Multi - tissue analysis of gene expression patterns in

24 tissues and changes in their cancers.

One problem of experimental research on molecular processes in cells is, that often only a few genes, gene products, cell types or tissues are investigated. Thus it is difficult to decide if a gene is a general cancer gene or only typical for the specific tissue represented by the investigated cell type, which might have been chosen not for basic, but for technical reasons.

We have decided obtain, in cooperation with B.Altenberg, EMBL, an overview of cancer - relevant molecules in essentially all cancers. This means, one has to combine experimental work with excessive datamining. We used NIH’s database dbEST, which provides data on the expression of essentially all human genes in 51 normal tissues and their cancers. We selected 24 of these tissues, for which the statistical quality of the data is already sufficient to obtain robust results. The 24 tissues represent appox. 70 % of all clinical cancer cases

It is known since almost a century that in certain cancers glycolysis is upregulated (Warburg effect). However, this has been definitely proven only in a few selected cancers. We could classify 24 cancers according this criterion and show, that most genes of glycolysis are up - regulated in the majority of them (Altenberg and Greulich, 2004 GENOMICS 84, 1014 -1020)

In a second line of investigation we have shown that among the 20 000 human genes only approx. 50 (including the just mentioned glycolysis genes) are upregulated in the majority of cancers. All other genes are up –regulated only in a few cancers. (Altenberg, Gemuend and Greulich 2006 PROTEOMICS 6, 67 -71)

Lung cancer (and others) have many genes which are highly up- or down- regulated, whereas in colon cancer, only few genes are up - regulated by more than a factor of 2 or down - regulated to less than a factor of 1/2. The latter finding is in line with the well known observation, that in colon cancers mutations in the coding region of genes play an important role.

If one looks at the degree of expression of single genes in 24 normal tissues and their cancers, no consistent trend is seen. If one looks, however, at whole groups of genes (for example at all 10 genes of glycolysis or at sixty two p53 binding proteins), their cumulated expression is surprisingly uniform. The expression in cancer is independent of the expression in normal tissue and is more uniform in cancer than in the 24 normal tissues

(Altenberg, Rapp, Schmitt and Greulich 2006 GENOMICS, submitted)

In the case of colon cancer, we, in cooperation with B.Pool –Zobel, J.Sauer and T.Hoffmann, FSU, could show that, at least for a number of some selected genes, predictions on the under – or overexpression made from the dbEST database correlate well with colon cancer patient data (University Hospital Jena, W.W.Richter), when the latter are averaged over several patients, i.e. when individual features of single patients are no longer relevant.