Morten Krogh and Kasper Astrup Eriksen
A connection between co-expression and the strength of transcription factor binding.


Abstract
Background:
An important factor in the cellular machinery is the relation between the combination of transcription factors binding upstream of a gene and the resulting expression level of the gene. Unravelling the interactions and interplay between different transcription factors is thus very important. Here we attack this issue by first asking what is most important in determining the influence of a transcription factor on its downstream target: The intrinsic binding characteristics of the DNA - transcription factor complex or whether there are other stronger transcription factors capable of binding in the same upstream region. As an intrinsic measure of the DNA - transcription factor binding we use the p-value of binding reported in location experiments.

Results:
First we demonstrate that the p-value is indeed a biological meaningful measure by showing that its logarithm is almost linearly related to how often a gene and its transcription factor are co-expressed in microarray experiments. This connection is, as expected, strongest for microarray experiments performed in media resembling the medium where the location p-values were measured. Somewhat surprisingly we find that the existence of stronger transcription factors does not affect how often a gene and its transcription factor are co-expressed in Saccharomyces Cerevisiae.

Discussion:
Our results indicate that important aspects of the transcriptional regulation are already captured by the intrinsic properties of the DNA - transcription factor complex. In fact we failed to detect any signs of internal competition between transcription factors beyond what already is encoded in the p-value obtained from location experiments. Our results emphasize that the individual transcription factors do not simply act in an on/off fashion, but come with a weight. This weight is at least partially given by the p-value of binding as measured in location experiments.

LU TP 03-30