Transcriptional regulation in plants: Using omics data to crack the cis-regulatory code
Introduction
The rapid accumulation of omics data over the last decade provides an unprecedented opportunity for the system-level reconstruction of the molecular mechanisms governing plant growth and development [1,2,3∗∗,4]. Of all the regulatory events that shape the phenotypic manifestation of the plant genome, the transcriptional regulation of gene expression is a fountainhead. Chromatin immunoprecipitation-sequencing (ChIP-seq) techniques enable the annotation of genome-wide transcription factor (TF)–binding events, whereas the recently developed DNA affinity purification and sequencing (DAP-seq) approach [5∗∗] and the development of an efficient protoplast isolation/transformation system for expressing epitope-tagged TFs for ChIP-seq [6∗∗] have made feasible the large-scale profiling of TF binding. The application of these techniques revealed the binding profiles for a large number of Arabidopsis and maize TFs [5∗∗,6∗∗].
Deciphering the cis-regulatory code from these omics data provides opportunities for unveiling condition-specific regulatory programs [7, 8, 9, 10]. The main challenge is that the sequence is only one of many parameters determining whether a TF binds its target DNA and whether the binding event affects gene expression. One must also consider information about the epigenetic marks, chromatin accessibility, abundances of TFs and their interactors, specificity of protein–DNA biophysical interactions (which might be distinct for dimers and multimers of different compositions), and other factors that contribute to triggering a regulatory program. Nonetheless, the detection of TF-binding sites (TFBSs) in gene regulatory regions is a prerequisite for the study of regulatory events, including the cooperative action of TFs, which guide condition-specific gene expression [8, 9, 10]. This review summarizes the state-of-the-art bioinformatics approaches used to obtain the maximum information from the regulatory sequences to understand transcriptional regulatory patterns, with a focus on applications in plant research.
Section snippets
Input data: whole-genome annotation data and binding site models
Various genome-wide data sets can be used as input to characterize the functions of the transcriptional machinery and elucidate regulatory patterns (Figure 1). In addition, TFBS models provide valuable input for understanding the principles of transcriptional regulation.
From a single TFBS to a cis-regulatory syntax
A standard step for analyzing a single chromatin-precipitation peak set is de novo motif discovery, which detects overrepresented sequence patterns (motifs) using, for example, the HOMER tool [20]. In addition to identifying the most common variant of the TFBS, de novo motif discovery reveals other overrepresented motifs. These may include minor variants of the TFBS, reflecting specific regulatory modes. The crystal structures of the TF–DNA complexes recently solved for the developmental
Searching for TF-binding targets
To identify TF target genes from ChIP-seq data, it is necessary to associate the peak positions with the gene regulatory regions. The regulatory regions predominantly localize upstream of the transcription start site; however, they are also often found downstream of a gene, in introns, in untranslated regions, and even in protein-coding sequences. Such ‘non-upstream’ regulatory regions were identified in the TERMINAL FLOWER 1 (TFL1), AGAMOUS-LIKE 6 (AGL6), and LEAFY (LFY) floral development
Does gene regulation follow from TF binding?
TF–DNA binding near a gene does not necessarily induce changes in transcription of the gene. Therefore, it is crucial to test if the presence of the cis-regulatory element is associated with a transcriptional response. Coupling TF-binding targets deduced from the ChIP-seq peak set and data on differential gene expression (e.g. triggered by TF perturbation in knockout or overexpressing transgenic lines) is a straightforward approach to distinguish regulatory TF-binding events [37,45] (Figure 2).
Machine learning applications in plant functional genomics
Machine learning (ML) methods are a promising alternative to inferential statistics for deducing the cis-regulatory code from whole-genome annotation data [55,56]. ML offers a hypothesis-free approach to modeling the regulatory outcome by learning the transcription regulation signatures directly from functional genomics data sets. Taking into account the contribution of numerous features, this approach is beneficial for prediction of regulatory TF-binding events and gene expression. For this,
Future prospects
Because the cis-regulatory code is highly complex and manifests itself very specifically in each cell type, further progress in its comprehensive reconstruction requires a much larger amount of high-quality data from model plant species. In this regard, a plant initiative similar to the human ENCODE project would be invaluable [12∗, 67]. Rapid technological advancement in single-cell multiomics provides new powerful tools to study cis-regulatory patterns at the single-cell resolution. An
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgements
The authors thank Tatyana Merkulova and Pavel Borodin for fruitful discussions and anonymous reviewers for valuable advice. This work was supported by the Russian Science Foundation, grant no. 20-14-00140.
References (70)
- et al.
Cistrome and epicistrome features shape the regulatory DNA landscape
Cell
(2016) - et al.
Reconstructing the maize leaf regulatory network using ChIP-seq data of 104 transcription factors
Nat Commun
(2020) - et al.
Capturing auxin response factors syntax using DNA binding models
Mol Plant
(2019) - et al.
The UCSC Genome Browser database: 2021 update
Nucleic Acids Res
(2021) - et al.
Genome encode analyses reveal the basis of convergent evolution of fleshy fruit ripening
Nat Plants
(2018) - et al.
Mass-spectrometry-based draft of the Arabidopsis proteome
Nature
(2020) - et al.
Diversity of cis-regulatory elements associated with auxin response in Arabidopsis thaliana
J Exp Bot
(2018) - et al.
Widespread long-range cis-regulatory elements in the maize genome
Nat Plants
(2019) - et al.
Base-resolution models of transcription-factor binding reveal soft motif syntax
Nat Genet
(2021) - et al.
Network-based approaches for understanding gene regulation and function in plants
Plant J
(2020)