Virtual mutagenesis of the yeast cyclins genetic network reveals complex dynamics of transcriptional control networks

Abstract

Study of genetic networks has moved from qualitative description of interactions between regulators and regulated genes to the analysis of the interaction dynamics. This paper focuses on the analysis of dynamics of one particular network--the yeast cyclins network. Using a dedicated mathematical model of gene expression and a procedure for computation of the parameters of the model from experimental data, a complete numerical model of the dynamics of the cyclins genetic network was attained. The model allowed for performing virtual experiments on the network and observing their influence on the expression dynamics of the genes downstream in the regulatory cascade. Results show that when the network structure is more complicated, and the regulatory interactions are indirect, results of gene deletion are highly unpredictable. As a consequence of quantitative behavior of the genes and their connections within the network, causal relationship between a regulator and target gene may not be discovered by gene deletion. Without including the dynamics of the system into the network, its functional properties cannot be studied and interpreted correctly.

Figure 1. Time varying network motifs.

Figure 2. Transition of network structures given by Figure 1 .

Darkness of the gene node and its connection reflects the influence of the regulator to the regulated gene's temporal expression level.

Figure 3. Yeast cyclins network topology reconstructed from binding experiments, gene interactions modeling (Eq. 1) and confirmed by literature search.

A. Cyclic diagram. B. causal relationship diagram.

Figure 4. Expression profiles of the genes of the last layer of Figure 3B (rows) resulting from the virtual deletion of the genes of the first layer of Figure 3B - FKH1, FKH2, MCM1, SWI6 and CLN3 (columns).

Thin lines represent measured expression profiles, thick lines represent computed profiles after virtual deletion of genes in the columns.

Figure 5. Regulatory pathway of CLB4.

Figure 6. Expression profiles of the genes of the last layer of Figure 3B (rows) resulting from the virtual deletion of the genes of the intermediate layer of Figure 3B - SWI5, NDD1, SWI4 and MBP1 (columns).

Thin lines represent measured expression profiles, thick lines represent computed profiles after virtual deletion of genes in the columns.