Hierarchical graphical model reveals HFR1 bridging circadian rhythm (昼夜节律)and flower development in Arabidopsis thaliana(拟南芥)
- Luonan Chen &
- npj Systems Biology and Applicationsvolume 5, Article number: 28 (2019) | Download Citation
Abstracts
To study systems-level properties of the cell, it is necessary to go beyond individual regulators and target genes to study the regulatory network among transcription factors (TFs). However, it is difficult to directly dissect(剖析) the TFs mediated genome-wide gene regulatory network (GRN) by experiment. Here, we proposed a hierarchical graphical model to estimate TF activity from mRNA expression by building TF complexes with protein cofactors and inferring TF’s downstream regulatory network simultaneously(同时地). Then we applied our model on flower development and circadian rhythm processes in Arabidopsis thaliana. The computational results show that the sequence specific bHLH family TF HFR1 recruits the chromatin regulator HAC1 to flower development master regulator TF AG and further activates AG’s expression by histone acetylation(组蛋白乙酰化). Both independent data and experimental results supported this discovery. We also found a flower tissue specific H3K27ac ChIP-seq peak at AG gene body and a HFR1 motif in the center of this H3K27ac peak. Furthermore, we verified(证实) that HFR1 physically interacts with HAC1 by yeast two-hybrid experiment. This HFR1–HAC1–AG triplet relationship may imply that flower development and circadian rhythm are bridged by epigenetic regulation and enrich the classical ABC model in flower development. In addition, our TF activity network can serve as a general method to elucidate (阐释、说明)molecular mechanisms on other complex biological regulatory processes.
Fig. 1
TFA network (FDCRNet) construction workflow for Flower Development and Circadian Rhythm processes. Key molecules (TFs) for circadian rhythm and flower development are curated from GO annotation database. Functional linkage and gene expression data are integrated step-by-step to construct the transcriptional regulation network, modulation regulatory network, and TFA network for those key molecules
Fig. 2
Modulators affect the pattern of the transcriptional regulation. a Expression pattern of HAC1 and AG. Each dot corresponds to the gene expression of HAC1 and AG in one sample. MI represents the mutual information score. b Top, expression pattern of a modulator–TF–target triplet: HFR1, HAC1, and AG. Each column represents a sample, which is sorted according to the expression level of the modulator (HFR1). The mutual information (MI) score of the AG and HAC1 on all the samples is 0.53. Middle, expression pattern of TF-target (HAC1 and AG) in HFR1 lowly expressed samples (bottom 35% samples). MI score is 0.34. Bottom, expression pattern of TF-target (HAC1 and AG) in HFR1 highly expressed samples (top 35% samples). TF and target gene are strongly correlated across those samples and the MI score is 0.72. The conditional mutual information (CMI) score is 0.38. c Significance test of the conditional mutual information of the HFR1-HAC1-AG. We randomly chose two groups of samples (each group contains 35% samples) and computed the mutual information between TF and target (HAC1-AG) in each group. We repeated this procedure 1000 times to generate the null distribution. Given HFR1’s expression level, the mutual information of HAC1 and AG is significantly changed (p-value < 0.001). d A summary for the example of the HFR1-HAC1-AG