Abstract Background Recently, the distinction between left- and right-sided colon cancer (LCC and RCC) has been brought into focus. RCC is associated with an inferior overall survival and progression-free survival. We aimed to perform a detailed analysis of the diversity of extracellular vesicles (EV) between LCC and RCC using quantitative proteomics and to identify for new diagnostic and prognostic biomarkers. Methods We isolated EVs from patients with LCC, RCC and healthy volunteers, and treated colorectal cancer cell line with serum-derived EVs. We then performed a quantitative proteomics analysis of the serum-derived EVs and cell line treated with EVs. Proteomic data are available via ProteomeXchange with the identifiers PXD012283 and PXD012304. In addition, we assessed the performance of EV SPARC and LRG1 as diagnosis and prognosis biomarkers in colon cancer. Findings The expression profile of the serum EV proteome in patients with RCC was different from that of patients with LCC. Serum-derived EVs in RCC promoted cellular mobility more significantly than EVs derived from LCC. EV SPARC and LRG1 expression levels demonstrated area under the receiver-operating characteristic curve values of 0.95 and 0.93 for discriminating patients with colon cancer from healthy controls. Moreover, the expression levels of SPARC and LRG1 correlated with tumour sidedness and were predictive of tumour recurrence. Interpretation We identified differences in EV protein profiles between LCC and RCC. Serum-derived EVs of RCC may promote metastasis via upregulation of extracellular matrix (ECM)-related proteins, especially SPARC and LRG1, which may serve as diagnosis and prognosis biomarkers in colon cancer. Keywords: Colon cancer, Extracellular vesicles, Proteomics, Metastasis, Tumour location Abbreviations: CRC, colorectal cancer; NCCN, National Comprehensive Cancer Network; LCC, left-sided colon cancer; RCC, right-sided colon cancer; mCRC, metastatic CRC; PFS, progression-free survival; EGFR, epidermal growth factor receptor; OS, overall survival; EV, extracellular vesicles; EMT, epithelial–mesenchymal transition; TMT, Tandem Mass Tagging; LC, liquid chromatography; MS, mass spectrometry; HPLC, High performance liquid chromatography; TOF, Time of Flight; GO, Gene Ontology; FDR, false discovery rate; PCA, Principal component analysis; NTA, nanoparticle tracking analysis; FBS, fetal bovine serum; ROC, receiver operating characteristic curve; AUC, area under the receiver-operating characteristic curve; DEPs, differentially expressed proteins; ECM, extracellular matrix; SPARC, secreted protein acidic and cysteine rich; LRG1, leucine rich alpha-2-glycoprotein 1; VCAM1, vascular cell adhesion molecule 1; THBS1, thrombospondin 1; FN1, fibronectin 1; VTN, vitronectin __________________________________________________________________ Research in context. Evidence before this study Primary tumour sidedness has been found to be prognostic in colorectal cancer, with right-sided tumours having a worse prognosis than left-sided tumours, even after controlling for known negative prognostic factors. In addition, recent analysis suggests that sidedness may also be a predictive marker of the response to epidermal growth factor receptor (EGFR) inhibitor, with right-sided tumours having a poor response. Added value of this study The diversity between left- and right-sided colon cancer is still controversial. And molecular underpinnings of this difference remain unclear. Here, we present a detailed analysis of the diversity in extracellular vesicles between left- and right-sided colon cancer using quantitative proteomics. Our study confirms the difference between left- and right-sided colon cancer at the serum extracellular vesicles level. Implications of all the available evidence The present study identify difference between left- and right-sided colon cancer at the serum extracellular vesicles level. And we found that extracellular vesicles derived from patients with right-sided tumours promote metastasis more significantly than those derived from patients with left-sided colon cancer. We hypothesize that serum-derived extracellular vesicles from right-sided colon cancer promote metastasis by upregulation of extracellular matrix-related proteins, especially SPARC and LRG1, which may serve as potential diagnostic and prognostic biomarkers in colon cancer. Alt-text: Unlabelled box 1. Introduction It has long been recognized that colorectal cancer (CRC) is molecularly heterogeneous, and its clinical behaviour differs if the primary tumour is located in the right or left side of the colon [[39]1,[40]2]. According to the National Comprehensive Cancer Network (NCCN) guidelines, tumours located in the splenic flexure, descending colon, sigmoid colon, and rectum are defined as left-sided colon cancer (LCC). In contrast, tumours located in the region from the hepatic flexure to the cecum are defined as right-sided colon cancer (RCC) [41][3]. Over the past few years, the distinction between LCC and RCC has been brought into focus due to their diversities in biology, clinical characteristics, prognosis and treatment response [[42]2,[43]4,[44]5]. Multiple retrospective analyses of randomized controlled trials [45][6], [46][7], [47][8], [48][9] have revealed that RCC leads to an inferior prognosis. A retrospective analysis of the NCIC CO.17 trial [49][6] indicated that tumour location could be used to predict treatment effectiveness. In accordance with this conclusion, a retrospective analysis of CRYSTAL and FIRE-3 studies [50][7] indicated the value of primary tumour location in predicting metastatic CRC (mCRC), with right-sided tumours associated with a worse prognosis than left-sided tumours regardless of the first-line treatment regimen. However, with respect to progression-free survival (PFS), a significant interaction between primary tumour location and treatment was observed. Analysis of tumour location subgroup data from the Phase Ⅲ CALGB/SWOG 80405 trial [51][8] showed similar prognostic and predictive impacts of tumour location to those of a retrospective analysis of CRYSTAL and FIRE-3 studies. Consequently, the NCCN guidelines now recommend against using first-line epidermal growth factor receptor (EGFR) inhibitors in patients with RCC, regardless of KRAS status. To investigate the prognosis between left- and right-sided non-metastatic colon cancer, we performed a retrospective study of 175 patients with histologically proven stage Ⅲ colon cancer undergoing radical resection at our institute from 2005 to 2012. Our study [52][10] demonstrated that patients with a right‐sided tumour carried a greater number of negative prognostic factors such as mucinous adenocarcinoma, and had inferior overall survival (OS) and PFS compared with those with tumours originating on the left side. In another retrospective study of 1,869 patients with stage Ⅲ colon cancer [53][11], patients with right-sided tumours had shorter survival after relapse and shorter OS compared with the patients with left-sided tumours. A more comprehensive understanding of the biological differences between tumours in different locations may help to develop more efficacious therapies. Further research is now required to clarify the reasons for these differences and to identify better treatments tailored to the patient. We believe that there are molecular characteristics that can be used to track right versus left disease, although they are not yet well described. Recently, the roles of extracellular vesicles (EV) and their contents as potential contributors to oncogenesis, metastatic disease, and resistance to chemotherapy is a rapidly expanding area of research in cancer biology [[54]12,[55]13]. EVs are extracellular vesicles released from the cell membrane that play a critical role in cell-cell communication [56][14] through the transmission of molecular messengers that alter the phenotype of recipient cells [[57]15,[58]16]. Previous studies have indicated that EVs are involved in metastasis by harbouring molecules that are involved in the epithelial-mesenchymal transition (EMT) or preparing target tissues for metastasis [[59]17,[60]18]. In general, previous studies of EVs at the levels of basic and clinical research are important in elucidating their role in cancer. To date, most studies of vesicle-mediated carcinogenesis were performed with EVs isolated from supernatants of tumour cell lines and, less frequently, from serum of patients with cancer. Here, we present the first detailed analysis of the diversity in EVs between LCC and RCC using quantitative proteomics. The procedure is schematically outlined in [61]Fig. 1. Fig. 1. [62]Fig 1 [63]Open in a new tab Schematic workflow. The workflow shown the TMT-based quantitative proteomic analysis of extracellular vesicles (EV) isolated from the pooled serum of patients with colon cancer and healthy volunteers. “RCC” refers to right-sided colon cancer; “LCC” refers to left-sided colon cancer. Original elements used in this diagram are from Servier Medical Art ([64]http://smart.servier.com/). 2. Materials and methods 2.1. Reagents Details of reagents are given in Supplementary Materials and Methods. 2.2. Patients and serum samples A total of 78 patients diagnosed with stage Ⅲ moderately differentiated colon cancer and 40 healthy volunteers were recruited in Peking Union Medical College Hospital (PUMCH, Beijing, China) from 2015 to 2017. The following exclusion criteria were applied: (a) with diabetes, autoimmune diseases, or blood diseases; (b) with a history of chemo- or radiotherapy, or a previous history of malignancy; (c) transverse colon cancer; (d) rectal cancer; (e) mucinous adenocarcinoma. The definition of LCC and RCC is consistent with that described in the Introduction. Rectal cancer was not included in the present study because they are treated differently to colon cancer. There was no statistical difference between these two patient groups. Details are shown in Table S1. All patients with stage Ⅲ colon cancer were treated with an oxaliplatin-based chemotherapy regimen followed by curative resection. Long-term follow-up was performed according to the NCCN guidelines [65][19]. The patients’ follow-up cut-off was January 2019. The healthy volunteers were recruited from people that had their medical check-up in our centre. Most of them underwent a thorough physical exam, electrocardiography, some laboratory tests, such as complete blood count and serum tumour marker tests, as well as medical imaging such as chest X-ray and abdominal ultrasound. They were defined as healthy individuals according to their check-up results. An additional 75 patients with other malignancies including thyroid cancer (n = 25), cervical cancer (n = 25) and gastric cancer (n = 25) enrolled for validation in March 2019. Clinical information about patients and volunteers are shown in Table S1-4. This study was approved by the Ethics Committee of PUMCH (No. S-k655) and was conducted in accordance with the most recent version of the Declaration of Helsinki. All patients and healthy volunteers provided written informed consent to participation in this study. Venous blood samples were obtained (using a 21 G gauge needle; the first 1 ml was discarded) from both patients and healthy volunteers in the fasting state. Serum samples were collected from patients before they underwent radical surgery. For quantitative proteomics analysis, serum samples from 14 patients with LCC and 14 patients with RCC were pooled, respectively. Serum samples obtained from 15 healthy volunteers were pooled and allocated to the normal control group. These three groups of pooled serum were applied to EV isolation by ultracentrifugation. Serum samples from the rest patients and volunteers were used for validation. 2.3. EV isolation from human serum EV were isolated from serum samples by ultracentrifugation or Total Exosome Isolation Reagent. For proteomics analysis, serum samples were pooled respectively and then applied to ultracentrifugation as previously described [66][20]. A detailed description of ultracentrifugation experiments is given in Supplementary Materials and Methods. Due to the low volumes of sera available from patients, we isolated EV using Total Exosome Isolation Reagent for other assays according to the manufacturer's instructions. In brief, pooled or individual serum samples were first diluted with an equal volume of PBS to decrease viscosity, followed by the addition of 0.2 vol of the Total Exosome Isolation Reagent. Mixtures of serum and reagent were vortexed and incubated at 4 °C for 30 min and then centrifuged at room temperature to isolate EV pellets. Samples were centrifuged at 10,000 × g for 30 min and the pellet was then resuspended in PBS containing 1% penicillin/streptomycin. EVs isolated from 100 μL serum were resuspended in 20 μL PBS. The protein content of the isolated EV was measured using the BCA assay after lysis with RIPA. 2.4. Tandem mass tagging (TMT) labelling For TMT labelling, the lysates of EVs from the three sample groups (Normal, LCC and RCC) were diluted to 1 mg/mL with 8 M urea. Labelling was performed using the TMT kit according to the manufacturer's protocol with slight modifications. Details are described in Supplementary Materials and Methods. 2.5. Liquid chromatography (LC)-mass spectrometry/mass spectrometry (MS) analysis The TMT-labelled peptides were fractionated by High performance liquid chromatography (HPLC). For LC-MS/MS analysis, peptides were separated using a 135-min gradient elution at a flow rate 0.3 μL/min with the Ultimate U3000 system, which was directly interfaced with the Thermo Orbitrap Fusion Lumos mass spectrometer. A detailed description of HPLC and LC-MS/MS experiments is given in Supplementary Materials and Methods. 2.6. Data processing Proteins were identified using Proteome Discoverer 2.2 software (Thermo Scientific) with the SEQUEST search engine. The raw MS data files were searched against the UniProt/SwissProt human proteome database (released on February 5, 2018). The search criteria and details are described in Supplementary Materials and Methods. In the current study, identified proteins were defined as proteins with at least two unique peptides. 2.7. Protein identification using MS/MS data Representative MS/MS spectral identification was performed as previously described [67][21]. Briefly, MS/MS spectral data of identified peptides and the intensity of TMT precursor ions were used for protein quantification. The masses of the resulting peptides were measured to obtain a Time of Flight (TOF) spectrum. Peaks from the TOF spectrum were selected for sequencing by fragmentation (MS/MS). 2.8. Bioinformatics analysis For proteomic analysis of human serum-derived EVs, relative protein abundances were presented as the ratios to TMT-129/131 (LCC/normal group), 126/131 (RCC/normal group), and 126/129 (RCC/LCC). The differential expression threshold was set as a 1.5-fold change. Details of the MS proteomics data are available from the ProteomeXchange Consortium [68][22] via the PRIDE partner repository (dataset identifier PXD012283). For proteomic analysis of CRC cell line SW480 treated with serum-derived EVs, relative protein abundances were presented as the ratios to TMT-127/126 (normal/PBS group), 129/126 (LCC/PBS group), 131/126 (RCC/PBS group), 129/127 (LCC/normal group), 131/127 (RCC/normal group), and 131/129 (RCC/LCC). Proteins were considered differentially expressed when fold change >1.2. The MS proteomics dataset was submitted to the ProteomeXchange Consortium with the identifier PXD012304. To stratify the proteome, a list of cancer-related proteins was downloaded from The Human Protein Atlas database ([69]https://www.proteinatlas.org/) [70][23]. Gene Ontology (GO) functional enrichment analysis was conducted using the clusterProfiler package [71][24] in R program (R Foundation for Statistical Computing, Vienna, Austria. [72]http://www.R-project.org/). False discovery rate (FDR) <0.05 was set as the threshold for statistical significance for GO enrichment analysis. Pathway enrichment analysis was performed using the Cytoscape plug-in ClueGO [73][25] based on WikiPathway database (released: November 05, 2018). Statistical significance of pathways was based on adjusted P-values of <0.05 and the presence of at least five target genes. Hierarchical Ward-linkage clustering was performed based on Spearman correlation coefficients using JMP Pro (version 13.0, SAS Institute, Cary, NC, USA). Principal component analysis (PCA) was performed using JMP Pro. The STRING database ([74]http://string-db.org) [75][26] was introduced for protein-protein interaction network analysis. And the results were graphically represented with Cytoscape (version 3.2.1, Cytoscape Consortium, USA, [76]https://cytoscape.org/) [77][27]. 2.9. Nanoparticle tracking analysis (NTA) The size distribution and concentration of EV were calculated by NTA using a Nanosight LM10 (Nanosight, Amesbury, UK) equipped with a fast video capture and particle tracking software. Nanoparticles were illuminated by a 635 nm laser and their movement under Brownian motion was record for 60 s. Then videos were collected and analysed with NTA 3.2 software (Nanosight, Amesbury, Wiltshire, UK). 2.10. Western blotting analysis Serum EV and primary tumour tissue were applied for immunoblotting. Western blot analyses were performed with the following primary antibodies: ALIX, CD63, TSG101, PSMA5, HMGB1, SPARC and LRG1. Details are described in Supplementary Materials and Methods. Coomassie-stained SDS-PAGE was used as loading control for EV. Human β-actin was used as an internal reference for tissue proteins. Densitometry analysis was performed using the ImageJ software (National Institutes of Health, USA). 2.11. ELISA For each target protein detection, 300 μL serum-derived EVs were resuspended in 70 μL RIPA and diluted to 100 μL with sample diluent provided in the ELISA kit. The procedure was performed following the manufacturer's instruction with no modifications. 2.12. Cell culture SW480 and HCT116 cells were obtained from China Infrastructure of Cell Line Resources (Beijing, China). Cell line authentication and mycoplasma testing were not performed. SW480 was cultured in Dulbecco's Modified Eagle Medium (DMEM) containing 10% fetal bovine serum (FBS). HCT116 was cultured in Iscove's modified Dulbecco's medium (IMDM) containing 10% FBS. All cell cultures were maintained at 37 °C under at 5% CO[2] in a humidified atmosphere. Cells were passaged approximately every 2–3 days. 2.13. Cell proliferation assay Cells (2 × 10^3) were seeded into 96-well plates with medium contained 40 μg EVs. Cell growth was determined every 24 h by using the Cell Counting Kit-8 (CCK-8) assay. Three replicates per condition were assayed. 2.14. Cell invasion and migration assay Transwell assays were used for studying the motility of cells treated with serum EVs as previously described [78][20]. Briefly, 175 μg EVs isolated from serum of colon cancer patients or healthy volunteers were added to the upper chamber transwell insert; an equal volume of EV-free PBS was added as the blank control. More details are described in Supplementary Materials and Methods. At first, EVs isolated from pooled samples were applied for transwell assays. Then, we performed a further validation using another group of individual samples. We randomly selected four healthy volunteers, four patients with LCC, and four patients with RCC from the validation cohort. None of these 12 individuals were included in the previous analysis of pooled samples. These 12 individuals were allocated to four groups. Each group included one healthy control, one LCC sample, one RCC sample and one PBS as a blank control. In other words, these 12 individuals were regarded as four groups of biological replicates. 2.15. Statistical analysis All statistical analyses were performed with SPSS statistics (version 23.0, International Business Machines Corp., Armonk, New York, USA), JMP Pro, and GraphPad Prism (version 7.04; Nashville, TN, USA). Continuous variables are expressed as the mean ± SD. Differences between groups were compared using Student's t-test. Receiver-operating characteristic (ROC) analysis was used to assess the specificity and sensitivity of the biomarkers and the area under the ROC curve (AUC) was estimated for each individual protein. The Kaplan–Meier curves were generated to analyse the cumulative probability of PFS and statistical significance was evaluated using log-rank tests. The Cox proportional hazards regression was carried out to identify proteins with expression correlating with PFS. The Benjamini–Hochberg procedure was used to control the FDR. P < 0.05 (two-sided) was considered to indicate statistical significance. 3. Results 3.1. Verification of serum EV isolation Under transmission electron microscopy, serum-derived EVs showed typical cup-shaped round morphology (Fig. S1a). The commonly reported EV-enriched surface markers ALIX, CD63 and TSG101 were detected by Western blot analysis in biological replicates for both groups (LCC, RCC and healthy controls; [79]Fig. 2a). These vesicles showed high concentrations of ALIX, CD63 and TSG101 compared to the whole serum and supernatants of pooled serum samples after ultracentrifugation. Further verification of successful EV isolation was performed by proteomic detection of EV-enriched markers including CD9, CD63, CD81 and ALIX. Among the identified EV proteins, there was an overlap with 843 proteins (83.7%) in the ExoCarta protein list ([80]http://www.exocarta.org, release date: July 29, 2015) ([81]Fig. 2c). Of the top 100 exosome-associated proteins from Exocarta, 92% were identified in the present study (Fig. S1b). Analysis of the serum EV size by NTA revealed slightly lower EV size in the RCC group versus the LCC group (Student's t-test, P < 0.05; [82]Fig. 2d-e). There were no differences in serum EV concentrations between LCC patients and healthy controls; however, the concentrations were slightly higher in LCC patients than in RCC patients (Student's t-test, P < 0.05; [83]Fig. 2d-e). According to GO classification, most of the EV proteins identified were derived from the cytoplasmic vesicle lumen, with important molecular functions in cell adhesion molecule binding, cadherin binding, and glycosaminoglycan binding as well as the biological processes of neutrophil activation and extracellular structure organization; these categories are consistent with the reported functions of EVs (Fig. S1c-e; a complete list of all GO terms is shown in Table S5). Fig. 2. [84]Fig 2 [85]Open in a new tab Verification of extracellular vesicles (EV) isolated from pooled serum. (a) Immunoblot analyses confirmed the presence of EV markers ALIX, CD63 and TSG101 among the harvested proteins. W: whole serum. S: supernatants after ultracentrifugation. E: extracellular vesicles. (b) The Coomassie-stained SDS-PAGE of EV lysates as a loading control for (a). (c) Venn diagram showing the overlap of identified proteins with ExoCarta proteins. (d-e) Nanoparticle tracking analyses of serum EVs. “Normal” refers to healthy volunteers, “LCC” refers to left-sided colon cancer, “RCC” refers to right-sided colon cancer. *** P < 0.001. * P < 0.05 (Student's t-test). 3.2. LCC- and RCC-derived serum EVs present distinct proteomic profiles The protein composition of EVs isolated from serum samples of RCC and LCC patients and healthy individuals was determined using TMT-based quantitative MS technology. The MS raw data was processed by Proteome Discoverer (version 2.2). In total, 1,007 proteins were identified (Table S6). PCA revealed differences in the proteome profiles between RCC, LCC and normal controls ([86]Fig. 3a). With a 1.5-fold change cut-off were regarded as differentially abundant in EVs, a total of 930 proteins were found to be differentially expressed in colon cancer versus normal controls, and 495 proteins were identified as differentially expressed proteins (DEPs) in RCC versus LCC. Among the 495 DEPs identified in serum-derived EVs isolated from RCC patients relative to those from the LCC patients, 57 were upregulated and 438 were downregulated ([87]Fig. 3b). Nineteen percent of these DEPs were cancer-related proteins ([88]Fig. 3c). Fig. 3. [89]Fig 3 [90]Open in a new tab Comparative analysis of proteome expression profiles of serum EV from LCC, RCC and healthy individuals. (a) Principal component analysis revealed differences in the proteome profiles between RCC, LCC and normal controls. “LCC” refers to left-sided colon cancer. “RCC” refers to right-sided colon cancer. “Normal” refers to healthy volunteers. (b) Scatter plot showing the distribution of up-regulated (red dots) and down-regulated (green dots) differentially expressed proteins (DEPs). (c) Venn diagram showing the overlap of the group of DEPs between the LCC and RCC groups (TMT-126/129) with the cancer-related proteins. (d) Hierarchical clustering analysis and heatmap of DEPs. The heatmap was constructed based on a log2 transformation of relative abundance ratios. (e) GO analysis of the upregulated proteins enriched in Cluster 1. “C1”- “C8” refers to Cluster 1 - Cluster 8. (f) Pathway analysis of the sidedness-related DEPs (TMT-126/129) of serum EVs. (g) The cancer-related DEPs in the protein-protein interaction networks are shown as nodes (MS data presented as the ratios to 126/129 were matched to STRING networks). Up- or downregulation of identified proteins is indicated by colours in the networks (upregulated in red, downregulated in green). (For interpretation of the references to colour in this