Cancer proteomics : characterisation of protein expression in human epithelial tumours

Abstract

Defining differences in protein expression between tumour cells of various degrees of malignancy is a first step in the development of markers for tumour diagnosis. In this thesis, the expression of multiple proteins in benign and malignant human ovarian, breast and prostate tumours was examined using two-dimensional gel electrophoresis (2-DE).

We were able to show a similar pattern of expression of a set of ten proteins between benign and malignant cells in all three tumour types examined. Cytokeratins and high molecular weight tropomyosins were consistently down-regulated in carcinomas, whereas stress proteins (HSP90, HSP60 and calreticulin) were upregulated. This finding suggests a high degree of homology in the expression of these proteins among different tumour types of epithelial origin.

An attempt to apply principal component analysis of quantitative 2-DE data for diagnosis of ovarian cancer is presented. Data of the expression of 170 polypeptides was compiled from 22 tumours and used to construct a model for classification into benign, borderline and malignant ovarian tumours. When the model was tested using 18 tumours, 11 tumours (61%) were correctly classified. We were encouraged by this result and intend to increase the number of tumours used to construct the model. Future work will show whether it is possible to accurately classify tumours by their gene expression profiles.

Twenty proteins in the 2-DE maps of breast, lung and ovarian tumours were identified using mass spectrometry. Some of these proteins were found to consist of polypeptide fragments, suggesting the occurrence of proteolytic processing of polypeptides in these tumours.

The process of tumour progression leads to the development of tumour heterogeneity. The 2-DE technique was used to study intra- and intertumour heterogeneity. Our results suggested that the degree of intertumour heterogeneity was substantial, whereas intratumoural variations were less pronounced.

We conclude that 2-DE separation of proteins in human tumours can yield new information relevant to the understanding of tumour biology. We believe that this line of research will lead to improved diagnostic and predictive tools.