Comprehensive characterization of ubiquitinome of human colorectal cancer and identification of potential survival-related ubiquitination

Approved by the Ethics Committee of Shenzhen People's Hospital, the cancerous intestinal mucosa tissues from six colon adenocarcinoma patients were obtained from Shenzhen People’s Hospital. We fully explained the purpose and experimental procedures to each patient. They all volunteered to participate in the program and signed the informed consent. All issues of CRC patients were collected with reference to a reported experimental method [14]. Tumor tissues were taken from the colon segment, and the normal adjacent colon mucosa was removed within 5 cm of the tumors. All tissues were collected within 0.5 h after surgery, frozen in liquid nitrogen immediately for more than three hours, and stored at − 80 °C.

Trichloroacetic acid (TCA) was slowly added in each sample, followed centrifugation at 4500 g at 4 °C for 5 min, then the supernatant was discarded. The remaining debris was washed with pre-cooled acetone once, dried in the air, and dissolved with 200 mM tetraethyl ammonium bromide (TEAB). Subsequently, the trypsin was added to the remaining debris and reacted with the sample overnight. After incubating with the dithiothreitol at 56 °C for 30 min, the remaining debris was added with iodoacetamide (IAA) and then incubated at room temperature for 15 min in the dark.

The LC–MS/MS analysis was performed following a standard protocol [15]. The peptides were dissolved with mobile phase A (0.1% formic acid, 2% acetonitrile), then the mixtures were separated using the Nano Elute ultra-high performance liquid system. The gradient of mobile phase B (containing 0.1% formic acid and 100% acetonitrile) was set as follows: 0–44 min, 6–22% mobile phase B; 44–56 min, 22–30% mobile phase B; 56–58 min, 30–80% mobile phase B; 58–60 min, 80% mobile phase B. The flow rate was set as 450.00 nl/min. The separated peptides were then injected into the capillary ion source for ionization and analyzed using the times-TOF Pro mass spectrometer. The ion source voltage was set into 2.0 kV. The scanning range of secondary MS was set into 100–1700 m/z. Parallel accumulation serial fragmentation (PASEF) mode was used for data acquisition. After the first mass spectrum, the secondary mass spectrum was collected ten times in the PASEF mode, and the dynamic exclusion time of tandem MS scanning was set into 30 s.

The secondary MS data were searched against the Homo sapiens 9606 (20,366 sequences) with the common contamination database using the Maxquant search engine (1.6.6.0). The reverse database was added to calculate the false discovery rate (FDR) caused by random matching. All parameters were set as follows: the FDR and peptide spectrum matches (PSM): 1%; Enzyme digestion: trypsin/P; Number of missed sites: 4; Minimum length: 7; Maximum modification number: 5; Mass error tolerance of primary parent ion and the main search: 20 ppm in the first search; Fragment ions: 0.02 Da; Fixed modification: cysteine alkylation carbamidomethyl; Variable modification: acetyl (protein N-term), oxidation (m); glygly (k).

Gene Ontology (GO) annotation was performed using the UniProt-GOA database (http://​www.​ebi.​ac.​uk/​GOA). Kyoto Encyclopedia of Genes and Genomes (KEGG) database (http://​www.​genome.​jp/​kegg/​) was utilized to annotate the differentially ubiquitinated proteins into pathways. The mRNA expression was analyzed using the datasets from Gene Expression Profiling Interactive Analysis (GEPIA) database (https://​www.​ncbi.​nlm.​nih.​gov/​geo/​). The protein expression was investigated using the datasets from The National Cancer Institute’s Clinical Proteomic Tumor Analysis Consortium (CPTAC) database (https://​portal.​gdc.​cancer.​gov/​). The immunohistochemical detection was obtained from The Human Protein Altas (HPA) database (www.​proteinatlas.​org).

The ratio of the relative quantitative value of each protein in the tissue was considered as a fold change (FC). The calculation formula was as follows: FC A/B, k = RAk / RBk (‘‘A’’/‘‘B’’ represented different tissues, ‘‘R’’ represented the relative quantitative value of the protein, and ‘‘k’’ represented the protein). Based on the above analysis of differences, the change of differential expression more than 1.5 was regarded as the threshold of significant up-regulation, and that less than 1/1.5 was regarded as the threshold of significant down-regulation. Fisher’s exact tests were used to evaluate the significance of enrichment results. A P value of less than 0.05 was considered being statistically significant.

To characterize the ubiquitinome of CRC patients, we collected the cancerous and para-cancerous tissues from six CRC patients and performed proteomic and ubiquitinomic analyses on them. The clinical information of the six CRC patients was summarized in Table 1, including age, type, gender, tumor/node/metastasis (TNM) stage, and chemotherapy. As shown in Additional file 1: Figure S1, most of the peptide lengths were distributed in 7–20 amino acids, which met the quality control requirements. For proteomic examination, we identified 86,259 matched spectrums, 43,073 peptides, 41,367 unique peptides, 6080 identified proteins, and 4712 quantifiable proteins. For ubiquitinomic examination, we identified 16,615 matched spectrums, 9295 peptides, 5537 identified sites, 5494 modified peptides, 2410 identified proteins, 1690 quantifiable sites, and 870 quantifiable proteins (Additional file 1: Figure S2, Additional file 2: Table S1,S2).

In eukaryotes, ubiquitination is essential for many biological processes, including cell survival and differentiation, as well as for innate and adaptive immunity. The analysis of the biological processes and pathways in which the highly-ubiquitinated proteins are involved may provide a preliminary exploration of the role of ubiquitination. Therefore, we enriched all the ubiquitinated proteins and found that 98.3% of total proteins (2409 proteins) harbored less than 10 ubiquitination modifications and 1.7% (42 proteins) harbored 10 or more ubiquitination modifications (Fig. 1A, Additional file 2: Table S3), such as PRKDC, PARP14, SLC12A2, SPG10, HKDC1, MYH9, HSPA8, AHNAK, HSP90AB1, POLR2B, and GCN1. Among the highly-ubiquitinated proteins (n ≥ 10), PRKDC was the most modified one among all proteins, which had 49 ubiquitination sites in total (Fig. 1B). To better understand the functions of high-ubiquitinated proteins, we performed GO and KEGG analyses of the high-ubiquitinated proteins and the low-ubiquitinated proteins, respectively. After removing the same modules shared with low-ubiquitinated proteins, we found that the high-ubiquitinated proteins were especially enriched in modules of G-protein-coupled receptors, ATPase activity, immunity, and cytoskeleton (Fig. 1C–F).

Subsequently, we searched for the motifs frequently modified by the ubiquitin in normal and cancerous colonic tissues using Motif-X (Fig. 1G). As a result, 20 motifs were screened out Additional file 2: Table S4) and the top five motifs were LKQ, KxL, KL, IxK, and KV (Fig. 1H). These five motifs were distributed in 56, 775, 585, 315, and 355 unique peptides, respectively (Fig. 1I).

To explore the functions of disturbed ubiquitination in CRC, we analyzed the differentially expressed ubiquitination in CRC cells versus normal cells. As a result, the ubiquitination contents of 1172 proteins were up-regulated, and the ubiquitination contents of 1700 proteins were down-regulated. Meanwhile, the ubiquitination contents of 2242 sites were up-regulated, and the ubiquitination contents of 1204 sites were down-regulated (fold change > 1.5, the ubiquitination solely detected in cancer samples or normal adjacent samples were also included) (Fig. 2A, Additional file 2: Table S5, S6). Subsequently, we performed GO and KEGG enrichment analyses of these different ubiquitinations, and found that the proteins with altered ubiquitination were mainly enriched in secretory vesicle, myelin sheath, and cytoplasmic vesicle part, etc. (Fig. 2B). This result suggested that secreted proteins were more likely to be modified by ubiquitin, and thus ubiquitination might play an important role in protein transporting and secretion. Meanwhile, the ubiquitination-altered proteins were also enriched in processes including cytoskeletal protein binding, small molecule binding, adenyl nucleotide binding, and coenzyme binding, etc. (Fig. 2C), and involved in the purine nucleotide metabolic process, nucleotide metabolic process, neutrophil mediated immunity, neutrophil activation, etc. (Fig. 2D). This result revealed that the differentially expressed ubiquitination in CRC cells versus normal cells might be associated with metabolic and immune regulation. Next, the KEGG enrichment analysis uncovered that the differentially expressed ubiquitination was enriched in pathways including carbon metabolism, pyruvate metabolism, leukocyte transendothelial migration, cell cycle, viral carcinogenesis, ribosomes, citrate cycle, etc. (Fig. 2E). As observed in Fig. 2B, the proteins harboring differential ubiquitination were mostly secretory proteins. To further explore the functions of these proteins, we performed functional analysis of these secretory proteins (Additional file 2: Table S7) using the Metascape database, and the results showed that the functions of secretory proteins were closely related to neutrophil degranulation and JAK-STAT pathways. The JAK-STAT pathway is closely related to inflammation [16], so this result further confirmed a probably new function of ubiquitination modification in the immune responses (Fig. 2F). At the same time, we also analyzed the cellular sources of secretory proteins through PaGenBase database, and found that these proteins were more located in blood cells (Fig. 2G), which further revealed that ubiquitination modification might play a role in the immune responses.

To further analyze the functional network of differentially-ubiquitinated proteins, we annotated the genes using the Metascape database (Fig. 3A). Consistently, we enriched the pathways including myeloid cell activation involved in immune response and cytokine signaling in the immune system, highlighting the role of ubiquitination in regulating the immune microenvironment of CRC patients (Fig. 3A). Subsequently, we searched for the functional modules within the network using MCODE. As a result, modules named Ribosome, Endocytosis, Metabolism 1, and Metabolism 2 were discovered (Fig. 3B–E).

Numerous previous studies have revealed that ubiquitination has a variety of functions, such as protein degradation, receptor internalization [17], protein assembly [18], transporting [19], signal transduction [20], and enzymatic activity modulation [21]. In the study, we focused on the function of protein degradation, and thus we searched for the differentially expressed proteins simultaneously with the opposite changes of ubiquitination. As a result, 53 proteins had up-regulated expression simultaneously with down-regulated ubiquitination, and 116 proteins had down-regulated expression simultaneously with up-regulated ubiquitination (169 in total) (Additional file 1: Figure S3, Additional file 2: Table S8). Then, the expression of the above 169 proteins were verified using the data set from CPTAC, and 59 proteins were confirmed (7 proteins were up-regulated and 52 proteins were down-regulated) (Additional file 1: Figure S3, Additional file 2: Table S9, S10). Subsequently, to screen out the functional ubiquitination potentially associated with CRC progression, we performed survival analyses using the proteomic datasets and the clinical information from the CPTAC database (n = 98). We found that 101 proteins were probably associated with the OS of CRC patients (Additional file 2: Table S11). We overlapped the 59 proteins selected in Additional file 1: Figure S3, and the 101 OS-relevant proteins and obtained two proteins ultimately, including FOCAD and DOCK2.

As shown in Fig. 4A–B, the protein expression of FOCAD and DOCK2 were both down-regulated in CRC tissues versus normal tissues (n = 100). Besides, the lower protein expression of FOCAD and DOCK2 were correlated with poor OS of CRC patients (Fig. 4C–D). Meanwhile, the ubiquitination of FOCAD at Lys583, Lys587, and DOCK2 at Lys151, Lys1006, and Lys1359 were all up-regulated in CRC cells versus normal cells (Fig. 4E–F). We considered that the decreased protein expression might be caused by the decreased mRNA content or increased ubiquitination. To clarify the protein down-regulation of FOCAD and DOCK were induced by up-regulated ubiquitination, rather than mRNA down-regulation, we analyzed the mRNA expression of FOCAD and DOCK in CRC cells versus normal cells using the RNA-Seq datasets from The Cancer Genome Atlas (TCGA). As a result, the mRNA expression of FOCAD was significantly up-regulated in CRC cells versus normal cells, while the mRNA expression of DOCK2 was significantly down-regulated in CRC cells versus normal cells (Fig. 4G–H). This result indicated that the down-regulation of the FOCAD protein was more likely to be caused by an increase of ubiquitination than DOCK2. Moreover, we analyzed the correlation between the protein level and the mRNA level of these two genes in a large cohort using data sets from the cbioportal database, and found that the DOCK2 mRNA had a higher correlation with its protein level (correlation coefficient = 0.49), while FOCAD mRNA had a lower correlation with its protein level (correlation coefficient = 0.15) (Fig. 4I–J). This further proved that FOCAD was more likely to be modified by ubiquitination to affect its protein level. Ultimately, we verified the protein expression of FOCAD using the immunohistochemical results from the HPA database and confirmed a reduced expression of FOCAD in CRC patients (Fig. 4 K–L).

Patients with germline deletions of FOCAD are more likely to develop intestinal polyps and tumors [12]. To further-in-depth reveal the role of FOCAD in CRC, we analyzed the mutation frequency and loci of FOCAD using the datasets from cbioportal database. The results showed that FOCAD was a frequently mutant gene (6%) in colon adenocarcinoma (Fig. 5A). Moreover, the loci of A381V, Q1179K, C66R, A1037V, R1546L, H1302P, G632C, R556Q, and N686S of FOCAD were more prone to be mutated (Fig. 5B).

Subsequently, to further explore the functions of FOCAD in CRC, we investigated the co-expressed proteins of FOCAD in CRC patients (correlation coefficient > 0.3 or < -0.3) using the datasets from the Linkedomics database. Figure 6A, B showed the top 50 positively and negatively correlated proteins of FOCAD. Next, the Gene Set Enrichment Analysis (GSEA) disclosed that FOCAD might be associated with gene transcription and translation, PI3K-AKT pathway, immune pathways, and metabolism pathways (Fig. 6C–F)