Dynamic Proteins Expression Patterns Reveal Prognostic Molecular Classification In Different Stages Of Human Colorectal Cancer

1 Background and purposeThe initiation, promotion and progression of human cancer are considered to be a complex, polygenic, multi-factors process. CRC is one of the common malignant tumors in the world, and the incidence rates increase every year. The recent "omics" revolution with high-throughput technologies has unveiled the law underlying the fact of cancer formation and progression through different levels of molecular profile. The application of molecular classifications, which is based on all kinds of molecular profile, will provide prerequisites for individualized treatment of colorectal cancer. In order to explore the tumor pathogenesis, illustrate the dynamical expression pattern of protein in CRC and establish molecular classifications, we are trying to use "Omics docking strategies" to establish molecular profile and reveal protein dynamic expression patterns in different stages of CRC. And then the CRC molecular staging classifier for CRC patients is obtained by Kaplan-Meier survival analysis. By using the molecular classifications, the diagnoses of CRC will regulate into the molecular level from the routine clinical diagnosis, which will provide prerequisites for individualized treatment of colorectal cancer.2 Materials and methods1) The specimen storeroom was built by collecting the specimen and the clinical pathological data. Twenty tissues were separated into 5 experiment groups including TNM I～IVclinical stages of CRC (each 5 samples) and 5 normal control groups before two-dimensional differential in gel electrophoresis (2D-DIGE). The molecular profiles of different stages of CRC were established by using DeCyder automatic analysis and statistical difference analysis system.2) The regularity of protein variation was discovered by Pattern Analysis and Principal Component Analysis with DeCyder EDA software. Meanwhile, Hierarchical clustering was performed by Gene Cluster 3.0 software. "Omics docking strategies" were employed to integrate the information of gemoe, transcriptome, proteome by using the same CRC samples.3) Proteins were identified by MALDI-TOF-MS/MS; classified by biological information analysis about its function, cellular localization and networks interaction; finally validated by Western blots and immunohistochemistry experiments.4) Colorectal cancer tissue microarray of 103 samples were followed up to obtain their survival information. The Kaplan-Meier survival analysis was used for the survival analysis of clinical pathological data. Discriminant analysis and survival analysis were combined to achieve the discriminant-function, which established the molecular classifications for prognosis at the different clinical stages of CRC. The accuracy, sensitivity and specificity were also assessed.3 Results1) After 2D-DIGE, we scanned qualified fluorescent gel maps.4863 matched protein spots were found by DeCyder analysis system, and paired student's t test was used to find the differentially expressed proteins between any two groups (p≤0.05; Av. Ratio≥2). The numbers of differentially expressed proteins in any two groups were exhibited. The differences between stage I and normal control were significant, while there were only small differences between the stage II and normal control. One-way ANOVA was applied to data analysis in the 5 experiment groups. And 199 differentially expressed proteins were obtained (One-way ANOVA, p< 0.05), in which 67 spots were significantly up-regulated in the tumor groups,114 were down-regulated, and 18 were up and down-regulated. Only spots that altered consistently in at least 3 out of the 5 patients were selected.2) Four Kmean's and eight Som expression patterns were obtained by patterns analysis among the 199 spots with DeCyder EDA module. Pattern clustering results show the dynamic changes of the protein in different stages of CRC and reveal the regularity of protein variation. The molecule in the advanced groups (stage III and stageIV) didn't form stage-specific clusters by PCA and Hierarchical clustering. CA1,CA2, ANXA4 perform the same expression pattern in mRNA, miRNA, protein in different stages of CRC by "Omics docking strategies" study.3) Forty proteins were unambiguously identified by MALDI-TOF-MS/MS, including 10 up-regulated spots,27 down- regulated spots, and 3 up-down-regulated spots. These proteins were mainly involved in disease mutation, folding of proteins, signaling pathway, reversible hydration of carbon dioxide. Parts of them were involved in energy metabolism, including Glycolysis/gluconeogenesis pathway, acetylation, Oxidation-reduction reaction and so on. GAPDH,TPI1,ALDOA,YWHAZ are the central elements of the protein- protein interaction network in 23 different expression proteins. The pattern of proteins expression, validated by Western blots and immunohistochemistry experiments, were consistent with the 2D-DIGE experiments.4) Colorectal cancer tissue microarray of 103 samples were followed up,84 samples survival information were obtained. The median follow-up time was 30.56 months and OS (overall survival) survival rate was 71.4%(60/84)。The clinical and pathological of "Age" and "lymph node metastasis" are closely related with the CRC prognosis(p<0.000). TNM staging system has good potential to predict the prognosis for the early and advanced stages of CRC(p<0.001). But it fails to make prognosis for TNMⅢandⅣpatients(p=0.483), while significant difference(p<0.000) exists between the actual prognosis of them.The proteins of ALDOA, PPIA, ANXA4, GRP78, CA1, YWHAZ, ALDOA were clustered to the Som 1,2,5 expression model, among which GRP78 and YWHAZ proteins can predict the prognosis for TNM stageⅢ-ⅣCRC patients(p=0.015; p=0.039). Except ANXA4, the expression of other 5 proteins are related to the prognosis of TNM stageⅠ-ⅣCRC patients(p<0.05).Discriminant-functions were established by leave-one-out cross-validation (LOOCV) analysis and survival analysis in CRC, Four kinds of molecular classifications, which can make good prognosis for different stages of CRC, were acquired by the discriminant-functions among them. The molecular classed with GRP78, ALDOA, CAland PP1A proteins could act as the molecular prognosis for TNM stage I-IVCRC patients(p<0.001) with accuracy of 74%. Co-expression of GRP78, ALDOA and CA1 proteins could act as the molecular prognosis for TNM stageIII-IVCRC patients(p<0.001) with accuracy of 76.9%. Molecular classification with a total of 6 proteins has 81.3% accuracy to predict prognosis for patients in stage III. ALDOA and YWHAZ proteins molecular can make prognosis for CRC stageIV patients with accuracy of 84.2%.4 Discussion1) The expression profile of different proteins in different stages of CRC was established:the regularity of protein variation in different stages of CRC by pattern analysis shows that the protein expression changes dynamically during the progression of CRC; molecular events occur before histological changes; there is a "Regression phenomenon" in stage II; different expression patterns of proteins represent their different roles in CRC.2) The "Omics docking strategies" study reveals that the consistency of mRNA, miRNA and proteins expression can increase the credibility of molecular change in CRC. However the further docking strategies should pay much more attention to the function and regulated mechanism in CRC. Not only the expression difference of molecular but also the consistency and inconsistency of the molecular expression in genoem, transcriptoem and proteom are significant to the biological CRC carcinogenesis.3) The traditional TNM staging can not provide a good prognosis for advanced stage CRC patients. However, the molecular classifications based on the differential expression of proteins can make good prognosis for CRC patients, especially for the patients with lymph node metastasis or distant metastasis.5 Conlcusion1)In this study, the molecular profile of different stages of CRC was established and the dynamic expression patterns for differentially expressed proteins in different stages of CRC were reported for the first time. 2) We integrated the information about gemoe, transcriptome, proteome in the same CRC samples. It is the first time that "Omics docking strategis" were implied to compare the expression of mRNA, miRNA and protein in CRC with different stages.3) Molecular classifications for different stages of CRC patients was obtained. Moreover, the molecular classifications, which were based on the different expression of proteins, were good supplementary for TNM stage III and IV patients, especially for the patients with lymph node metastasis or distant metastasis.