Identification of differentially expressed proteins in colorectal cancer

Abstract

Colorectal cancer (CRC) is the third most common cancer and a major public health issue worldwide. Survival rate of patients with CRC depends upon early diagnosis of the disease. In that CRC is mostly asymptomatic at an early stage and typically diagnosed only at advanced stages, implementation of screening strategies for early detection is highly desirable. Notwithstanding the fact that continuous development in screening protocols and available treatment options has resulted in a considerable decline in the CRC mortality rate in the developed nations, the scope of benefits so far has not extended to the developing nations like Pakistan where most of the patients present at the advanced stage with resultant poor survival rates. Currently available invasive, semi-invasive and non-invasive diagnostic/screening methods [e.g., carcinoembryonic antigen (CEA), faecal occult blood test (FOBT), colonoscopy, sigmoidoscopy, computed tomography (CT) scan or barium enema, etc.] have significantly improved the patient survival rates for CRC. However, some of the blood- and stool-based clinical tests, currently in practice, lack sufficient level of sensitivity and/or specificity. Hence, there remains a compelling need for developing more reliable, sensitive and specific non-invasive methods and the molecular markers for earlystage diagnosis CRC, to ensure increased survival rate of the patients and better disease management. Proteomic strategies to identify markers for the diagnosis of cancers (such as lung, liver, pancreas breast and ovarian cancers) at an early stage have been employed with noteworthy results. To extend these studies, we have utilized two dimensional gel electrophoresis (2D-PAGE) and mass spectrometry (MS) for expression profiling of proteins extracted from the freshly frozen human colorectal cancer tissue specimens and the comparable regions of adjacent normal mucosa (serving as controls) with an aim of identifying novel CRC associated proteins. Based on the anatomic/Duke’s staging, the vi collected CRC tissue samples (n=12) were grouped as 1) Stage B, 2) Stage C, and 3) Stage D samples. Equal amounts of the protein from tissue lysates of tumor and adjacent normal mucosa were resolved by 2D-PAGE followed by colloidal coomassie blue staining. On average, 505±216 spots in normal and 497±221 in tumor tissue lysate appeared when the colloidal-coomassie stained gels of the three CRC patient groups were compared. To select statistically significant, differentially stained gel spots, the quantification of individual spots was performed using ImageMaster 2D platinum (v. 7.0) software program. During the initial screening, 49 gel spots were found to have at-least one fold change in staining with p-value ≤0.05. False discovery rate (FDR) estimation of the data was performed with value set as ≤0.1 reflecting that more than 90% of the findings are accurate. The selected spots were subjected to identification by MALDI-TOF and/or liquid chromatography-based tandem mass spectrometry (LC-MS/MS). In the case of MALDITOF- MS data analysis, the threshold value for positive hits of peptide mass finger printing (PMF) score was set as ≥79 whereas in the case of LC-MS/MS, only the proteins identified with >95% probability were considered for further analysis. These analyses led to the identification of 21 distinct proteins. In silico characterization of the proteins identified by MS analysis (MALDI-TOFMS and LC-MS/MS) analysis was also performed. More specifically, the proteins of data sets were uploaded in Protein ANalysis THrough Evolutionary Relationships (PANTHER) database that categorized all proteins into molecular functions, biological processes and cellular components. Gene ontology (GO) data search for molecular functions classified these proteins based on of their binding, receptor activity, structural molecule activity, catalytic activity, antioxidant activity and transporter activity. In the context of biological processes, these proteins were found to be involved in ten major categories of biological processes viz., cellular component organization or biogenesis, cellular processes, vii localization, biological regulation, response to stimulus, developmental process, multicellular organismal process, biological adhesion, metabolic process and immune system process. Data analysis using Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) database was also performed for an overview of the functionally connected proteins. It was found that two proteins PSME1 (proteasome activator subunit 1) and FABP5 functionally separated proteins than rest of proteins in our data set. So we performed individual network analysis of these two proteins and our protein of interest, ACTBL2. Collectively ACTBL2 appears to interact with proteins in cells that have potential to arrest cell growth, elicit tumorigenesis and promote cell migration. PSME1 directly involved in regulation of proteasome activity and FABP5 is important in cellular lipid catabolic process. Further for differentially expressed proteins, disease assessment was performed by Ingenuity Pathways Analysis (IPA). Most proteins showed up to be important in cancer. These prediction analyses by different tools all support together that experimentally identified proteins are vital in health and disease and could be beneficial if carefully studied in disease assessment. In the present study about 50% of the identified proteins were represented by more than one spots, generally 2-5 which could be due to difference in PTMs or alternative splicing. Therefore, the intensities of multiple gel spots that corresponded to the same protein were pooled/summed up and their statistical significance was re-calculated. Four gel spots that qualified four tier criteria (fold change >1.5, p-value ≤ 0.05, PMF score ≥79 and FDR analysis ≤0.05), were selected for further validation analysis. Two gel spots showed greater staining in the CRC samples compared to the matched samples from the normal tissue region while two showed decreased staining for the same comparison. These four spots include six proteins; beta-tropomyosin (TPM2), ACTBL2, PSME1, prohibitin viii (PHB), annexin A2 (ANXA2) and myosin light chain 9 (MYL9). Out of these six proteins five had previously been reported to be associated with colorectal cancer. In that ACTBL2 was identified in CRC at greater abundance than in normal tissue and had not previously been associated with CRC, we verified the mass spectrometry data by performing immunohistochemistry (IHC) for ACTBL2 in six paraffin embedded formalin fixed (FFPE) CRC patients samples and respective controls which were obtained from the Biorepository and Tissue Research Facility, University of Virginia, USA for validation of selected proteins.. From the mass spectrometry experiments above, PSME1 also appeared to be in high abundance in CRC samples compared to normal tissue. Given that PSME1 has previously been associated with CRC in some patients, we decided to perform IHC on another panel of CRC patient samples as well. For ACTBL2, significantly higher staining was observed in all six patient samples with p-value ≤0.005 and fold change +3.5. These results are similar to those obtained from the 2DE/MS data. However, in the case of PSME1, the IHC results showed marginally higher staining only in three out of six CRC patient samples when compared with the normal epithelium. This suggested that PSME1 is not upregulated in CRC patients. The higher abundance observed in 2DE/MS data for PSME1 could be due to the greater staining of spot CRC-05 containing two other upregulated proteins i.e., ANXA2 and PHB. Thus, ACTBL2 association and differential upregulation in colorectal cancer is novel, and as such may contribute to our understanding of the colorectal carcinogenesis and potentially serve a function in developing markers for colorectal cancer. Uncovering the differentially expressed proteins in colorectal cancer observed in this study will be important to understand the CRC carcinogenesis. However for more precise statistically significant proteins, much larger number of samples are needed along with testing the samples in serum/plasma samples. Respective studies have been started in our laboratory.