Introduction

Breast cancer is a major global health concern, representing a significant proportion of cancer cases in women1. It typically originates in breast tissue, specifically affecting the inner lining of milk ducts or the lobules2. Breast cancer can be classified into different subtypes on the basis of hormone receptor status and genetic characteristics, resulting in diverse pathophysiological and clinical profiles3. The etiology of breast cancer is multifactorial and involves both genetic and environmental factors4. Understanding the complexity of breast cancer is essential for developing effective prevention strategies, optimizing treatment approaches, and improving patient care.

Advances in high-throughput sequencing technologies have significantly enhanced our understanding of gene‒phenotype associations in tumor tissues. Comprehensive transcriptomic analyses of various tumor types have provided valuable insights into the mechanisms of tumor proliferation, invasion, and recurrence5,6,7. Despite this progress, it is important to note that the correlation between mRNA and protein abundance is only 0.48. This low correlation indicates that transcript levels alone are often insufficient to predict protein levels, complicating the understanding of the genotype‒phenotype relationship.

A comprehensive understanding of cancer development and progression requires in-depth proteomic research. Extensive proteomic data can ultimately guide the development of more effective therapeutic strategies, bridging the gap between transcriptomic data and clinical outcomes9. An Integrated analysis of transcriptome, proteomics and phosphor-proteomics clearly identified LIG1 as a triple-negative breast cancer chemotherapy-resistance and multicancer-type poor prognosis marker10. In a prospective study of deep-scale proteomics and post-translational modification (PTM) proteomics of breast cancer, researcher uncovered hypoacetylation leaded to increased activity of the glycolysis pathway in breast cancer tissue11.

PTMs are generally enzymatic modifications of proteins following protein biosynthesis that significantly impact cancer occurrence and progression by influencing key mechanisms such as signal transduction, protein stability, subcellular localization, protein interactions, and metabolic rates12,13. Although more than 300 types of PTMs are known14, only a few have been thoroughly investigated, including phosphorylation, acetylation, ubiquitination, and malonylation.

Phosphorylation, which occurs on serine, threonine, and tyrosine residues, is a common PTM involved in cell signaling, cell cycle control, and apoptosis15,16. It is one of the most extensively studied PTMs, with many proteins undergoing this modification. The phosphorylation of key proteins such as STAT317, RhoA18, and mTOR19 has been linked to cancer cell development, invasion, and drug resistance.

Malonylation is an emerging form of protein acylation that occurs on lysine residues20. Malonyl-CoA serves as a precursor for de novo fatty acid synthesis and a key inhibitor of fatty acid oxidation, as well as a precursor to malonylation21. Malonylation is closely related to energy metabolism, particularly fatty acid synthase and the TCA cycle22. In lipopolysaccharide (LPS)stimulated macrophages, malonyl-CoA increases the malonylation of GAPDH and stimulates GAPDH activity and mRNA binding capacity, which in turn promotes the production of proinflammatory cytokines23. In malonyl-CoA decarboxylase (MCD)-deficient diseases, elevated malonylation impairs mitochondrial function and fatty acid oxidation24. Malonylation of mTOR inhibits pathological angiogenesis, suggesting the potential for malonylation in tumor microangiogenesis25. NAD-dependent protein deacylase sirtuin-5 (SIRT5) demalonylates SDHA and TPI, contributing to drug resistance and recurrence in wild-type Kras colorectal cancer22. However, there is still a lack of research on the role of malonylation in breast cancer.

To further elucidate the mechanisms underlying breast cancer, we systematically integrated proteomics, succinylproteomics, and phosphoproteomics to identify critical features of proteins and posttranslational modifications. This comprehensive profiling could contribute to the development of novel biomarkers and targeted therapeutics.

Results

Overview of multilevel proteomics

To investigate the posttranslational modifications (PTMs) involved in breast cancer, we analyzed the changes in the whole-cell proteome (WCP), phospho-proteome, and malonyl-proteome of breast cancer tumor tissues and adjacent healthy tissues (Fig. 1a). All tumor samples and healthy tissues were distinguishable on the basis of the first two components (Fig. 1b, Supplementary Fig. S1a). Consistently, the correlation matrix of tumor and healthy tissues was significantly different, highlighting their distinct protein expression patterns (Fig. 1c, Supplementary Fig. S1b). To determine whether different PTMs target different proteins, we analyzed the overlap between proteins identified in the two PTM proteomes and WCP. As shown in Fig. 1d, approximately half of the proteins identified in the WCP did not contain PTM sites, whereas 461 proteins contained both PTMs.

Fig. 1
figure 1

Overview of multilevel proteomics. (a) Flowchart of three proteomes, created with BioGDP.com66; (b) PCA of WCP (n = 5); (c) PCC analysis of WCP (n = 5); (d) Venn diagrams of identified proteins in three proteomes, WCP: Whole-cell proteome; P: phospho-proteome; Ma: malonyl-proteome. T: breast cancer tumor tissues. N: adjacent healthy tissues.

Full size image

Profiles of deps in the whole-cell proteome

In the WCP, we identified a total of 7,459 proteins, with 1,630 notably upregulated and 787 significantly downregulated in tumor tissues (Fig. 2a and Supplementary materials). The top 5 upregulated proteins were S100A8, S100A9, S100P, SPRR1B, and AZU1, whereas the top 5 significantly downregulated proteins were ZBTB2, OXTR, PLIN1, IGSF1, and DHRS2 (Fig. 2b). Subcellular localization revealed that the upregulated proteins were located primarily in the cytoplasm (27.83%), nucleus (26.47%), and extracellular space (14.86%), whereas the downregulated proteins were located mainly in the extracellular space (33.33%), cytoplasm (18.58%), and nucleus (17.68%) (Fig. 2c). The changes in the subcellular localization of proteins between breast cancer tissues and adjacent healthy tissues may indicate significant alterations in cellular functions.

Fig. 2
figure 2

Characteristics of the differentially expressed proteins (DEPs). (a) Heatmap of DEPs; (b) Volcano plot of DEPs; (c) Subcellular localization prediction of DEPs, Up: Upregulated DEPs; Down: Downregulated DEPs.

Full size image

Functional enrichment analysis of deps

Next, we performed functional enrichment analysis on the differentially expressed proteins (DEPs). KEGG enrichment analysis (Fig. 3a) revealed that the DEPs were enriched in immune regulation and cell adhesion-related pathways (glycosaminoglycan biosynthesis-keratan sulfate, ECM-receptor interaction). Similarly, GO enrichment analysis (Supplementary Fig. S2) revealed that the differentially expressed proteins were enriched in immune signal regulation (leukocyte proliferation, T-cell binding, and antigen binding) and extracellular matrix regulation processes (extracellular matrix, integrin binding, and extracellular matrix structural constituents).

Fig. 3
figure 3

Functional analysis of DEPs. (a) KEGG pathway analyses of the DEPs; (b) Statistical analysis chart of four groups of DEPs, which are divided according to fold change. Q1 (FC < 0.5), Q2(0.5 < FC < 0.667), Q3(1.5 < FC < 2.0), and Q4(FC > 2.0); (c) KEGG pathway analyses of the four groups; (d) Biological processes of the four groups, blue represents high enrichment, white represents low enrichment, “*” p value < 0.05, “* *” p value < 0.01, “* * *” p value < 0.001.

Full size image

To further investigate the potential biological functions of these proteins, we divided the differentially expressed proteins into four groups on the basis of FC (Fig. 3b). Group Q1 (FC < 0.5) included 391 proteins, group Q2 (0.5 < FC < 0.667) included 396 proteins, group Q3 (1.5 < FC < 2.0) included 973 proteins, and group Q4 (FC > 2.0) included 657 proteins. Proteins in groups Q1 and Q2 were enriched mainly in the ECM‒receptor interaction, focal adhesion, and gap junction pathways, whereas proteins in groups Q3 and Q4 were enriched mainly in immune-related pathways such as the phagosome, lysosome, and infection pathways (Fig. 3c‒d). These findings suggest that DEPs with high fold changes may play important roles in remodeling of the tumor immune microenvironment.

Profiles of DPPs and motif analysis of phosphorylation sites

In the phospho-proteome, we identified a total of 4,870 proteins (569 upregulated and 787 downregulated) and 18,108 phosphorylation sites (787 upregulated and 1,154 downregulated) (Fig. 4a and Supplementary materials). The top 5 upregulated DPP included SNAP23, PDCD11, SNPH, TANC2, and TNKS1BP1, whereas the top 5 downregulated DPP included TRIM16, TUBA1B, SMC3, IRF2BPL, and ADD3 (Fig. 4b).

Fig. 4
figure 4

Characteristics of the differentially phosphorylated proteins (DPPs). (a) Statistical analysis of DPPs and phosphorylation sites; (b) Volcano plot of DPPs; (c) Motif analysis of phosphorylation from positions − 6 to + 6 (the phosphorylation sites were set to zero). Left: serine phosphorylation site. Right: Threonine phosphorylation site; (d) Subcellular localization prediction of DPPs, Left: Upregulated DPPs; Right: Downregulated DPPs.

Full size image

The phosphorylation sites exhibited a certain degree of conservation, and we used Motif-X to analyze the frequency of sequence motifs around the modification site (Fig. 4c). For serine phosphorylation sites, proline (P) and glutamic acid (E) were notably enriched upstream and downstream of the modification site. Arginine (R) substantially increased in frequency from positions − 6 to -1, whereas aspartic acid (D) and serine (S) were considerably more frequent downstream. For threonine phosphorylation sites, proline (P) and serine (S) were enriched both upstream and downstream, whereas phenylalanine (F) and leucine (L) were notably infrequent. These findings suggest a preference for certain kinases in breast cancer cells. Then, we predicted the subcellular localization of the DPPs (Fig. 4d). Both upregulated and downregulated proteins showed similar localization patterns, primarily in the nucleus (up 51.49%, down 50.91%), followed by the cytoplasm (up 21.79%, down 21.3%), plasma membrane (up 10.37%, down 9.35%), and mitochondria (up 4.75%, down 5.45%). More than half of the DPPs are localized in the nucleus, indicating that phosphorylation may play a significant role in the nucleus. Protein phosphorylation represent a mechanism that is frequently employed by cells to regulate transcription factor activity26. Phosphorylation of CREB at Ser133 in response to cAMP stimulus is sufficient to induce target gene expression27. The nuclear factor-kappa B (NF-κB) family of transcription factors has recently emerged as a major regulator of the growth and elaboration of neural processes. NF-κB is activated by a variety of extracellular signals, and either promotes or inhibits growth depending on the phosphorylation status of the p65 NF-κB subunit28.

In our study, we found some transcription factors, such as CREB1, STAT3, FOXA1, were DPPs. For example, STAT3 phosphorylation at tyrosine 705 (pSTAT3 Y705) was downregulated. STAT3 dimerization, nuclear translocation and transcriptional activity required for pSTAT3 Y70529.

Profiles of DMPs and motif analysis of Kma sites

In the malonyl proteome, we identified 107 differentially malonylated proteins (DMPs), with 116 upregulated and 110 downregulated malonylation-lysine (Kma) sites (Fig. 5a and Supplementary materials). Top upregulated DMPs included EZR, DDTL and EIF3CL, whereas the top downregulated DMPs included COL1A1, CAVIN2, FBN1, ANXA2 and APOD (Fig. 5b). Notably, the EZR had multiple Kma sites among these DMPs.

Fig. 5
figure 5

Characteristics of the DMPs. (a) Statistical analysis of differentially malonylated proteins (DMPs) and malonylation sites; (b) Volcano plot of DMPs; (c) Motifs analysis of malonylation from positions − 10 to + 10 (the malonylation sites were set to zero); (d) Subcellular localization prediction of DMPs, up: Upregulated DMPs; Down: Downregulated DMPs.

Full size image

Motif analysis revealed that alanine (A), glycine (G), and valine (V) were enriched around the Kma sites, whereas proline (P), arginine (R), tryptophan (W), and glutamine (Q) were less common near the Kma sites (Fig. 5c). These findings indicate distinct amino acid preferences around Kma sites, suggesting that different malonylation enzymes may be involved in regulating the malonylation of breast cancer. The partial biochemical preferences of enzymes for specific substrates may be determined by the residues surrounding the modification sites30. These findings indicate distinct amino acid preferences around Kma sites, which may suggest different malonylation enzymes may be involved in regulating the malonylation of breast cancer.

Moreover, we predicted the subcellular localization of DMPs (Fig. 5d) and found that most DMPs were mainly located in the cytoplasm (up 45.31%, down 53.49%), nucleus (up 21.88%, down 11.63%), mitochondria (up 7.81%, down 13.95%) and endoplasmic reticulum (up 6.25%, down 4.65%). Reports suggest that Kma primarily regulates cellular glycolipid metabolic activities29,31,32. However, the cytoplasm is the main site of metabolism, which aligns with our localization prediction. Our localization prediction found that DMP was main located in the cytoplasm, and enrichment analysis showed DMPs were enriched in galactose metabolism, glycolysis metabolism and glycerolipid metabolism (Fig. 6d).

Fig. 6
figure 6

Statistical analysis chart of the four groups DPPs (a) and DMPs (b); KEGG pathway analyses of the four groups DPPs (c) and DMPs (d).

Full size image

Functional enrichment analysis of DPPs and DMPs

Similarly, we classified DPPs and DMPs into four groups on the basis of their fold changes (Fig. 6a-b). Clustering analysis (Fig. 6c) further revealed that the DPPs in the Q1 (FC < 0.5) and Q2 (0.5 < FC < 0.667) groups were enriched mainly in the glycolysis/ gluconeogenesis, thyroid hormone synthesis, and insulin secretion pathways related to carbohydrate and lipid metabolism; the DPPs in the Q3 (1.5 < FC < 2.0) group were enriched in inositol phosphate metabolism, PPAR signaling pathway, and viral carcinogenesis; and the proteins in the Q4 (FC > 2.0) group were enriched in the base excision repair, beta-alanine metabolism, and fat digestion and absorption pathways.

DMPs in the Q1 group were enriched mainly in the ECM‒receptor interaction, focal adhesion, and PI3K‒Akt signaling pathways; those in the Q2 group were enriched in metabolic pathways such as galactose metabolism, glycolysis and glycerolipid metabolism; those in the Q3 group were not enriched in any pathway; and those in the Q4 group were enriched in base excision repair, NF-κB pathway, and pyrimidine metabolism (Fig. 6d). Given that tumor cells are characterized by mismatch repair defects and abnormal NF-κB activation, the Q4 group of DMPs might represent potential targets for breast cancer treatment.

Network analysis of PTMs

We performed PPI network analysis on the DPPs and DMPs. As shown in Fig. 7a, the DPP network consists of 1006 nodes and 4683 edges, whose hub proteins include ACTB, CTNNB1, EP300, HDAC1, and CREBBP. We found that kinases such as MAPK3 and PIK3R1, as well as the acetyltransferase EP300 and deacetylase HDAC1, interact with many DPPs. The DMP network, composed of 74 nodes and 122 edges, which are hub proteins, were identified primarily as enzymes, including TPI1, LDHA, and TKT. Interestingly, one of the hub protein phosphatases, PKM, presented increased malonylation levels (Fig. 7b). These results suggest potential crosstalk between phosphorylation and malonylation modifications.

Fig. 7
figure 7

Protein–protein interaction network of DPPs (a) and DMPs (b).

Full size image

Moreover, we performed Kinase-Substrate Enrichment Analysis (KSEA) for analysis kinase activity, which scores each candidate kinase based on the relative hyperphosphorylation or dephosphorylation of most substrates. KSEA results indicated marked activation of the kinases CSNK1D, ROCK1, ROCK2, and CDK2 and inhibition of CSNK2A2, PDPK1, CLK1, CSNK2A1, and SRC (Supplementary Fig. S3, Supplementary Table S2). The regulatory enzymes involved in the Kma modification are not yet fully understood. Our proteomics data revealed no significant difference in the total protein expression of the acetyltransferases EP300 and CBP or the deacetylases SIRT2 and SIRT5 between tumor and adjacent normal tissues. Malonylation levels were also unchanged. However, we observed decreased phosphorylation of acetyltransferase EP300 at S2309 and increased phosphorylation of SIRT2 at S368. However, the specific mechanisms and effects of these changes require further investigation.

Discussion

Proteomics has provided insights into breast cancer. Anurag et al. performed proteogenomics to explore the molecular basis for triple-negative breast cancer and reported the activation of DNA repair, E2F targets, the G2-M checkpoint, interferon-gamma signaling, and immune checkpoint pathways10. In another study, researchers performed proteomics on formaldehyde-fixed, paraffin-embedded (FFPE) samples of malignant tumors and normal breast tissue samples. They reported that the main categories of proteins upregulated in tumor samples were RNA-binding proteins, heat shock proteins, and DNA repair proteins33. In our study, we identified a total of 7,459 proteins and found significant upregulation of S100 family proteins (S100A8, S100A9, and S100P). The S100 family has been shown to play significant roles in immune responses34,35. A previous study reported that S100A8/A9 may promote breast cancer cell growth by remodeling the tumor immune microenvironment36. High expression of S100A8/A9 predicts poor prognosis in breast cancer patients37. Our functional enrichment analysis also revealed that the DEPs were significantly enriched in terms of immune regulation and the extracellular matrix.

The enrichment of these differential pathways may be closely related to significant changes in the tumor immune microenvironment (TIME) of breast cancer, indicating extensive immune cell infiltration in tumor tissues. Different types of infiltration have distinct effects on tumor tissues. For example, extensive infiltration of Treg cells hinders the host antitumor immune response and promotes tumor growth and metastasis38, whereas increased infiltration of helper TH1 cells and cytotoxic T lymphocytes can enhance the effectiveness of immunotherapy39.

Posttranslational modifications (PTMs) are ubiquitous and play crucial roles in various cellular functions40. Phosphorylation is a common, reversible, and highly regulated PTM of proteins. Studies have shown that the proto-oncogene Src promotes the development of breast cancer by directly binding and phosphorylating lipin-141. AXL stimulates the phosphorylation of a network of focal adhesion (FA) proteins, promoting tumor invasion42. Malonylation is a newly discovered PTM that has been shown to be involved in energy metabolism, macrophage inflammatory signaling, and chondrocyte metabolic regulation43. In mouse models of cancer cachexia, protein lysine malonylation was systemic declined44. Lysine malonylation of TKT inhibited its activity, the activation of TKT could accelerate CRC tumorigenesis in mice model45.

We performed PTM proteomics to investigate the characteristics of phosphorylation and malonylation in breast cancer and revealed its potential significance. There have been no reports about malonylation in breast cancer.

In the phosphorylation proteome, we identified a total of 4,870 phosphorylated proteins, among which PDCD11, SNPH, and TNAC2 presented significantly increased phosphorylation levels. PDCD11 can promote the G2/M transition in colorectal cancer (CRC) cells, facilitating tumor progression46. However, the mechanisms by which phosphorylation regulates PDCD11 function remain unclear and require further investigation.

Syntaphilin (SNPH) functions as an anchor, immobilizing mitochondria47. Phosphorylation by PAK5 allows SNPH to mobilize damaged mitochondria for replacement with healthy mitochondria, promoting neuron survival and regeneration48. The increased phosphorylation of SNPH in tumor cells may be related to a higher mitochondrial turnover rate. While the mitochondrial network is mobile and not evenly distributed throughout the cell49, researchers have reported that breast cancer cells with more mitochondria at their leading edge exhibit faster migration velocities50. Thus, we propose that phosphorylating SNPH to unanchor mitochondria might promote their accumulation at the leading edge of tumor cells, potentially regulating tumor invasion. This assumption requires further investigation.

Our KSEA results revealed significant upregulation of casein kinase 1 delta (CSNK1D) activity in breast cancer tissues. CSNK1D plays a tumor-promoting role in various cancers. Inhibition of CSNK1D can delay the growth of breast tumors51. Its substrates include TACC1, TACC2, XPO1, MTA2, and SP1. The transforming acidic coiled coil (TACC) family comprises tumor-promoting factors that are involved in the microtubule-dependent coupling of the nucleus and the centrosome. TACC2 is a prognostic predictor, with increased expression correlating with poor prognosis in breast cancer patients52,53. Phosphorylation is suggested as a mechanism for TACC activation54,55. In our study, TACC2 has 15 phosphorylation sites, with significant increases in phosphorylation at S2321 and S2512. These findings suggest that TACC2 may be a potential therapeutic target, with S2321 and S2512 as active sites.

In the malonylation proteomics analysis, 107 DMPs were identified, with 116 upregulated Kma sites and 110 downregulated Kma sites. Among these, Ezrin (EZR) has multiple Kma sites, suggesting that it may play a significant role in cellular processes. EZR, a member of the Ezrin-Radixin-Moesin (ERM) family, interacts with cell surface adhesion molecules such as CD44, E-cadherin, and ICAM-1, regulating cell morphology and migration56. EZR has been identified as an oncogene57 and is associated with high levels of metastatic behavior in various types of cancer56. Currently, there are no reports of Kma in the EZR. Many acyltransferases typically promote cancer progression20, and the acyltransferases largely overlap20. Moreover, five HDAC inhibitors are currently used in clinical tumor treatments22. We hypothesize that the malonylation of EZR also promotes breast cancer proliferation and invasion.

KEGG enrichment analysis revealed that phosphorylation and malonylation modifications primarily impact tumor cell glycolipid metabolism, ECM interactions, and mismatch repair. These findings indicate that phosphorylation and malonylation may be associated with typical tumor cell features, including glycolipid metabolism, increased de novo fatty acid synthesis, and mismatch repair deficiencies.

In summary, our study provides a proteomic overview of breast cancer and clarifies the changes in phosphorylation and malonylation in breast cancer tissues and paracancerous tissues. In addition, our study identified several novel protein markers that may have potential value in clinical diagnosis. Our study offers new insights into the complexity of breast cancer and lays the foundation for future research on breast cancer.

Materials and methods

Sample collection and clinical information

Fresh tumor samples were collected from breast cancer patients at the First Affiliated Hospital of Bengbu Medical University. Breast cancer tumor tissues and adjacent nontumorous tissues from five patients were used for proteomic analysis. This study was approved by the Bengbu Medical College Ethics Committee (No. 2021KY165), all methods were carried out in accordance with relevant guidelines and regulations, and all patients provided written informed consent. Supplementary Table S1 provides the basic clinical information of all patients.

Proteomics and PTM proteomics

The samples were ground into a powder via liquid nitrogen and then lysed in four volumes of lysis buffer (1% Triton X-100, 1% protease inhibitors, 3 µM TSA, 50 mM NAM). The lysates were sonicated (227 W, 2 s on, 3 s off, for 10 min) and then centrifuged at 12,000 × g for 10 min at 4 °C. The supernatant was transferred to a new centrifuge tube, and the protein concentration was determined via a BCA assay, adjusting to a uniform concentration. TCA was added to a final concentration of 20%, the mixture was vortexed, and the mixture was allowed to precipitate at 4 °C for 2 h. The mixture was centrifuged at 4500 × g for 5 min, the supernatant was discarded, and the pellet was washed with prechilled acetone 2–3 times. After air-drying the pellet, TEAB was added to a final concentration of 200 mM, the pellet was resuspended, and trypsin was added at a 1: 50 ratio overnight digestion. Dithiothreitol (DTT) was added to the peptides at a final concentration of 5 mM, and the mixture was incubated at 56 °C for 30 min. IAA was then added to the mixture at a final concentration of 11 mM, and the mixture was incubated in the dark at room temperature for 15 min.

Phosphorylated peptide enrichment

Peptide mixtures were first incubated with IMAC microsphere suspensions with vibration in loading buffer (50% acetonitrile/0.5% acetic acid). To remove the nonspecifically adsorbed peptides, the IMAC microspheres were sequentially washed with 50% acetonitrile/ 0.5% acetic acid and 30% acetonitrile/ 0.1% trifluoroacetic acid. To elute the enriched phosphopeptides, elution buffer containing 10% NH4OH was added, and the enriched phosphopeptides were eluted via vibration. The supernatant containing phosphopeptides was collected and lyophilized for LC‒MS/MS analysis.

Malonylated peptide enrichment

The peptide was dissolved in an IP buffer solution (100 mM NaCl, 1 mM EDTA, 50 mM Tris-HCl, 0.5% NP-40, pH 8.0). Then, the supernatant was transferred to prewashed antibody resin (PTM904, Jingjie Biotechnology) and placed on a rotating shaker at 4 °C with gentle shaking and an overnight incubation. After incubation, the resin was washed four times with IP buffer solution and twice with deionized water. Finally, the resin-bound peptide segment was eluted three times with 0.1% trifluoroacetic acid as the eluent. The eluent was collected and drained by vacuum freezing, followed by desalting according to the C18 ZipTips instructions before vacuum freezing and draining again for LC‒MS analysis.

LC‒MS/MS analysis

The peptides were dissolved in mobile phase A via liquid chromatography and separated via the Easy-nLC1000 ultrahigh-performance liquid phase system. The mobile phase consisted of solvent A (0.1% formic acid, 2% acetonitrile/in water) and solvent B (0.1% formic acid in acetonitrile). Peptides were separated with following gradient: 0–18 min, 2-22%B;18–22 min, 22-35%B; 22–26 min, 35-90%B; 26–30 min, 90%B. The peptide segment was subsequently injected into the capillary ion source for ionization and introduced into the timsTOF Pro mass spectrometer for data acquisition. The ion source voltage was set to 1.75 kV, and time-of-flight (TOF) analysis was employed to detect and analyze both the parent ions of the peptide segment and its secondary fragments. The data acquisition mode utilized is known as data-independent parallel cumulative serial fragmentation (DIA-PASEF) mode. After setting the scanning range of primary mass spectrometry from 300 to 1500 m/z, samples were collected in PASEF mode 20 times for each primary mass spectrometry scan. The full MS scan was set as100-1700 (MS/MS scan range) and 22 PASEF (MS/MS mode) -MS/MS scans were acquired per cycle. The MS/MS scan range was set as 395–1395 and isolation window was set as 20 m/z.

Database search

The DIA data were retrieved via the Spectronaut (v17) search engine with default parameters. The database utilized was “Homo_sapiens_9606_SP_20220107.fasta,” comprising 20,389 protein sequences. Trypsin/P was specified as the cleavage enzyme, allowing up to 2 missed cleavages. Carbamidomethylation of cysteine was designated as a fixed modification. Variable modifications included methionine oxidation (proteome, phospho-proteome), N-terminal protein acetylation (proteome, malonyl-proteome, phospho-proteome), lysine malonylation (malonyl-proteome), and phosphorylation of serine, threonine, and tyrosine (phospho-proteome). An inverse library was also incorporated to calculate the false discovery rate (FDR) resulting from random matching. The FDR for protein, peptide, and peptide-spectrum match (PSM) identification was set at 1%. Search results were filtered at 1% FDR, and the peptide confidence level was set for at least one unique peptides per protein for protein identification. The corresponding spectral library was imported into Spectronaut (v.17.0) software to predicts the retention time by nonlinear correction and searched against with DIA data.

Bioinformatics methods

The fold change (FC) was used to calculate the difference between tumor tissue (T) and adjacent nontumorous tissues (N). Fisher’s exact test was used to analyze the significance of functional enrichment of differentially expressed proteins. For phospho-proteome and malonyl-proteome, FC was calculated as the ratio of the mean relative quantification values for each modification site between two groups. The relative quantification values of the modification site are divided by the relative quantification values of its corresponding protein to eliminate the influence of protein expression on the modification expression. Differentially expressed proteins (DEPs), differentially phosphorylated proteins (DPPs) and differentially malonylated proteins (DMPs) were identified by an absolute value of FC > 1.5 and a p value < 0.05, which divided identified protein into 4 groups according to fold change, Q1 (FC < 0.5), Q2 (0.5 < FC < 0.667), Q3 (1.5 < FC < 2.0), and Q4 (FC > 2.0). Gene Ontology (GO) annotation was performed via eggNOG-mapper (v2.1.6)58. Subcellular localization prediction of the proteins was performed via WoLF PSORT and PSORTb (v3.0). Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis and protein‒protein interaction network analysis were conducted via Diamond software (v2.2.11.149)59,60,61,62,63, all differentially expressed protein database accession or sequence were searched against the STRING database for protein-protein interactions. Only interactions between the proteins belonging to the searched data set were selected, thereby excluding external candidates. STRING defines a metric called “confidence score” to define interaction confidence; we fetched all interactions that had a confidence score > 0.7 (high confidence).

Motif analysis utilized the MoMo analysis tool64, employing the motif-x algorithm, to examine the motif characteristics of modification sites. The analysis focuses on peptide sequences consisting of 10 amino acids both upstream and downstream of all identified modification sites, while for phosphorylation modifications, sequences of 6 amino acids upstream and downstream are utilized. The background for this analysis comprises peptide sequences that include 10 amino acids upstream and downstream of all potential modification sites within the species, with the same 6 amino acids upstream and downstream specification for phosphorylation modifications. A characteristic sequence form is deemed a motif of the modified peptides when the number of peptides exhibiting this specific sequence exceeds 20, and the p-value from the statistical test is less than 0.000001. Kinase-Substrate Enrichment Analysis (KSEA) was performed via the KSEA app65.

For further hierarchical clustering based on differentially expressed protein functional classification. We collated all the categories obtained after enrichment along with their p values, and then filtered for those categories which were at least enriched in one of the clusters with p value < 0.05. This filtered p value matrix was transformed by the function x = − log10 (p value). These p values were then clustered by one-way hierarchical clustering (Euclidean distance, average linkage clustering) in Genesis.