Research Article

A transcription factor hierarchy defines an environmental stress response network

+ See all authors and affiliations

Science  04 Nov 2016:
Vol. 354, Issue 6312, aag1550
DOI: 10.1126/science.aag1550

You are currently viewing the abstract.

View Full Text

Complex transcription factor interactions

To respond to environmental changes, such as drought, plants must regulate numerous cellular processes. Working in the model plant Arabidopsis, Song et al. profiled the binding of 21 transcription factors to chromatin and mapped the complex gene regulatory networks involved in the response to the plant hormone abscisic acid. The work provides a framework for understanding and modulating plant responses to stress.

Science, this issue p. 598

Structured Abstract


Transcription factors (TFs) are often studied one by one or in clusters of a few related factors. However, the integration and networks of transcriptional changes to response to environmental stresses often involve many related TFs. In many organisms, such as plants, overlapping functions can make it difficult to understand how a biologically relevant end result can be achieved via the complex signaling networks controlled by these TFs. To better understand how the reference plant Arabidopsis deals with the stresses incurred by water limitation via the hormone abscisic acid (ABA), we characterized all DNA sequences that bind to the 21 ABA-related TFs in vivo.


There have been limited systematic studies of stress-responsive TF networks in multicellular organisms. We chose ABA, an essential plant hormone that is required for both development and responses to osmotic stress, as an elicitor to investigate complex gene regulatory networks under stress. Combining differential binding (DB) of 21 ABA-related TFs at a single time point measured by chromatin immunoprecipitation sequencing (ChIP-seq) with differentially expressed genes from a time-series RNA sequencing (RNA-seq) data set, we analyzed the relationship between DB of TFs and differential expression (DE) of target genes, the determinants of DB, and the combinatorial effects of multi-TF binding. These data sets also provide a framework to construct an ABA TF network and to predict genes and cis-regulatory elements important to ABA responses and related environmental stresses.


We found that, in general, DNA binding is correlated with transcript and protein levels of TFs. Most TFs in our study are induced by ABA and gain binding sites (termed “peaks”) after the hormone treatment. ABA also increases the binding of the TFs at most peaks. However, in some peaks, TF binding may be static or even decrease after ABA exposure, revealing the complexity of locus-specific gene regulation. De novo motif discovery enabled us to identify distinct, primary motifs often centrally localized in the ChIP-seq peaks for most TFs. However, it is not uncommon to find motifs, such as the G-box, that are shared by peaks from multiple TFs and may contribute to binding dynamics at these sites. DB of multiple TFs is a robust predictor of both the DE and ABA-related functions of the target genes. Using the DB and DE data, we constructed a network of TFs and canonical ABA pathway genes and demonstrated a regulatory hierarchy of TFs and extensive feedback of ABA responses. On the basis of a “guilt-by-association” paradigm, we further predicted genes whose functions were previously not linked to ABA responses, and we thus functionally characterized a new family of transcriptional regulators.


These data sets will provide the plant community with a roadmap of ABA-elicited transcriptional regulation by 21 ABA-related TFs. We propose that dynamic, multi-TF binding could be a criterion for prioritizing the characterization of TF binding events, cis-regulatory elements, and functionally unknown genes in both plants and other species. In our proof-of-principle experiments, ectopic expression of the transcriptional regulators ranked highly in our model results in altered sensitivity to both ABA and high salinity. Together with the fact that our modeling recovered genes related to seed development and osmotic stresses, we believe such predictions are likely applicable to a broad range of developmental stages and osmotic stresses.

Transcriptional landscape of the ABA response.

ABA response pathway gene targets were identified by large-scale ChIP-seq and time-series RNA-seq experiments. A network model was built to reveal the hierarchy of TFs and the impact of multi-TF dynamic binding on gene expression. A new family of transcriptional regulators was predicted by the model and was functionally tested to evaluate the role of these regulators in osmotic stress in plants.


Environmental stresses are universally encountered by microbes, plants, and animals. Yet systematic studies of stress-responsive transcription factor (TF) networks in multicellular organisms have been limited. The phytohormone abscisic acid (ABA) influences the expression of thousands of genes, allowing us to characterize complex stress-responsive regulatory networks. Using chromatin immunoprecipitation sequencing, we identified genome-wide targets of 21 ABA-related TFs to construct a comprehensive regulatory network in Arabidopsis thaliana. Determinants of dynamic TF binding and a hierarchy among TFs were defined, illuminating the relationship between differential gene expression patterns and ABA pathway feedback regulation. By extrapolating regulatory characteristics of observed canonical ABA pathway components, we identified a new family of transcriptional regulators modulating ABA and salt responsiveness and demonstrated their utility to modulate plant resilience to osmotic stress.

View Full Text

Related Content