Which of the following statements best defines combinatorial control of gene expression

Transcription is a highly regulated process controlled in large part by transcription factors (TFs) that specify when, where and how eukaryotic genes are expressed. TFs are operationally defined here as proteins that bind to DNA in a sequence-specific fashion. The number of TFs in an organism is significantly smaller than the number of genes that need to be controlled with exquisite temporal and spatial expression patterns. For example, Arabidopsis has ~ 28,000 protein-coding genes, but only ~ 2000 TFs [1], [2], [3], although it is likely that additional TF families remain to be identified in the ~ 8000 proteins of yet unknown function. Similarly, the maize genome encodes ~ 2700 TFs [4] from around 33,000 protein-coding genes [5]. Indeed, across the eukaryotes, TFs represent 5–10% of all genes [1], [6].

Combinatorial control provides a mechanism to explain the complexity of gene expression patterns. A TF may be part of different protein complexes that determine different types of regulation of different targets. Therefore, a composition of a protein complex and not a TF per se, represents an input leading to distinct gene expression patterns as an output.

Gene function is intimately linked to when and where genes are expressed. This information is hardwired in the gene regulatory regions formed in part by cis-regulatory elements (CREs) recognized by specific TFs [7], [8]. CREs are often located immediately upstream of the transcription start site (TSS) in what is generally known as the promoter. However, CREs are also integral part of enhancers, and can be found in 5′ UTR, introns [9], or 3′ of the genes they control [10]. The modular nature of gene regulatory regions is captured by the arrangement of CREs into cis-regulatory modules (CRMs), each responsible for executing a fraction of the overall gene regulation. Several TFs can come together and bind to each one of these regulatory modules. These DNA modules can cooperatively function following rules that often resemble digital logic with the output being the overall regulation of the gene [11]. Combinatorial control has been extensively studied from the perspective of cis-regulatory systems, i.e., how CRMs are arranged to produce distinct gene expression outputs [12]. In plants, the promoter of the viral CaMV 35S gene continues to provide one of the best examples of how CRM arrangements contribute to the expression of a gene in many plant tissues [13], [14], [15].

The identification of functionally relevant regulatory motifs starts with defining TSSs and other important gene landmarks (e.g., 3′ ends and introns). Similar to alternative splicing, a TSS can be affected by genetic variation or by development, as recently shown in maize [16]. As a consequence of alternative TSSs, genes with different TSSs are expected to have their own regulatory regions that may or may not involve shared CREs and CRMs. The identification of functionally important CREs often involves investigating conservation between co-regulated genes, or across related species in what is called phylogenetic shadowing or footprinting [17], [18]. A combination of these methods was recently used to identify CREs recognized by the ethylene response factors RELATED TO APETALA2.12 (RAP2.12) and RAP2.2 in hypoxia responsive genes [19]. Phylogenetic shadowing also permitted identification of evolutionarily conserved CREs that combine to control the expression of the GIGANTEA (GI) circadian clock protein [20].

TFs usually bind to short (5–8 bps long) DNA sequences that correspond to a consensus sequence and are represented for example by position weight matrices (PWMs). Indeed, a TF can recognize a broad range of DNA sequences in vitro with varying affinities, which range from the nanomolar to the micromolar range. Clearly, the short DNA sequences frequently recognized by a single TF in vitro, are insufficient to explain the affinity and specificity of binding in vivo [21]. For example, RAP2.2 and RAP2.12 bind in vitro the 5′-ATCTA-3′ sequence, which does not correspond to the CREs required for the regulation of RAP2.2/RAP2.12 targets [19]. While recent studies suggest a significant overlap between in vitro binding of TFs identified by DNA affinity purification sequencing (DAP-Seq) and chromatin immunoprecipitation (ChIP)-based experiments [22], there are many other examples in the literature of TFs that recognize a DNA motif in vitro which is not the top identified CRE in vivo. In many instances, this could be a consequence of indirect binding (i.e., through another TF) [23]. Not surprisingly, many TF families are identified by the presence of protein-protein interaction domains (e.g., the helix-loop-helix in bHLH, and the leucine zipper in bZip domains) that permit TFs to increase both affinity as well as specificity for DNA binding through the formation of homo- and heteromers. These interactions are often dynamic and central to combinatorial gene control.