Inferring cell cycle feedback regulation from gene expression data

Abstract

Feedback control is an important regulatory process in biological systems, which confers robustness against external and internal disturbances. Genes involved in feedback structures are therefore likely to have a major role in regulating cellular processes. Here we rely on a dynamic Bayesian network approach to identify feedback loops in cell cycle regulation. We analyzed the transcriptional profile of the cell cycle in HeLa cancer cells and identified a feedback loop structure composed of 10 genes. In silico analyses showed that these genes hold important roles in system's dynamics. The results of published experimental assays confirmed the central role of 8 of the identified feedback loop genes in cell cycle regulation. In conclusion, we provide a novel approach to identify critical genes for the dynamics of biological processes. This may lead to the identification of therapeutic targets in diseases that involve perturbations of these dynamics.

Figure 1. A dynamic Bayesian network and its translation into a gene regulatory network

A) Example of a simple dynamic Bayesian network representing the probabilistic dependencies of four variables (A-B-C-D) between two consecutive time points; B) The network in A) translated into a gene regulatory network. This representation facilitates the identification of the feedback loop involving variables A-C-B.

Figure 2. Gene network inferred analyzing human cell cycle expression data

Relying on the expression values for 1099 probes measured by Whitfield et al. [32] and on our dynamic Bayesian network inference algorithm, we inferred a gene regulatory network. This network contains a large group of 412 connected probes, shown in the Figure.

Figure 3. Inferred feedback loops

Twelve nodes in the network in Figure 2 are involved in interrelated feedback loops; these probes map to ten different genes. The Figure shows the relationships between the loop nodes.

Figure 4. Fitting assessment

The Figure shows the measured (blue) and fitted (red dashed) profiles for four loop probes. The data are shown starting from the second time point, as the first one is always taken equal to the first measured value.

Figure 5. Predictive accuracy assessment on an independent test set

The Figure shows the measured (blue) and predicted (red dashed) profiles for the same loop probes as in Figure 4 but relative to the independent expression dataset employed to evaluate our network model.

Figure 6. Assessment of the robustness of the inferred feedback loops

The Figure shows the relationships between the loop nodes annotated with the corresponding BF. In cases in which a gene has two parents, the BF of the first added parent (BF₁₀) is indicated with [1] and that of the second parent (BF₂₁) with [2].