Comparative Proteomic Analysis to Investigate the Pathogenesis of Oral Adenoid Cystic Carcinoma.

Adenoid cystic carcinoma (ACC) belongs to salivary gland malignancies commonly occurring in an oral cavity with a poor long-term prognosis. The potential biomarkers and cellular functions acting on local recurrences and distant metastases remain to be illustrated. Proteomics is the core content of precision medicine research, which provides accurate information for early detection of cancer, benign and malignant diagnosis, classification and personalized medication, efficacy monitoring, and prognosis judgment. To obtain a comprehensive regulation network and supply clues for the treatment of oral ACC (OACC), we utilized mass spectrometry-based quantitative proteomics to analyze the protein expression profile in paired tumor and adjacent normal tissues. We identified a total of 40,547 specific peptides and 4454 differentially expressed proteins (DEPs), in which HAPLN1 was the most upregulated protein and BPIFB1 was the most downregulated. Then, we annotated the functions and characteristics of DEPs in detail from the aspects of gene ontology, subcellular structural localization, KEGG, and protein domain to thoroughly understand the identified and quantified proteins. Glycosphingolipid biosynthesis and glycosaminoglycan degradation pathways showed the biggest difference according to KEGG analysis. Moreover, we confirmed 20 proteins from the ECM-receptor signaling pathway by a parallel reaction monitoring quantitative detection and 19 proteins were quantified. This study provides useful insights to analyze DEPs in OACC and guide in-depth thinking of the pathogenesis from a proteomics view for anticancer mechanisms and potential biomarkers.

Introduction

Adenoid cystic carcinoma (ACC) is an uncommon but aggressive malignancy. It often originates in the epithelial cells of mucous-secreting glands, particularly the salivary glands.[1,2] It accounts for 10% of all salivary gland neoplasms,[3] 22% of all salivary gland malignancies,[4] and 1% of the head and neck malignancies with a median/mean age between 47 and 56.[1] The solid, cribriform, and tubular are three major histological subtypes of ACC, with the oral cavity being the most frequent incidence location.[5] Oral ACC (OACC) frequently masquerades as a benign neoplasm with an early but not in-depth research, which even accounted for 34.5% of all head and neck examples.[6] The characters of ACC are the variable clinical patterns and protracted clinical course. Despite growing slowly, ACC shows a poor long-term prognosis (the 15- or 20-year survival is about 23 to 40%) due to the perineural invasion, aggressive nature, and high risk of recurrence.[7] Compared with other tumors, ACC develops slowly and spreads systemically, considering its high possibility of distant metastases. Large tumor size and locoregional treatment were reported to be predictive for distant metastasis.[8] Multiple metastatic sites are known and the lung is the most commonly involved.[9] As for ACC, surgical resection is the primary treatment option combined with post-operative radiotherapy.[10] Meanwhile, local and distant recurrence exists extensively.[11] Owing to the rarity of this cancer and the lack of in-depth mechanism research, its pathogenesis is still unclear and the detailed studies to investigate prognostic factors are needed even for multiple analysis.

Precise treatment of cancer has always been one of the most concerned research fields. Tumor genome analysis can provide guidance information for patients to choose treatment individually. However, based on clinical statistical observation and genomic sequencing, only a small number of patients can really benefit from the predictive therapy. Recent studies have shown that proteome contains more comprehensive information than genome alone. This leads to the concept of proteomics, which offers a wide methodology to study biological systems and diagnostic biomarkers in the process of tumorigenesis and cancer development. In the past few decades, mass spectrometry (MS)-based shotgun proteomics technology has developed into a powerful tool for high-throughput detection and quantification of proteomes in complex samples.[12,13] The multiple peptides decomposed from protein mixture could be analyzed by a mass spectrometer. Then, some peptide segments are selected for re-fragmentation (secondary mass spectrometry) to get smaller amino acid sequences. According to the secondary mass spectrum results, the retrieval software matches the peptides in the corresponding database to get the exact sequence and then splices into the complete protein sequence.[14,15] More mature technologies are needed to provide powerful tools for high-throughput proteomic analysis.

The peptide sequencing method containing liquid chromatography (LC) and MS is a major technology widely used not only in qualitative protein identification but also in the analysis of protein complex, cellular pathways, complex post-translational modifications, organellar protein compositions, and in vivo comprehensive biomarker discovery.[16,17] Compared with the traditional shotgun method, qualitative proteomics has the advantages of clear identification target, good quantitative repeatability, and high detection sensitivity. It can be used in a variety of biological samples, including serum or plasma, urine secretions, frozen tissues, archive samples (formalin), as well as immortalized and primary cells. Blood is a common material for proteomic analysis to screen and identify cancer biomarkers.[18] Meanwhile, a fresh frozen tissue is considered a better choice for MS-based proteomics analysis for disease-specific molecular illustration.[19] Parallel reaction monitoring (PRM) is the mainstream method of targeted proteomics for data collection with the advantages of high specificity, high sensitivity, high throughput, and antibody independence.[20] Through the selective detection of target peptides, it can achieve the relative or absolute quantification of target proteins/modified peptides. Parallel accumulation serial fragmentation (PASEF) is a new generation of 4D proteomics, which can be combined with PRM to achieve enhanced label-free quantitative proteomics. TimsTOF can complete proteomic analysis more quickly, sensitively, and stably with its trapped ion mobility spectrometry (TIMS) technology.[21] PASEF enables ion accumulation, mobility separation, and quadrupole isolation in the mobility analyzer to proceed simultaneously.[22,23] Therefore, the sensitivity and resolution of timsTOF are not lost when the scanning speed is higher than 100 Hz. The combined 4D proteomics technologies can break the new record of data collection speed and bring higher sensitivity and speed in proteomics area.

In this study, we utilized a quantitative proteomics approach combined with LC/MS, PRM and PASEF to obtain enhanced quantitative information. Based on this 4D label-free technique, differentially expressed proteins (DEP) in OACC tumor and adjacent normal tissues were identified and analyzed by comprehensive bioinformatics. The ECM-receptor-related proteins were quantitatively confirmed by PRM to ensure the validation results are consistent with the omics data, and effectively avoid the interference of false-positive results. Because the proper materials from OACC tumors are difficult to obtain in clinics and the related research is still in an early stage, this study is helpful to reveal the possible molecular pathway in tumorigenesis and progression more comprehensively, find potential therapeutic targets more accurately, and provide clinical guidance more specifically for OACC diagnosis and treatment.

Results

Protein Identification in OACC Tissues

To investigate the pathogenesis of OACC with proteomic identification, we selected four OACC patients who had not received any treatment before surgery. The tumor tissues (OACC_T) and adjacent normal tissues (OACC_N) were surgically removed for the study. A total of 420,012 secondary spectra were obtained through MS in this project. After searching the protein theoretical data database, the number of available useful spectra was 80,659 and the spectrum utilization rate was 19.2%. The analysis identified a total of 41,985 peptides, of which the specific peptide number was 40,547. We identified a total of 5715 proteins, of which 4454 can be quantified (Figure a). Proteins that were upregulated by >1.5 fold or downregulated by <1/1.5 were considered as a significant difference. According to this standard, 1024 proteins were increased, and 940 proteins were decreased in the OACC_T compared with OACC_N (Figure b; Table S1). The top 20 upregulated and top 20 downregulated proteins based on OACC_T/OACC_N ratio are listed in Table . These proteins showed significant difference in quantitative tissues in the comparison between OACC_T and OACC_N. In addition, the expression levels of up- and downregulated proteins in tumors were more than 10- and 5-folds higher than adjacent normal tissues, respectively.

Figure 1

Protein identification. (a) Basic statistical diagram of mass spectrum results. (b) Histogram of quantity distribution of DEPs. (c) Gene ontology (GO) secondary classifications of the DEPs of tissues in OACC_T and OACC_N based on BP, CC, and MF. (d) Subcellular structure localization and classification of DEPs.

Table 1

Top 20 Upregulated and Downregulated DEPs

gene name	OACC_T/OACC_N ratio	regulated type	gene name	OACC_T/OACC_N ratio	regulated type
HAPLN1	420.496	Up	BPIFB1	0.013	down
FNDC1	348.467	Up	MYL3	0.025	down
LAMC2	119.919	Up	ACTA1	0.031	down
MDK	46.291	up	ACTN2	0.049	down
SBSN	38.308	up	MYBPC1	0.054	down
PXDN	33.137	up	BPIFB2	0.072	down
CLSTN1	31.1	up	MUC5B	0.078	down
THBS2	29.87	up	CKM	0.099	down
ITGA9	29.543	up	MYH2	0.099	down
MFGE8	22.717	up	TNNC1	0.107	down
COL7A1	22.678	up	MB	0.108	down
FABP7	21.936	up	CASQ1	0.133	down
MATN2	21.001	up	B3GNT3	0.133	down
SPARC	20.359	up	AGR2	0.139	down
LAMB3	19.446	up	PRR27	0.155	down
LAMA5	18.018	up	CA4	0.157	down
VCAN	17.68	up	TMEM41B	0.16	down
COL5A1	15.761	up	COX7A2L	0.167	down
GAS6	15.593	up	KIAA0513	0.169	down
CLEC11A	15.59	up	SLC35B2	0.169	down

Functional Classification of DEPs

We did a gene ontology (GO) secondary classification analysis of the DEPs, including a biological process (BP), a cellular component (CC), and a molecular function (MF). GO annotation proteome was derived from the UniProt-GOA database (http://www.ehi.ac.uk/GOA), and the upregulation and downregulation results are shown in an additional file (Figure S1a,b). In the BP classification, most of the proteins are related to cellular process, biological regulation, and metabolic processes. In the CC classification, most of the proteins are located in cell, organelle, and membrane-enclosed lumen. In the MF classification, most of the proteins are involved in the binding and catalytic activity (Figure c). Subcellular structure localization prediction and classification statistics of DEPs were performed and the subcellular structural changes were mainly observed in the nucleus, cytoplasm, and extracellular regions (Figure d). The upregulation and downregulation results of subcellular localization are shown in an additional file (Figure S1c,d).

Functional Enrichment Analysis of DEPs

To find the significant enrichment property of DEPs, we carried out enrichment analysis at three levels: GO classification, KEGG pathway, and protein domain. GO is used to analyze various properties of genes and gene products, explaining the biological roles of proteins from different perspectives. Regarding fold enrichment of proteins based on BP analysis, co-translational protein targeting to membrane, spliceosomal complex assembly, and RNA 3′-end processing were mostly enriched. For CC analysis, precatalytic spliceosome, U2-type spliceosomal complex, and spliceosomal snRNP complex showed the biggest fold change. Four-way junction DNA binding, serine-type carboxypeptidase activity, and 7S RNA binding had the same fold enrichment and were mostly enriched through MF analysis (Figure a; Table S2). KEGG is an information network connecting the known intermolecular interactions, in which glycosphingolipid biosynthesis associated proteins were the most numerous. The KEGG-based enrichment revealed DEPs were mostly changed in glycosphingolipid biosynthesis—globo and isoglobo series (map00603), glycosphingolipid biosynthesis—ganglio series (map00604) and glycosaminoglycan degradation (map00531) (Figure b). The results also identified 25 pathways from upregulated DEPs and 34 pathways from downregulated DEPs. The most upregulated proteins focused on glycosaminoglycan biosynthesis—chondroitin sulfate/dermatan sulfate (map00532) and glycosaminoglycan degradation (map00531). Meanwhile, protein export (map03060), mucin type O-glycan biosynthesis (map00512), and collecting duct acid secretion (map04966) showed the biggest fold enrichment in downregulated proteins (Table S3). The protein domain refers to some components that repeatedly appear in different protein molecules, which have a similar sequence, structure, and function. It is the unit of protein evolution. The length of protein domains detected was between 25 and 500 amino acids. The data identified the top 20 differentially significant protein domains in DEPs, in which MCM2/3/5 family, MAC/Perforin domain, and Zinc carboxypeptidase enriched most (Figure c; Table S4).

Figure 2

Functional enrichment analysis. (a) GO-based enrichment analysis of DEPs. The red bars indicate a negative value of log10 (Fisher’s exact test p value). KEGG pathway-based enrichment analysis (b) and protein domain enrichment analysis (c) of DEPs. The top 20 signal pathways with the most significant enrichment are showed by a bubble chart. The color of the circle indicates the p-value of significant enrichment, and the size of the circle indicates the differential protein number. DEPs, differentially expressed proteins.

Clustering Analysis

We divided the DEPs into four parts according to the quantitative OACC_T/OACC_N ratio: Q1 (<0.5), Q2 (0.5–0.667), Q3 (1.5–2), and Q4 (>2) (Figure a). The ratios 0.5 and 0.667 correspond to the downregulated by 2-fold and 1.5-fold (1/2 and 1/1.5), respectively. According to the p-value of the enrichment test obtained by enrichment analysis, we use the hierarchical clustering method to gather the related functions in different clusters. For example, the most significant enriched are post-embryonic eye morphogenesis, hemidesmosome assembly, and embryonic eye morphogenesis based on GO classification of BP in the Q4 cluster (Figure b). The enriched protein domains in four clusters are drawn as a heatmap, which horizontally represents the enrichment test results of different groups and vertically represents the description of enrichment-related functions expressed differently (Figure c). The DEPs in each cluster are listed according to GO-based enrichment analysis, KEGG pathway-based enrichment analysis, and protein domain enrichment analysis (Tables S5–S7).

Figure 3

Functional enrichment of clusters. (a) Distribution histogram of DEPs in Q1–Q4. (b) Cluster analysis bubble chart based on BP in GO classification of the Q4 cluster. (c) Cluster analysis heat map based on protein domains.

Confirmation of the Up-/Downregulated Protein of LC–MS/MS by PRM

To confirm the LC–MS/MS experiment’s result, we quantified 19 proteins in the ECM signaling pathway by PRM with the same samples (Table ). Cluster analysis of the KEGG pathway by differential protein expression fold showed that an ECM–receptor interaction was significantly enriched in the Q4 group (Figure ). PRM was quantified by the peak area. In the experimental design, each protein was quantified by more than two unique peptides, and only one peptide was identified for some proteins due to sensitivity and other reasons. We detected 19 proteins by PRM and got the same results as those detected by LC–MS/MS identification (Table S8).

Table 2

Ratio of OACC_T/OACC_N Detected by LC–MS/MS and PRM

protein accession	protein gene	OACC_T relative abundance	OACC_N relative abundance	OACC_T/OACC_N ratio	OACC_T/OACC_N ratio (LQ)
O15230	LAMA5	1.87	0.13	13.96	18.02
P98160	HSPG2	1.49	0.51	2.93	3.67
P07942	LAMB1	1.83	0.17	10.58	7.02
P11047	LAMC1	1.85	0.15	12.69	5.64
P05556	ITGB1	1.39	0.61	2.29	2.61
P16144	ITGB4	1.48	0.52	2.86	2.60
O00468	AGRN	1.74	0.26	6.65	5.00
P35442	THBS2	1.96	0.04	45.87	29.87
P24821	TNC	1.75	0.25	7.10	5.07
Q16787	LAMA3	1.86	0.14	13.21	12.62
P23229	ITGA6	1.44	0.56	2.59	2.39
P17301	ITGA2	1.55	0.45	3.47	7.30
Q13751	LAMB3	1.76	0.24	7.23	19.45
P02452	COL1A1	1.23	0.77	1.59	1.97
P16070	CD44	0.77	1.23	0.62	0.64
Q16363	LAMA4	1.23	0.77	1.61	1.56
Q13753	LAMC2	1.88	0.12	15.01	119.92
P35443	THBS4	1.28	0.72	1.77	2.84
P25391	LAMA1	1.90	0.10	19.08	12.79

Figure 4

Differentially quantified proteins in the ECM signaling pathway. (a) Fisher’s exact test p-value in Q4 of KEGG pathway enrichment. (b). ECM–receptor interaction signaling pathway.

Discussion

ACC is a malignant tumor with epithelial and myoepithelial cells originating from the mandible, sublingual, and salivary glands,[2] which accounts for 1% of all malignant tumors of the head and neck region.[24] Oral ACC (OACC) occurs in oral salivary glands, and the boundary between tumor tissue and surrounding healthy tissue is unclear, which make the early detection, early diagnosis, and early treatment the key points of diagnosis and treatment. Because of its special biological and clinicopathological characteristics, OACC is not sensitive to radiotherapy and chemotherapy. Then, identifying protein biomarkers is crucial for diagnosis, treatment, and pathogenesis.

Clinical proteomics is still the preferred tool for cancer biochemical research and identification of differentially expressed biomarkers. An instrument is the best tool for individualized medical treatment. In precision medicine/personalized medicine, high-throughput and high-sensitivity MS technology is needed. The LC–MS/MS-based PRM combines the high selectivity of quadrupole with the high resolution and high precision of Orbitrap, which can identify the secondary spectrum independently and make the method flow more convenient. This method can achieve better anti-interference ability and detection sensitivity in a complex background and is a new targeted proteome detection method with more advantages and potential. In addition, the rapid scanning property of PASEF can obtain more data acquisition points at a short time to produce MS spectra with higher specificity and greatly improve the detection sensitivity.[21] Different from the traditional methods, ion mobility only depends on the electric field force; trapped ion mobility MS (TIMS) uses the synergistic effect of electric field force and air flow to obtain higher mobility separation in a smaller mobility device. TimsTOF innovatively uses a dual TIMS separation/enrichment device.[21] The ions are accumulated in the first part and then separated according to the mobility in the second part. The separated ions continue to be used for MS/MS fragmentation. This step is then repeated to achieve nearly 100% ion utilization.[25] The combination of multiple techniques can maintain the ultrahigh resolution of MS, obtain higher sensitivity, greatly increase the sequence coverage, and detect more post-translational modifications and low abundance peptides. This will greatly promote the further study of proteomics and bring more possibilities for clinical large cohort studies.

We performed proteomic analysis of four OACC and adjacent normal tissues by LC–MS/MS in this work and tested the mixed four samples of multiple individuals. Unfortunately, there was no justification and no replicates for the statistics because of the limitation of the number of samples. The fold change of each protein is calculated from the related value in OACC_T divided by that in OACC_N. We designed this experiment as an exploratory work and to focus on the diverse proteins and their biological functions, even the pathways they participated in, so we selected one sample to represent OACC_T and OACC_N, respectively. As stated in previous studies, MS-based proteomics should strive to maintain accuracy within a range of 1.3- to 2-folds. Then, we chose the 1.5-fold change as an acceptable threshold of the statistically significant difference.[26−33]

The analysis identified a total of 41,985 peptides, of which the specific peptide was 40,547. We identified a total of 5715 proteins, of which 4454 can be quantified. Hyaluronan and Proteoglycan link protein 1 (HAPLN1) showed the most significant upregulated between OACC_T and OACC_N (ratio: 420.496). HAPLN1 links proteoglycan with hyaluronic acid, thereby establishing a growth factor binding platform,[34] which has been reported to play a role in cell adhesion and extracellular matrix structure conformation.[35] The loss of HAPLN1 in the aged tumor microenvironment leads to disruption of ECM cross-linking,[36] thus promoting the migration of melanoma while limiting intravasation of tumor-infiltrating lymphocytes.[37] HAPLN1 is associated with liver cancer occurrence. In a study of 80 HCCs with respect to 307 nontumor hepatitis/cirrhosis livers, much higher levels of HAPLN1 were found in HCCs than in nontumor livers in their cohort.[38] HAPLN1 also regulates cell growth in developing cartilage and heart valves.[39,40] HAPLN1 has been associated with colorectal cancer[41] and bad outcome in pleural mesothelioma.[42] Fibronectin (FN) type III domain containing 1 (FNDC1) is the second most significant upregulated protein between OACC_T and OACC_N (ratio: 348.467). FNDC1 is a protein-coding and disease-related gene,[43,44] which contains the conserved FN type III domain of FN. FN is an important extracellular matrix (ECM) protein and a well-known tumor regulator. It is involved in cell proliferation, migration, and invasion of a variety of human tumors, such as renal cancer,[45] cholangiocarcinoma,[46] breast cancer,[47] and gastric cancer.[48] Laminin γ2 (LAMC2) is a subunit of heterotrimeric glycoprotein laminin 332, which can be used as a top mesenchymal marker.[49] It belongs to a vital protein in the ECM-receptor interaction pathway. In our detection, it has a tremendous difference between OACC_T and OACC_N (ratio: 119.919) and is also identified to show top difference by PRM testing. LAMC2 is an epithelial-specific basement membrane protein, which had been reported to promote cell invasion and migration by mediating EGFR activation and inducing epithelial–mesenchymal transition (EMT).[50] LAMC2 can upregulate the expression of the mesenchymal marker vimentin (VIM) in LUAD cells to enhance cell proliferation and migration.[51] The overexpressed LAMC2 associated with tumor growth, EMT, chemoresistance, and poor survival through modulating the extracellular acidic microenvironment in pancreatic ductal adenocarcinoma patients.[52,53] The promotion role of LAMC2 in cancer invasion reveals the possible reason of the high recurrence rate in OACC.

In addition, bactericidal/permeability-increasing fold-containing protein B1 (BPIFB1) was identified to be the most downregulated DEPs in OACC tumor (ratio: 0.013). BPIFB1 belongs to a highly conserved protein family and is considered to play regulation roles in innate immune response, especially in lung-specific autoimmunity.[54,55] Its abnormal expression had been found to be associated with nasopharyngeal, gastric, salivary gland, and lung cancer.[56] Mechanically, BPIFB1 increased the radiosensitivity and suppressed tumor migration through the direct interaction and inhibition with vitronectin and VIM.[57,58] The function of BPIFB1 in tumor inhibition needs further exploration and represents a novel therapeutic strategy by targeting BPIFB1-interacted oncogenes.

Our data in the KEGG pathway-based annotation analysis revealed that DEPs were significantly enriched in 20 pathways. ECM–receptor interaction was significantly enriched in the Q4 group, which may be a major reason for the local and distant recurrence. ECM–receptor interaction is the most abundant gene signal transduction pathway, which plays an important role in the process of tumor shedding adhesion degradation movement and proliferation. It was upregulated in prostate cancer[59] and involved in the invasion and metastasis of gastric cancer.[60] The study showed that the ECM might promote the development of EMT in the colorectal cancer.[61] The main pathological features of glioblastoma are abnormal neovascularization and diffuse infiltration, making it the fatal brain tumor in adults. Studies revealed that the interaction between EMT and glioblastoma microenvironment plays an important role in this process.[62]

Prevention of distant metastasis and recurrence is crucial for the treatment of ACC. The high expression of ECM proteins in ACC may provide new ideas for treatment strategies. We believe that these genes and pathways may be potential biomarkers of ACC, although their occurrence and development mechanism need further experimental verification.

Materials and Methods

Tissue Specimen

Four ACC tissues and adjacent normal tissues were obtained from Shenzhen People’s Hospital. The study was approved by the Medical Ethics Committee of Shenzhen People’s Hospital (no. LL-KY-2019173) and was carried out following the Declaration of Helsinki. The four participants signed an informed consent that allowed for the researchers to use their tissue during the tumor resection and conduct the study accordingly. A part of each tissue sample was prepared into paraffin wax for pathological diagnosis. The rest of the tissue was stored in the ultralow temperature freezer (−80 °C) for this study. The clinical characteristics were collected (Table ).

Table 3

Clinical Information of OACC and Normal Tissue Samples

sample ID	sex	age (years)	smoking	pathological diagnosis	tissue samples
1	female	32	no	right submandibular ACC	tumor tissue; tumor-adjacent normal tissue
2	female	23	no	palatal ACC	tumor tissue; tumor-adjacent normal tissue
3	male	58	no	left submandibular ACC	tumor tissue; tumor-adjacent normal tissue
4	male	64	no	left parotid gland ACC	tumor tissue; tumor-adjacent normal tissue

Protein Extraction

Four OACC tissue samples (n = 4; 200 mg per patient) were mixed as an OACC_T sample (800 mg), and four corresponding tumor-adjacent control tissue samples (n = 4; 200 mg per patient) were mixed as an OACC_N sample (800 mg). The fold change of each protein is calculated from the related value in OACC_T divided by that in OACC_N. Liquid nitrogen was added to grind the sample into powder. Each sample was added with four volumes of lysis buffer (1% Triton X-100, 1% protease inhibitor, 50 μM PR-619, 3 μM TSA, 50 mM NAM) followed by three times of ultrasonic lysis. The supernatant was then centrifuged at 12,000g in 4 °C for 10 min and was transferred to the new centrifuge tube. The protein concentration was determined using the BCA kit.

Trypsin Digestion

Each sample was taken in equal amounts for enzymatic digestion, and the volume of each sample was adjusted with the lysis buffer. The 20% TCA was added slowly and mixed with vortex to precipitate for 2 h at 4 °C. After the centrifuge at 4500 g for 5 min, the supernatant was discarded and the precipitate was washed 2–3 times with pre-cooled acetone. A final concentration of 200 mM TEAB was added to the pellet, and the residue was dispersed ultrasonically. A ratio of 1:50 trypsin (protease: protein, m/m) was added and hydrolyzed overnight. Dithiothreitol (DTT) was added in a final concentration of 5 mM to reduce the protein solution at 56 °C for 30 min. Iodoacetamide (IAA) was added to make the final concentration of 11 mM and incubated in the dark for 15 min at room temperature.

LC-Tandem MS Analysis

The peptides were dissolved in liquid chromatography phase A and then separated with a NanoElute ultrahigh performance liquid phase system. Phase A was an aqueous solution containing 0.1% formic acid and 2% acetonitrile. Phase B contains 0.1% formic acid and 100% acetonitrile solution. The liquid phase gradient was set as follows: 6–24% phase B for 0–70 min; 24–35% phase B for 70–84 min; 35–80% phase B for 84–87 min; and 80% phase B for 87–90 min with a 450 nL/min flow rate. The peptide segments are separated by an ultrahigh performance liquid system and then ionized by injecting into the capillary ion source. The peptide segments are analyzed by tims-TOF Pro MS. The ion source voltage was set as 2.0 kV, and the parent ion of the peptide segment and its secondary fragments were detected and analyzed by high-resolution TOF. The m/z scan range of secondary MS was set to 100–1700. The data acquisition mode is parallel cumulative serial fragmentation (PASEF) mode. After a first-stage MS acquisition, secondary spectrographs with the parent ions charge in the range of 0–5 were collected in PASEF mode for 10 times. The dynamic elimination time of tandem MS scanning was set at 30 s to avoid repeated scanning of the parent ions.

Database Search

Maxquant 1.6.6.0 database was used to retrieve the secondary MS data and set the retrieval parameters: the database was Homo_sapiens_9606 (20,366 sequences), an inverse library was added to calculate the false positive rate caused by random matching, and a common pollution library was added to the database to eliminate the influence of contaminated proteins in the identification results. The enzyme digestion method is set as Trypsin/P. The number of missing bits is set to 2. The minimum length of the peptide segment was set as seven amino acid residues. The maximum modification number of the peptide segment was set as 5. The first search and main search primary parent ion mass tolerance are set to 20.0 and 20 ppm, respectively, and the second fragment ion mass tolerance is 20.0 ppm. Cysteine alkylation (Carbamidomethyl (C)) is set as fixed modification and variable modification to [“Acetyl (Protein N-term”, “Oxidation (M))”, “Deamidation (NQ)”].

PRM Quantitative Proteomic Analysis

The peptides were injected into the NSI source, followed by MS analysis in Q Exactive Plus. The electrospray voltage was set at 2.1 kV and the peptide parent ion and its secondary fragments were detected and analyzed with high resolution Orbitrap. The scanning range of primary MS was set at 495–1175 m/z, and scanning resolution was set at 70,000. The scanning resolution energy of secondary MS was set at 27. The automatic gain control (AGC) of a primary mass spectrometer was set to 3E6, and the maximum IT was set to 50 ms. The AGC of secondary mass spectrometer was set to 1E5, the maximum IT was set to 120 ms, and the isolation window was set to 1.6 m/z. The peptide parameters were set as follows: protease was trypsin [KR/P]; maximum number was 0; peptide length was 7–25 amino acids; and cysteine alkylation was fixed modification. The transition parameters were set as follows: parent ion charge was 2,3; child ion charge was 1, and ion type was b, y. The fragment ion selection starts from the third to the last, and the mass error tolerance of ion matching was set to 0.02 Da.

Data Analysis

The change threshold of differential expression over 1.5 was considered as upregulated, and less than 1/1.5 was significantly downregulated. Fisher’s exact test was used to obtain the functional classification and significant enrichment pathways of different proteins. Differential protein immunofluorescence intensity was analyzed using SPSS 8.0 software and Origin 8.5 statistics. The mean ± standard deviation (SD) was presented. Any significant differences were determined by variance analysis (P ≤ 0.05).