyperglycemia [62]. The role of the immune system in anaerobic glycolysis of tumor tissue can alter the energy supply of immune cells, such as NKTh1, which play a major killing role in tumor tissue, weakening their ability to kill tumor tissue, promoting immune evasion, and maintaining the growth and reproduction of tumor cells. The low immune function in patients might further reduce the immune defense against tumors [63]. During hypoxia, tumor cells can adapt to this unfavorable environment through various cellular mechanisms, making them more resistant and capable of survival. This is a common phenomenon in solid tumors. Recent studies have shown that immunosuppressive cells, tumor stem cells, and circulating tumor cells in the tumor hypoxia microenvironment can mediate immunosuppression and immune tolerance and promote the development of tumors [64].

Strength and limitations

The main strength of this study is as follows: First, quantitative phosphoproteomics based on TMT-labeling and TiO2 enrichment of phosphopeptides was used to detect, identify, and quantify phosphoproteins and phosphosites in early-stage primary hepatic carcinoma tissues (n = 6) compared to the paired tumor-adjacent normal control tissues (n = 6). In total, 858 DPPs with 1326 phosphosites were identified. These DPPs were enriched in 33 statistically significant signaling pathways, such as the spliceosome, glycolysis/gluconeogenesis, B-cell receptor signaling pathway, HIF-1 signaling pathway, and fatty acid degradation. These signaling pathways are also cross-talking to form a complex phosphorylation-mediated regulatory network for early-stage primary hepatic carcinoma. These resulting data are a solid and precious molecular resource for early-stage primary hepatic carcinoma because the cancer tissue samples (n = 6) and paired tumor-adjacent normal control tissue samples (n = 6) were obtained from the same hepatic carcinoma patients (n = 6), and tumor-adjacent tissue was the normal liver tissue around the edge of the tumor with a distance that was greater than 1 cm from tumor-adjacent tissue to the tumor margin envelope. Thus, the quality of quantitative phosphoproteomics data can be guaranteed because cancer tissues and control tissues were from the same patients and a sufficient distance (> 1 cm) was guaranteed from the tumor-adjacent control tissue to the tumor margin envelope. The bias factors between individuals were maximally overcome to affect the results of quantitative phosphoproteomics. Therefore, this set of DPP data is reliable, although only 6 paired samples were used. Second, the transcriptomics data of 858 DPPs were obtained from liver cancer tissue samples (n = 373) and non-paired tumor-adjacent normal liver tissue samples (n = 50) in TCGA database, and 417 of those 423 liver cancer patients had complete clinical data and survival data. The survival analysis of the corresponding gene of each DPP was performed among 417 liver cancer patients, which offered the clinical relationship of the identified DPPs with primary hepatic carcinomas. Third, the association analysis between OS-related DEGs and drug sensitivity was performed with Corrplot R package plus Spearman method (p < 0.05) based on the drug sensitivity data of 263 drugs among 60 tumor cell lines and transcriptomic data (n = 23,805 genes) among these 60 tumor cell lines from CellMiner (NCI-60 Analysis Tools) in hepatic carcinoma, which provided the potential drug sensitivity to primary hepatic carcinomas. Thereby, this study clearly provided phosphorylation-related prognostic biomarkers and potential therapeutic drugs for early-stage primary hepatic carcinomas in the frame of the PPPM approach.

However, one must clearly realize that some limitations remain in this present study. First, the sample size (cancers: n = 6; paired controls: n = 6) was small and was used for quantitative phosphoproteomics analysis. For the future deep investigation of the biological functions and molecular mechanisms of the key phosphoproteins and phosphorylation-mediated signaling pathways in early-stage primary hepatic carcinoma, a significantly expanded sample size will be needed to verify and confirm those results. Second, for the survival analysis of the identified DPPs that were from the Chinese race, the transcriptomics and clinical data from the public TCGA database were from the foreign race. To overcome race differences, it would be best to establish transcriptomics and clinical data from the Chinese race for the survival analysis of the identified DPPs. Third, although potential drugs sensitive to primary hepatic carcinomas were found in this study, it is necessary to further test the real efficiency and biological roles of these drugs against hepatic carcinoma in different levels of experiments, including cell models, animal models, and even clinical trials for hepatic carcinomas in the future. Fourth, precise molecular subtyping of hepatic carcinoma needs to use phosphoproteomics data from different stages of patients, and one should do more targeted analysis and mining of combined medication strategies based on phosphorylation signaling pathway network for patient stratification and molecular subtypes. Protein phosphorylation and dephosphorylation are reversible processes that rely on multiple kinases and phosphatases to achieve complex and precise biological regulation. In the future, we need to screen potential kinases, phosphatases, and phosphodiesterases based on identified DPPs and construct a kinase and kinase signal transduction network. Moreover, we also need to do further studies to detect and validate the functions of specific phosphorylation sites in a phosphoprotein and reveal how the phosphorylation process affects the phenotypes of hepatic carcinoma. Fifth, with the development of genomics, transcriptomics, proteomics, PTMomics, proteomics, bioinformatics, and other omics, cancer research has become more and more systematic and in-depth. Compared with the genome, the proteome is more complex in composition, more active in function, and closer to the essence of life’s activities. However, compared to single-omics data analysis, multiomics big data integration analysis provides a solid foundation for a deeper understanding of the pathogenesis of diseases and targeted drug research and development. Our study should further add multiomics big data to reveal functions and mechanisms of phosphorylation in 3P medical practice of hepatic carcinoma.

Conclusion and expert recommendation in the framework of PPPM

The application of phosphoproteomics has made significant progress in various fields. For cancer research, phosphoproteomics can study gene expressions, phosphorylation status abnormalities, and their relationship with cancer phenotypes. This study provided a specific and effective research method for types and distribution of phosphoproteins in human early-stage primary hepatic carcinoma tissues, which identified 858 DPPs with 1326 phosphosites, 33 statistically significant signaling pathways, 105 overlapped molecules between DPPs and DEGs, 125 OS-related DPPs, 14 common molecules among DPPs, DEGs, and OS-related DDPs, and sensitive drugs associating 26 OS-related DPPs. These research results can serve as the basic data for in-depth research on molecular mechanisms of phosphorylation regulation, the discovery of target drugs, and the construction of phosphorylation-related biomarkers in hepatic carcinoma.

Currently, major advances have been achieved in hepatic carcinoma [65, 66]. The international Barcelona Clinic Liver Cancer (BCLC) diagnosis, treatment, and prognosis strategy [65] (Supplementary Fig. 3) and the China liver cancer staging (CNLC), treatment, and prognosis strategy [66] were published in 2022, respectively. However, BCLC staging [65] (stages 0: very early stage; A: early stage; B: intermediate stage; C: advanced stage; and D: terminal stage) and CNLC staging [66] (stages Ia, Ib, IIa, IIb, IIIa, IIIb, and IV) are all mainly based on performance status (PS), the number and size of liver tubers, liver functions, and some biochemical parameters such as AFP, ALBI score, Child–Pugh, etc. The corresponding treatment effects and types of effective therapeutic drugs are still very limited, and survival time is still relatively short. The in-depth molecular changes at a systematic level remain unclear for hepatic carcinoma. Multiomics is an effective approach to achieving those molecular changes at a systematic level. Here, we recommend emphasizing a quantitative phosphoproteomics study in human primary hepatic carcinoma, which can reveal the phosphoproteomic alterations, phosphorylation-mediated signaling pathway network changes, and kinase profile abnormalities in primary hepatic carcinomas compared to normal liver tissues. These data, in combination with other omics such as transcriptomics and clinical data, offer great promise for in-depth insight into the molecular mechanisms of primary hepatic carcinoma, the discovery of effective therapeutic targets/drugs, and the construction of reliable phosphorylation-related biomarkers for predictive diagnosis, prognostic assessment, and personalized treatment of primary hepatic carcinoma in the framework of 3P medicine. The PPPM innovation is present in the following three aspects.

  1. (i)

    Predictive approach. Phosphorylation is an important protein modification in the human body, and its abnormal changes exist in the entire pathophysiological process of primary hepatic carcinoma, crossing the occurrence and development of primary hepatic carcinoma. DPPs, in combination with DEGs and clinical data, can construct a phosphorylation-related predictive diagnosis model and prognostic assessment model to innovate the predictive approach of primary hepatic carcinoma. For example, one study constructed a prognosis-related histone phosphorylation-regulated (HPR) gene signature and elucidated whether HPR genes can predict overall survival in hepatocellular carcinoma patients. The HPR gene risk score is closely related to the prognosis of HCC, tumor immune process, and tumor cell progression [67]. A large number of phosphorylation databases have been established based on existing high-throughput and low-flux analysis data, consisting of a large amount of protein phosphorylation site information from humans, mice, and rats [68], such as Phospho.ELM (http://phospho.elm.eu.org/), which is a database of experimentally verified phosphorylation sites in eukaryotic proteins with 2166 tyrosine sites, 13,320 serine sites, and 2766 threonine sites, and PhosphoSitePlus™ (http://www.phosphosite.org/homeAction.do) that includes literature and Qualcomm Protein phosphorylation sites discovered during the development process, and currently this database contains 66,045 phosphorylation sites. This present study could continue to fill in disease-related DPPs and specific protein phosphorylation sites to enrich the existing database. Moreover, some databases provide a tool for understanding the kinase and substrate relationship in vivo, which helps to gain a deeper understanding of kinases and phosphorylated proteins. It is an important resource and gateway for the development of phosphorylation regulation networks such as NetworKIN (http://networkin.info/search.php).

  2. (ii)

    Targeted prevention. Phosphorylation-mediated signaling pathway network changes and their regulation by kinases and phosphatases benefit the deep understanding of the molecular mechanism of primary hepatic carcinoma and the discovery of effective therapeutic targets and drugs, which directly contribute to the targeted prevention and therapy of early-stage primary hepatic carcinoma. For example, Met tyrosine kinase, a receptor for a hepatocyte growth factor (HGF), plays a critical role in tumor growth, metastasis, and drug resistance. One study found that Met directly phosphorylated the outer mitochondrial membrane protein Fis1 at Tyr38 (Fis1 pY38). Fis1 pY38 promoted mitochondrial fission by recruiting the mitochondrial fission GTPase dynamin-related protein-1 (Drp1) to facilitate cell metastasis in hepatocellular carcinoma cells both in vitro and in vivo. These findings may provide a therapeutic target for metastatic hepatocellular carcinoma [69]. Currently, it is known that many human diseases are caused by abnormal phosphorylation. The research on the prevention and treatment of hepatic carcinoma with protein phosphorylase inhibitors has achieved remarkable results, including a marker drug for the targeted treatment of liver cancer, sorafenib (a representative of tyrosine kinase inhibitors), but more clinical trials are needed to confirm new findings [70]. For example, brivanib is a small-molecule tyrosine kinase inhibitor that inhibits vascular endothelial growth factor receptors and fibroblast growth factor receptors. Phase II clinical studies suggest that this drug is safe and tolerable, with a 6-month tumor progression-free survival rate of 18.2%. It is also a promising targeted drug for liver cancer treatment [71]. The phase II clinical trial of another MEK tyrosine kinase inhibitor, seluminib (AZD6244), found that its efficacy was average during mid-term analysis. Although the side effects of this drug were acceptable and inhibition of ERK phosphorylation in tumor tissue was observed, no imaging relief was observed in the patient [72]. Now, the first-line drugs of small-molecule tyrosine kinase inhibitors that have been listed are only sorafenib and lovatinib, and the second-line drugs are regofinib and cabotinib in hepatic carcinoma. As a new class of drugs, protein phosphorylase inhibitors are promising to open up a new way for the prevention and treatment of hepatic carcinoma.

  3. (iii)

    Personalization of medical services. Integrative analysis of DPPs, DEGs, and clinical data can establish phosphorylation-related molecular biomarkers to stratify patients with primary hepatic carcinoma and personalize predictive diagnosis, prognostic assessment, and treatment of primary hepatic carcinoma to realize personalized medical services. For example, one used isobaric tags for relative and absolute quantitation (iTRAQ)-based quantitative phosphoproteomics approach to identify biomarkers associated with hepatoma recurrence/metastasis in hepatoma cell lines with increasing metastasis ability. In total, 75 phosphopeptides corresponding to 60 phosphoproteins were significantly dysregulated, and the biological processes that these phosphoproteins participated in were tightly associated with tumor metastasis [73]. In terms of phosphorylation of insulin receptor substrate 1-IRS1 (ratio T/N = 1.80), the phosphorylation of IRS1 at preneoplastic stages during experimental hepatocarcinogenesis and its protective effect against apoptosis suggest that IRS1 contributes to liver tumor progression [74]. In terms of phosphorylation of 5′-AMP-activated protein kinase catalytic subunit alpha-1-AMPK (ratio T/N = 0.65), AMPK is dysfunctional in patients with hepatic carcinoma, and low p-AMPK staining is correlated with aggressive clinicopathologic features and poor prognosis, as determined by immunoblotting and immunostaining in specimens from 273 patients with hepatic carcinoma [75]. These findings may lead to new biomarkers: the phosphorylation state of cancer-related proteins may help to understand the tumor transformation process and therefore choose the appropriate treatment.

In summary, quantitative phosphoproteomics can identify phosphoproteins, phosphosites, and phosphorylation-mediated signaling pathway changes, which is an innovative area in the research field of early-stage primary hepatic carcinoma. Furthermore, integrative analysis of DPPs, DEGs, and clinical data reveals phosphorylation-based survival, hub molecules, and sensitive drugs against the key DPPs in early-stage primary hepatic carcinoma toward predictive diagnosis, targeted prevention, and personalized medical service. This present study demonstrated an innovative, phosphorylation-based state-of-the-art, contributing to the paradigm shift from reactive medicine to PPPM in human primary hepatic carcinoma.