Regulons and Regulatory Motif Characterization

G C A T T A C G G C A T genes

Review Beyond Trees: Regulons and Regulatory Motif Characterization

Xuhua Xia 1,2

1 Department of Biology, University of Ottawa, Ottawa, ON K1N 6N5, Canada; [email protected] 2 Ottawa Institute of Systems Biology, Ottawa, ON K1H 8M5, Canada

Received: 24 May 2020; Accepted: 24 August 2020; Published: 25 August 2020

Abstract: Trees and their seeds regulate their germination, growth, and reproduction in response to environmental stimuli. These stimuli, through signal transduction, trigger transcription factors that alter the expression of various genes leading to the unfolding of the genetic program. A regulon is conceptually defined as a set of target genes regulated by a transcription factor by physically binding to regulatory motifs to accomplish a specific biological function, such as the CO-FT regulon for flowering timing and fall growth cessation in trees. Only with a clear characterization of regulatory motifs, can candidate target genes be experimentally validated, but motif characterization represents the weakest feature of regulon research, especially in tree genetics. I review here relevant experimental and bioinformatics approaches in characterizing transcription factors and their binding sites, outline problems in tree regulon research, and demonstrate how transcription factor databases can be effectively used to aid the characterization of tree regulons.

Keywords: gene expression; transcription factor; regulon; regulatory motifs; comparative genomics; Gibbs sampler

1. Introduction Research in tree genetics, especially in gene regulation networks, has benefitted much from looking beyond trees. The finding of animal c-MYB family of genes resulted in the discovery of the R2R3-MYB family of transcription factors in Arabidopsis thaliana [1,2], and the understanding of how A. thaliana responds to cold through gene regulation sheds light on how tree species such Malus demestica respond to cold [3]. I present a conceptual framework here to highlight some shortcomings in tree regulon studies to facilitate better experimental designs in the future. The term regulon was originally proposed [4] as an extension to operon, in the context of a repressor controlling the expression of multiple genes that are not located within the same operon. For example, the eight genes involved in the biosynthesis of arginine are scattered in five separate genomic locations in the genome of Escherichia coli, but controlled by the same arginine repressor [4]. Since its early usage, “regulon” has been associated with specific biological processes, e.g., the argR (the E. coli gene encoding the arginine repressor) regulon for arginine biosynthesis, or the CO-FT regulon for flowering timing and seasonal growth in Arabidopsis thaliana [5] and in trees [6]. The adoption of Arabidopsis thaliana as a model for research in plant molecular biology has resulted in the discovery and characterization of regulons in Arabidopsis thaliana and related plants, but also in various tree species. Some regulons function similarly, such as the CBF regulon in plant response to low temperature [3,7,8] which is gated by a circadian clock in both Arabidopsis thaliana [9,10] and peach trees [11]. However, some other regulons appear to function differently. For example, the CO-FT regulon controls genes involved in seasonal growth and flowering time in Arabidopsis thaliana [5], but flowering and growth cessation appear to be two different processes in Populus species. While

Genes 2020, 11, 995; doi:10.3390/genes11090995 www.mdpi.com/journal/genes Genes 2020, 11, 995 2 of 17

flowering time is still under the control of the CO-FT regulon [6], seasonal growth cessation is regulated by anGenes additional 2020, 11, x FOR set ofPEER genes REVIEW [12 ]. 2 of 17 Many different experimental and bioinformatics methods have been used to discover and characterizePopulus species. regulons, While but flowering the large-scale time is approachstill under of the combining control of ChIP-seqthe CO-FT and regulon RNA-seq [6], seasonal technology has beengrowth the cessation most frequently is regulated used by in an recent additional years set [13 of,14 genes] evolving [12]. from the microarray method that had Many different experimental and bioinformatics methods have been used to discover and been used successfully in the characterization of the CESA regulon for cellulose biosynthesis [15,16]. characterize regulons, but the large-scale approach of combining ChIP-seq and RNA-seq technology Newhas derivatives been the most from frequently ChIP-seq, used such in asrecent ChIP-exo years [13,14] [17,18 evolving] ChEC-seq from [ 19the] microarray and Cut&Run method [20 ],that as well as newhad technologiesbeen used successfully such as in DamID-seq the characterization [21], have of the been CESA developed regulon for in recentcellulose years. biosynthesis I evaluate these[15,16]. experimental New derivatives methods from and ChIP-seq, bioinformatics such as ChIP-exo algorithms [17,18] after ChEC-seq a formal [19] definition and Cut&Run of regulon [20], and regulonas well network. as new technologies such as DamID-seq [21], have been developed in recent years. I evaluate these experimental methods and bioinformatics algorithms after a formal definition of regulon and 2. Aregulon Formal network. Definition of Transcription Regulon and Regulon Network A conceptually ideal transcription regulon should have four features: (1) a transcription factor, 2. A Formal Definition of Transcription Regulon and Regulon Network (2) one or more target genes whose expression is regulated by the transcription factor, (3) a set of conservedA transcriptionconceptually ideal factor transcription binding sitesregulon (TFBS) should on have the four target features: genes, 1) and a transcription (4) specific factor, biological 2) one or more target genes whose expression is regulated by the transcription factor, 3) a set of functions accomplished by altering the expression of the target genes. All regulons I mention in this conserved transcription factor binding sites (TFBS) on the target genes, and 4) specific biological review are transcription regulons. functions accomplished by altering the expression of the target genes. All regulons I mention in this reviewThe importance are transcription of feature regulons. 3 above may be illustrated in Figure1. Suppose a transcription factor A (tfA) canThe bind importance to TFBS of Xfeature and regulate3 above may three be targetillustrated genes, in FigureT1, T2 1.and Suppose a transcription a transcription factor factor B (tfB), that shareA (tfA) TFBS can bind X. tfB to proteinTFBS X and can regulate in turn bindthree target to TFBS genes, Y to T1 control, T2 and the a transcription expression offactor target B (tfB genes), T4 and T5thatthat share share TFBS TFBS X. tfB Y protein in their can regulatory in turn bind region. to TFBS Criterion Y to control 3 implies the expression that the of tfA-regulon target genes includes T4 targetand genes T5 thatT1, shareT2, and TFBStfB Y, butin their not regulatoryT4 and T5 region.whose Criterion TSBS Yis 3 boundimplies bythat tfB the but tfA-regulon not by tfA. includes In contrast, the termtarget “coregulated genes T1, T2, genes” and tfB would, but not typically T4 and includeT5 whoseT1, TSBS T2, tfB,Y is T4, boundand byT5 tfBas coregulatedbut not by tfA. directly In by tfA orcontrast, indirectly the term by tfA “coregulated through tfB. genes” would typically include T1, T2, tfB, T4, and T5 as coregulated directly by tfA or indirectly by tfA through tfB.

tfA Cold stress (A) tfB Chr13 MdMYB124

T1 T2 tfA tfB T4 T5 MYB88

Chr16 MdMYB88 ? MdCSP3 T1 T2T4 T5

(B) (C) regulon 1 MYB124 Chr1 ? MdCCA1 Transcription, translation, and tfA nuclear import CCA1 Protein-DNA binding T1 T2 tfB Chr7 ? MdCBF3 TFBS X for tfA

TFBS Y for tfB T4 T5 (D) ? Unknown TFBS Cold hardiness regulon 2 FigureFigure 1. Illustration 1. Illustration of of a regulona regulon network network ofof twotwo interacting interacting regulons. regulons. (A) ( AtfA-regulon) tfA-regulon with withtarget target genesgenesT1, T2T1, T2 and, andtfB ,tfB and, and tfB-regulon tfB-regulon with with targettarget genes genes T4T4 andand T5T5. (B.() BLegends) Legends of graphic of graphic elements. elements. TFBS—TranscriptionTFBS—Transcription factor factor binding binding site. site. ((C) A simplified simpliﬁed representation representation of the of thetwo two regulons. regulons. (D) A ( D)A partialpartial regulon regulon network network [3 ][3] for for coping coping withwith cold stress stress in in apple apple (Malus (Malus demestica demestica), in), which in which TFBSs TFBSs remainremain poorly poorly characterized. characterized.

In rareIn rare cases, cases, two two didifferentﬀerent TFs TFs represent represent nearly nearly identical identical paralogues paralogues and feature and the feature same DNA- the same binding affinity. Examples include MSN2 on chromosome 13 and MSN4 on chromosome 11 of the DNA- binding aﬃnity. Examples include MSN2 on chromosome 13 and MSN4 on chromosome yeast Saccharomyces cerevisiae [22], and MdMYB124 (MYB88, transcription factor MYB51) on 11 of the yeast Saccharomyces cerevisiae [22], and MdMYB124 (MYB88, transcription factor MYB51)

Genes 2020, 11, 995 3 of 17 on chromosome 13 and MdMYB88 (LOC103402919, transcription factor MYB88) on chromosome 16 of the apple [3] shown in Figure1D. They should be designated as MSN2 /MSN4 regulon and MdMYB124/MdMYB88 regulon, respectively. Such redundancy may enhance regulon reliability in responding to critical environmental stimuli [22]. Regular research aims to characterize the four features of a regulon and to understand its biological function. Among the four features, TFBS is typically the worst characterized, partly through negligence and partly through poor experimental design. For example, the partial regulon network Figure1D) for developing cold hardiness in response to cold stress does not feature established TFBS because, although the researchers [3] took the ChIP-seq approach, the ineffective experimental design thwarted the effort. I discuss this in more detail later. Feature 1 of the regulon definition is essential for us to understand the dynamic nature of regulon operation. A transcription factor does not bind to TFBS of its target genes all the time, but only at specific time under specific environment, e.g., CO-FT regulon in response to seasonal change in photoperiod [5] or bZIP60 regulon in response to unfolded protein response (UPR) in endoplasmic reticulum [23,24]. This problem is particularly serious for lowly or transiently expressed TFs, and it is partly for this reason that the DAP-Seq approach with TFs highly expressed in vitro can often identify more TFBSs than alternative methods [25]. Thus, the arrow of protein-DNA binding results from collapsing over different time and intracellular/extracellular conditions, and consequently is not useful in predicting dynamic interaction of the regulon. Knowing the function of the regulon helps us understand when gene regulation should take place. The definition of transcription regulon above does not include gene regulation mediated by mechanisms other than transcription factor binding to TFBSs, i.e., it does not include gene regulation involving posttranscriptional or posttranslational modification, DNA methylation, or histone modification. For example, yeast HAC1 mRNA is exported into cytoplasm in an unspliced form, and is translated to produce functional proteins only after the HAC1 mRNA is spliced by Ire1 to remove the translation inhibition mediated by its intron binding to its 5’ UTR [26–28] However, we do not consider Ire1-Hac1 as a regulon because the regulation is not achieved by Ire1 binding to the TFBS of HAC1 to regulate the transcription of HAC1. A regulon could have just a single gene. For example, HAC1 is a transcriptionally autoregulated gene with its protein binding to its TFBS forming a positive feedback loop until Ire1 stops splicing HAC1 mRNA [29]. Thus, a transcriptionally autoregulated gene, with its protein regulating the expression of its own mRNA, can be a regulon without involving any other genes. A regulon network involves two or more interacting regulons. The interacting tfA-regulon and tfB-regulon (Figure1) constitute a regulon network. In this framework, the CO-FT regulon regulating flowering time in Arabidopsis thaliana [5] and in Populus species [6] would be considered a regulon network instead of a regulon. In fact, CO and FT are equivalent to tfA and tfB, respectively, in Figure1. Another example of a regulon network is the ICE1–CBF–COR regulon network [30–32] in which ICE1, in response to cold, binds to TFBS of CBFs to upregulate their expression, and CBFs bind to TFBS of their downstream cold-regulated genes (CORs) to enhance cold tolerance. The key difference between ICE1-CBF-COR regulon network and the regulon network in Figure1 is that CBFs include multiple genes instead of a single tfB. A regulon network, as represented in Figure1C with the interacting tfA-regulon and tfB-regulon, is a tree with height of 2 (Tree height is the maximum number of edges from the root to the tip). If T4 protein (Figure1) were a transcription factor regulating its own target genes, then the regulon network would be a tree of height 3. Because biological systems are typically noisy, a robust regulon network represented as a tree should have low height. In other words, if we have a long chain of tfA tfB tfC tfD ... , and if each link has a chance of failure, then the reliability of tfA triggering the → → → last transcription factor in the chain is reduced as the chain length (tree height) is increased. However, information propagation in biology is rarely through a single chain but more often through redundant pathways to increase robustness. Genes 2020, 11, 995 4 of 17

A regulon network (RN) differs from the conventional gene regulation network (GRN) in the basic unit of the network. A basic unit in RN is a regulon with its four features (TF, TFBS, target genes, and biological function), but that in a GRN is not explicitly defined, although two nodes in GRN connected by a directed or undirected line are assumed to be in regulator–regulatee relationship. An RN highlights which feature of a regulon is missing. For example, Figure1D shows clearly the missing TFBS without which one cannot scan the genome for candidate target genes and verify the regulator–regulatee relationship. While some regulons in tree species are well characterized [3,6,12], the regulon definition above is best exemplified by the Hac1-regulon in yeast Saccharomyces cerevisiae [33,34]. First, Hac1 is a well-characterized master transcription regulator with a basic-leucine zipper motif that controls the initiation of unfolded protein response triggered by an abnormal accumulation of unfolded/misfolded proteins in the endoplasmic reticulum (ER) [27,35–37]. Second, the regulator has a fairly well documented set of 381 target genes [38] serving two key functions related to restoring ER homeostasis: (1) increasing the folding capacity by upregulating the ER-resident chaperone proteins [39], and (2) decreasing folding demand by selectively degrading unfolded/misfolded proteins [38,40], extracellular export of unfolded/misfolded proteins [41], and reduction of protein synthesis [42]. Third, it features two well-validated TFBSs (unfolded protein response element, UPRE): UPRE1 (GACAGCGTGTC) [35,39,43,44] and UPRE2 (TACGTGT) [38,39,45,46] that is present in target genes, e.g., KAR2 [35]. Fourth, its function within UPR is well defined and specific, i.e., to restore ER homeostasis or, if the restoration fails, pass the signal to other cellular machinery to trigger apoptosis. Multiple approaches have been used to verify the association between TF (e.g., Hac1) and TFBS of target genes (e.g., KAR2). The first is to check if TF fails to bind to, and regulate, the target gene after removing or mutating the putative TFBS [35], or TF-binding occurs when the putative TFBS is inserted into another gene that does not originally have a TFBS and does not bind to TF. This cannot be done in large-scale and is done less frequently today. Second, if target genes T1, T2, and T3 harbor this TFBS, then their expression should differ between the wild type with a functional TF and a mutant without the TF. This is typically done as part of a ChIP-Seq experiment. One such experiment is exemplified by BioProject PRJNA633509 in the NCBI SRA (Sequence Read Archive) database, with the objective to characterize transcription factor narrow sheath1 (NS1), its target genes, and its binding sites in Zea mays. The ChIP-Seq data set would allow one to identify putative TFBSs in a large number of putative genes. Transcriptomic data were obtained from the wild type with a function NS1 and an ns1 mutant. A gene is likely NS1-regulated if it not only harbors a putative TFBS, but also differs in gene expression between the wild type and the mutant. The third approach is to compare sequence conservation of orthologous genes among related species. A TFBS should be strongly conserved but its flanking sequences should not. These approaches have been used to characterize the two UPREs above and the target genes of Hac1. That Hac1 can bind to two separate motifs is particularly relevant to regulon characterization and interpretation of experimental results. Yeast researchers were once puzzled by the observation that many genes whose transcription is driven by Hac1 do not have UPRE1 [38,45,46] Instead, they appear to share a different motif, later named UPRE2. UPRE1 was an experimentally validated TFBS for Hac1 [35,43,44]. In the conventional wisdom of one specific regulatory motif for one transcription factor, it was hypothesized that another transcription factor was involved, playing the role of tfB in Figure1, with Hac1 equivalent to tfA (Figure1). The equivalent of tfB was identified (or perhaps misidentified) as Gcn4 [45], i.e., Hac1 regulates GCN4 transcription and the resulting Gcn4 regulates the transcription of those yeast genes that do not have UPRE1 but have UPRE2. However, GCN4 does not have UPRE1, and its transcription remains unchanged during UPR [45], so it cannot possibly be equivalent to tfB in Figure1. It turns out that Hac1 is capable of binding to either UPRE1 or UPRE2 [ 39]. This example illustrates the importance of characterizing regulatory motifs of transcription factors. The detailed elucidation of yeast UPR demonstrates the success of long-term, persistent and focused research since early 1990s [33,34]. An equivalent success story in plant sciences is the characterization of Genes 2020, 11, 995 5 of 17

AtHB1 during roughly the same period [47–49], although much remains unknown in AtHB1-mediated regulation based on the four-feature criterion of a regulon. The skeletal scheme in Figure1 serves only as a starting point for conceptualization of regulons and regulon networks. Gene regulation can occur in unexpected ways. For example, yeast Hac1 can upregulate mRNA expression of a set of target genes involved in UPR, but downregulate protein production of a subset of these genes [50]. It accomplishes this by altering the transcription start site to include a short open reading frame in the upstream of some target genes, so that the resulting mRNA cannot be translated eﬃciently, leading to reduced protein production. A standardized set of conceptual deﬁnitions is need to avoid confusion in regulon research.

3. Discovery and Characterization of New Plant Regulons There are two categories of methods for discovery and characterization of new plant regulons. The ﬁrst is to check the “regulon dictionary” (transcription factor databases) upon ﬁnding a new gene in the same way as we come upon a new word. If the dictionary does not contain the new word, then we make use of the second approach of inferring the meaning of the word based on the context and our grammatical knowledge. This is the approach of de novo discovery of transcription factors based on their shared features, but not their presence in the dictionary. If the new word (gene) turns out to be a transcription factor, then it goes to enrich the regulon dictionary.

3.1. Regulon Discovery by Checking the Regulon Dictionary There are now many databases of characterized transcription factors, from the early TRANSFAC [51] to more recent JASPAR [52], CIS-BP [53], Cistrome DB [54], ChIPBase [55], GTRD [56], the last three all being based on ChIP-Seq data. Plant-specific databases of transcription factors and their TFBS have also become available recently [57,58]. Unfortunately, but as expected, an overwhelming majority of transcription factors in plants are from Arabidopsis thaliana. The rest are mostly from crop species. Thus, the field of tree regulon is essentially a virgin land. Note that these databases do not feature an exhaustive compilation of known TFBSs. They may include TFBS from some large-scale studies, but ignore TFBS detected and verified through conventional painstaking experiments. For example, they include one TFBS of yeast Hac1p [38,39,45,46] but not the other (GACAGCGTGTC) that was well-characterized earlier by conventional experiments [35,39,43,44]. Such incompleteness could be misleading. If a new researcher identifies GACAGCGTGTC as a TFBS for Hac1p, he may search the databases for Hac1p TFBS. Upon finding no such an entry, he may be misled to think that he has identified a TFBS new to science. In spite of the incompleteness, the regulon databases are useful for two main purposes. The first is to check if one’s gene is homologous to a transcription factor already characterized in the database. The second, which should have been more commonly employed, is to retrieve the compiled binding sites to scan new sequences for binding sites [59–61], which I illustrate in detail below. Almost all these databases output experimentally characterized regulatory motifs of specific transcription factors in the form of a matrix. The one shown in Figure2A is for AtHB1 (GenBank locus_tag AT3G01470) as retrieved from PlantPAN [57]. The first three and the last two sites are less informative than sites 4 to 12 (Figure2B) which may be simplified as a palindromic CAATYATTG. From these nine highly informative sites, one can obtain a position weight matrix or PWM (Figure2C) with entries specified as Equation (1) ! pij PWMij = log2 (1) pi where pij is the frequency of nucleotide i (i = A, C, G, or T) at site j, sometimes with pseudocounts to avoid taking logarithm of zero, and pi is background frequencies. The pros and cons of various pseudocount schemes, and of different background frequencies, for PWM have been reviewed and numerically illustrated in detail [62,63]. Genes 2020, 11, 995 6 of 17

The resulting PWM (Figure2C) can now be used to scan new sequences for candidate TFBS. This is done with a sliding window of nine sites (same as the number of sites in PWM) by computing a window-speciﬁc PWM score (PWMS). A 9mer with a high PWMS implies a greater likelihood of being a TFBS. PWM allows us to obtain PWMS of a 9mer by summing up relevant PWM entries. For example, if theGenes nine 2020 sites, 11, x in FOR a window PEER REVIEW happen to be CAATCATTG, then its PWMS is simply the summation6 of 17 of the bolded numbers in Figure2C. This would result in a PWMS of 14.30496, the highest PWMS we can obtainexample, given if the the nine PWM sites (Figure in a window2C). Suppose happen weto be use CAATCATTG, this PWM to then scan its the PWMS 2000 is nt simply upstream the of a summation of the bolded numbers in Figure 2C. This would result in a PWMS of 14.30496, the highest micro-RNA gene miR164a. Ideally, the background frequencies, i.e., pi in Equation (1), should be the PWMS we can obtain given the PWM (Figure 2C). Suppose we use this PWM to scan the 2000 nt nucleotide frequencies of these 2000 nt upstream of miR164a (which are 0.3005, 0.1795, 0.1811, 0.3389 upstream of a micro-RNA gene miR164a. Ideally, the background frequencies, i.e., pi in Equation (1), for A, C, G, and T, respectively). These frequencies were in fact used to compute the PWM (Figure2C). should be the nucleotide frequencies of these 2000 nt upstream of miR164a (which are 0.3005, 0.1795, The highest0.1811, 0.3389 PWMS for ( =A,10.2048) C, G, and was T, foundrespectively). at site Th 343ese (with frequenciesmiR164a werestarting in fact atused site to 2001), compute with the a 9mer CAATCATTA.PWM (Figure This 2C). 9mer The highest matches PWMS eight (=10.2048) out nine was sites found in the at consensussite 343 (with (Figure miR164a2B) starting and its at PWMS site is much2001), higher with than a 9mer those CAATCATTA. upstream or This downstream 9mer matches from eight this outsite nine (Figure sites in 2theD). consensus The expected (Figure PWMS 2B) is zeroand for sequencesits PWMS is assembled much higher randomly than those from upstream the background or downstream frequencies. from this A site PWMS (Figure of 10.2048 2D). The means that theexpected hypothesis PWMS of is azero site-speciﬁc for sequences pattern assembled is 1180 timesrandomly as likely from asthe the background hypothesis frequencies. of no site pattern,A i.e., CAATCATTAPWMS of 10.2048 is means highly that likely the ahypothesis TFBS for of AtHB1. a site-specific A detailed pattern experimental is 1180 times as study likely [64 as] the showed the 9merhypothesis to be indeedof no site a functional pattern, i.e., TFBS CAATCATTA for AtHB1 is to highly regulate likely the a expression TFBS for AtHB1. of miR164a A detailedexpression. experimental study [64] showed the 9mer to be indeed a functional TFBS for AtHB1 to regulate the Through such a combination of bioinformatics and experimental studies, the last two features of a expression of miR164a expression. Through such a combination of bioinformatics and experimental regulonstudies, (i.e., the the last regulated two features target genesof a regulon and the (i.e key., the regulon regulated function) target may genes be and elucidated. the key Aregulon number of programsfunction) can may scan be sequences elucidated. for A putativenumber of TFBSs programs based can on scan an inputsequences PWM, for andputative the performanceTFBSs based on of four suchan programs input PWM, has and recently the performance been evaluated of four [65 such]. programs has recently been evaluated [65].

1234567891011121314 The pattern is strong in A71432225002500328 sites 4 to 12, with C354131005000181 consensus CAATYATTG (B) G337400000001243 T41043 1 0 25 20 0 25 25 0 11 (A) (C) Site123456789 A -1.4112 1.3451 1.5272 -4.8684 -4.8684 1.5272 -4.8684 -4.8684 -1.4112 C 2.4978 -0.9842 -3.7003 -3.7003 1.1504 -3.7003 -3.7003 -3.7003 -0.9842 G 1.1463 -3.3827 -3.3827 -3.3827 -3.3827 -3.3827 -3.3827 -3.3827 2.6890 U -1.7346 -3.0529 -5.0350 1.1894 0.8722 -5.0350 1.1894 1.1894 -5.0350

...... CAATCATTA...... 10 (D) 0

-10

PWMS -20

-30

-40 323 333 343 353 363 Site

FigureFigure 2. Construction2. Construction andand use use of of position position weight weight matrix matrix (PWM) (PWM) to discover to discover new transcription new transcription factor factorbinding binding sites sites (TFBS). (TFBS). (A) ( ASite-specific) Site-specific compilation compilation of experimentally of experimentally validated validated TFBS TFBSfor AtHB1, for AtHB1, retrievedretrieved from from PlantPAN PlantPAN [57 ].[57]. (B )(B The) The consensus consensus sequence sequence from from ninenine stronglystrongly informative sites sites in in (A). (C) PWM(A). (C obtained) PWM obtained from data from in (Adata). ( Din) PWM(A). (D score) PWM (PWMS) score (PWMS) for 9mers for along 9mers the along 2000 the nt 2000 upstream nt of micro-RNAupstream gene of micro-RNAmiR164a which gene miR164a starts at which site 2001. starts Theat site expected 2001. Th PWMSe expected for randomPWMS for sequences random is 0. The 9mersequences with is the 0. The highest 9mer PWMSwith the ishighest CAATCATTA PWMS is CAATCATTA (at Site 343) which (at Site was 343) verifiedwhich was to verified be an e fftoective TFBSbe for an AtHB1. effective PWM TFBS for computation AtHB1. PWM and computation sequence scanning and sequence were scanning done with were software done with DAMBE software [66 ,67]. DAMBE [66,67].

Genes 2020, 11, 995 7 of 17

The characterized binding sites in transcription databases may also help us determine the presence or absence of direct regulator–regulatee relationship. For example, AtHB1 is upregulated in hypocotyls and roots of seedlings in response to short day length, and its expression was claimed to be regulated by PIF1/PIL5 [68]. The binding site of PIL5 (locus_tag AT2G20180) is the G-box transcriptional regulatory code (CACGTG) which is shared among many different plant transcription factors [69]. However, there is no CACGTG in the 2000 nt upstream of the AtHB1 gene. The closest 6mer one can get is CTCGTG at site 781 (AtHB1 start codon at site 2001) and CACGAG at site 1270. Scanning an additional 1000 nt further upstream does not find better matching. Because TFBS is rarely located more than 3000 nt upstream of a coding sequence, PIL5 is unlikely to be a direct regulator for AtHB1 expression. If the regulation is indirect through another transcription factor, then PIL5 and AtHB1 may belong to the same regulon network but not the same regulon according to the definition laid out in the previous section. This highlights the importance of using well-characterized regulons to build regulon networks.

3.2. De Novo Regulon Discovery

3.2.1. De Novo Discovery of TF When we encounter a new word, cannot find it in dictionaries, and wish to judge if it is a verb, we have to use our grammatical knowledge. Transcription factors, like verbs, also share certain features that allow them to be identified and classified [70]. All transcription factors feature a nuclear localization signal [71], a specific structure, such as a leucine zipper [72], for binding to TFBS, and a positively charged DNA-binding domain to facilitate favorable electrostatic interaction with the negatively charged DNA backbone. The nuclear localization signal (NLS) is typically rich in basic amino acids Lys and Arg. The first NLS is PKKKRKV [71]. The NLS is 29RKRAKTK35 in the UPR master transcription factor Hac1 in yeast [73], 75RRKLKNRVAAQTARDRKK92 in XBP1 in human [74], and 645RKRGSRGGKKGRK657 yeast Ire1 [75]. The minimum consensus in mammalian NLS is KR/KxR/K[76]. Transcription factors fall into several major structural classes such as leucine zipper, helix-loop-helix, helix-turn-helix, and many others [70]. Leucine zippers feature their characteristic heptad repeats (Figure3) with repeated 7mers represented as (abcdefg) n. Positions a and d are hydrophobic. In leucine zipper transcription factors such as GCN4 in yeast [77], and XBP1 in human [74], the fourth position in each heptad (position d) is occupied by leucine (Figure3A,B). Heptad repeats are relatively poor in glycine (which would permit too much bending flexibility). They form helices, contain no helix-breaking proline and few clustered charged residues. Hydrophobic residues at positions a and d are on the same side of the helix (Figure3C and D) and form a hydrophobic interface in a homodimer. Plants also have a Hac1 functional homologue, e.g., bZIP60 in Arabidopsis thaliana [23,24] with a leucine zipper structure. It is significant to note that yeast Hac1p [38,39,45,46] mammalian Xbp1 [78,79] and plant bZIP60 [80] all have a TFBS with a core ACGT, although little sequence homology exists among the three. For example, there is only weak homology in a short stretch of amino acid sequences between Hac1p and Xbp1 corresponding to sites 29–53 in the yeast Hac1p [28]. This short stretch is the nuclear localization signal in Hac1p [73]. In addition to the basic NLS, transcription factors typically feature a basic DNA-binding domain favoring electrostatic interaction with the negatively charged DNA backbone. The domain could be in the middle as in AtHB1 (Figure4) or at the N-terminal as in Hac1 and XBP1 from yeast and human, respectively (Figure4). The last two are functional homologues serving as master regulator of UPR, but share no detectable sequence homology at either nucleotide or amino acid level. While none of the features above can identify a protein as a TF, the presence of all these features jointly does increase the likelihood of a protein being a TF. If a protein not only has all these features, but its expression in the cell also consistently results in altered expression of a set of genes, then we can conclude with reasonable confidence that the protein is indeed a transcription factor. Unfortunately, there is no software at present that integrates all available information in de novo discovery of TFs. Genes 2020, 11, x FOR PEER REVIEW 8 of 17

GCN4 XBP1 250 95 MKQLEDK MSELEQQ VEELLSK VVDLEEE Genes 2020, 11, 995 (D) 8 of 17 Genes 2020, 11, x FOR PEER REVIEWNYHLENE NQKLLLE 8 of 17 VARLKKL NQLLREK a b GCN4abcdefg XBP1THGLVVE NQELRQR c 250277 95 d abcdefg e MKQLEDK MSELEQQ

hydrophobic f

136 g hydrophilic VEELLSK VVDLEEE a (A) N(B)QKLLLE (D)(C) b NYHLENE VARLKKL NQLLREK a Figure 3. Examples ofabcdefg leucine zipper TtranscriptionHGLVVE factors with heptadb repeats, with the seven amino c acids in each heptad277 labelled as abcdefg.NQE HydrophoLRQR bic residues at positionsd a and d are shown in bold (A) GCN4 (sites 250 to 277) in yeast Saccharomycesabcdefg cerevisiae. (B) XBP1e (sites 95 to 136) in human. (C)

Relative position of the seven amino acids in an α helix. (D) Top viewhydrophobic of af leucine zipper homodimer

136 g hydrophilic a contacting at their hydrophobic(A) side. (B) (C) b

In addition to the basic NLS, transcription factors typically feature a basic DNA-binding domain Figure 3. Examples of leucine zipper transcription factorsfactors with heptad repeats, with the seven amino favoring electrostatic interaction with the negatively charged DNA backbone. The domain could be acids in each each heptad heptad labelled labelled as as abcdefg. abcdefg. Hydropho Hydrophobicbic residues residues at positions at positions a and a andd are d shown are shown in bold in in the middle as in AtHB1 (Figure 4) or at the N-terminal as in Hac1 and XBP1 from yeast and human, bold(A) GCN4 (A) GCN4 (sites (sites 250 to 250 277) to 277)in yeast in yeast SaccharomycesSaccharomyces cerevisiae. cerevisiae. (B) XBP1(B) XBP1 (sites (sites 95 to 95 136) to 136)in human. in human. (C) respectively (Figure 4). The last two are functional homologues serving as master regulator of UPR, (RelativeC) Relative position position of the of the seven seven amino amino acids acids in inan an α αhelix.helix. (D (D) Top) Top view view of of a a leucine leucine zipper zipper homodimer homodimer but share no detectable sequence homology at either nucleotide or amino acid level. contacting at their hydrophobic side.side.

11 12 13 In addition to the basicAtHB1 NLS, transcription factors typically Hac1 feature a basic DNA-binding XBP1 domain 12 favoring10 electrostatic interaction with the11 negatively charged DNA backbone. The domain could be in the middle as in AtHB1 (Figure 4) or at the N-terminal as in Hac1 and XBP1 from yeast and human, 11 10 respectively9 (Figure 4). The last two are functional homologues serving as master regulator of UPR, 10 but share no detectable sequence homology9 at either nucleotide or amino acid level. 8 9 11 128 13 7 AtHB1 Hac18 XBP1 12 10 117 7 Sliding window pI Sliding window pI 6 Sliding window pI 11 10 9 6 6 10 5 95 8 5 9 4 8 4 4 7 8 3 73 3 7 Sliding window pI Sliding window pI 6 0 50 100 150 200 0 50 100 150 Sliding window pI 0 100 200 300 mid-window site 6 mid-window site mid-window site 6 5 Figure 4. Protein isoelectric point (pI) profile5 showing different transcription factors having a positively Figure 4. Protein isoelectric point (pI) profile showing different transcription5 factors having a chargedpositively DNA-binding charged DNA-binding domain, with AtHB1domain, from withArabidopsis AtHB1 thaliana,from ArabidopsisHac1 from Saccharomycesthaliana, Hac1 cerevisiae from 4 4 andSaccharomyces XBP1 from cerevisiae human and (a functional XBP1 from homologue human (a functional of Hac1). Window-specifichomologue of4 Hac1).pI was Window-specific computed with pI software DAMBE [66,67] with window size of 80 and step size of 5. was3 computed with software DAMBE [66,67]3 with window size of 80 and3 step size of 5. 3.2.2. De0 Novo 50Discovery 100 of 150 TFBS 200 0 50 100 150 0 100 200 300 mid-window site mid-window site mid-window site While none of the features above can identify a protein as a TF, the presence of all these features jointlyThere does areincrease computational the likelihood methods of a protein for being predicting a TF. If DNA-binding a protein not only sites has from all these 3D features, protein Figure 4. Protein isoelectric point (pI) profile showing different transcription factors having a structurebut its expression [81–85] or in even the cell from also plain consistently amino acid results sequences in altered [86], expression especially of from a set sequence of genes, similarity then we positively charged DNA-binding domain, with AtHB1 from Arabidopsis thaliana, Hac1 from ofcan DNA- conclude binding with domains reasonable [53], although confidence the reliabilitythat the ofprotein these approaches,is indeed especiallya transcription the latter factor. one Saccharomyces cerevisiae and XBP1 from human (a functional homologue of Hac1). Window-specific pI withoutUnfortunately, structural there information, is no software is questionable. at present that A statistical integrates trend all available often does information not imply accuracyin de novo in was computed with software DAMBE [66,67] with window size of 80 and step size of 5. specificdiscovery predictions. of TFs. For example, Weirauch et al. [53] likely have overgeneralized the observation that “closely related DBDs almost always have very similar DNA sequence preferences” to the conclusion While none of the features above can identify a protein as a TF, the presence of all these features that “in general, sequence preferences can be accurately inferred by overall DBD AA identity”. I do jointly does increase the likelihood of a protein being a TF. If a protein not only has all these features, not disagree with the observation and can list examples more impressive than they did. For example, but its expression in the cell also consistently results in altered expression of a set of genes, then we TFs such as Hac1p in yeast [38,39,45,46], Xbp1 in mammals [78,79] and BZIP60 in plants [80] all have a can conclude with reasonable confidence that the protein is indeed a transcription factor. leucine zipper DNA-binding structure, and their TFBS all feature the core ACGT. However, I disagree Unfortunately, there is no software at present that integrates all available information in de novo discovery of TFs.

Genes 2020, 11, x FOR PEER REVIEW 9 of 17

3.2.2. De Novo Discovery of TFBS There are computational methods for predicting DNA-binding sites from 3D protein structure [81–85] or even from plain amino acid sequences [86], especially from sequence similarity of DNA- binding domains [53], although the reliability of these approaches, especially the latter one without structural information, is questionable. A statistical trend often does not imply accuracy in specific predictions. For example, Weirauch et al. [53] likely have overgeneralized the observation that “closely related DBDs almost always have very similar DNA sequence preferences” to the conclusion Genesthat2020 “in, 11 general,, 995 sequence preferences can be accurately inferred by overall DBD AA identity”. I9 do of 17 not disagree with the observation and can list examples more impressive than they did. For example, thatTFs such such observations, as Hac1p in yeast remarkable [38,39,45,46], as they Xbp1 are, in implymammals much [78,79] power and in BZIP60 making in plants specific [80] predictions. all have a leucine zipper DNA-binding structure, and their TFBS all feature the core ACGT. However, I If I states that “Because transcription factor X has a DBD nearly identical to that of transcription disagree that such observations, remarkable as they are, imply much power in making specific factor Y, I predict that transcription factor X has nearly identical DNA sequence preference as that of predictions. If I states that “Because transcription factor X has a DBD nearly identical to that of transcription factor Y”, few experimental biologists would take my prediction seriously and act upon it. transcription factor Y, I predict that transcription factor X has nearly identical DNA sequence In other words, few people would agree with the assertion that “in general, sequence preferences can preference as that of transcription factor Y”, few experimental biologists would take my prediction be accuratelyseriously and inferred act upon by it. overall In other DBD words, AA few identity”, people aswould one couldagree with list a the number assertion of counterthat “in general, examples in whichsequence sequence preferences preferences can be accurately cannot be inferred accurately by inferredoverall DBD by overall AA identity”, DBD AA as identity.one could In list short, a computationalnumber of counter predictions examples of TFBS,in which and sequence the resulting preferences databases cannot such be accurately as CIS-BP, inferred could beby valuable,overall butDBD one shouldAA identity. be aware In short, of their computational limitations predicti in makingons specificof TFBS, predictions. and the resulting databases such as CIS-BP,Experimental could be approaches valuable, but for one large-scale should be de aware novo of discovery their limitations of TFBSs in includemaking microarray,specific predictions. ChIP- Seq and itsExperimental derivatives such approaches as ChIP-exo for large-scale [17,18], ChEC-seqde novo discovery [19] and of Cut&Run TFBSs include [20]. microarray, All these methods ChIP- fragmentSeq and DNA, its derivatives either by such sonication, as ChIP-exo by Micrococcal[17,18], ChEC-seq nuclease [19] and as in Cut&Ru ChEC-seqn [20]. and All Cut&Run,these methods or by lambdafragment exonuclease DNA, either as in by ChIP-exo sonication, (Figure by Micrococcal5). The microarray nuclease approachas in ChEC-seq will hybridize and Cut&Run, the extracted or by DNAlambda fragments exonuclease to a microarray. as in ChIP-exo Because (Figure probe 5). The sequences microarray on the approach microarray will hybridize are known, the binding extracted to a probeDNA means fragments that the to a probe microarray. sequence Because either probe contains sequences a candidate on the TFBSmicroarray or is very are known, close to binding it along to the genome.a probe Such means information that the probe can be sequence used to either produce contains a position-specific a candidate TFBS affinity or is matrix very close (PSAM) to it by along using softwarethe genome. such as Such MatrixReduce information [ 87can]. be This used PSAM to produc can thene a position-specific be used to scan newaffinity sequences, matrix (PSAM) in the sameby wayusing as PWM, software for candidatesuch as MatrixReduce TFBS. [87]. This PSAM can then be used to scan new sequences, in the same way as PWM, for candidate TFBS.

(A) gene1 gene2 gene3 gene4 gene5 gene6 gene7

Cross-link and sonication (B)

binding by antibody (C)

isolation by protein A/G-coupled agarose beads (D)

Reversal of cross-linking (E)

Sequencing (F) ..AGTACGTGTCC.. ..TCTACGTGTGG.. ..GCTACGTGTTC.. ..GATACGTGTCT.. ..AGTACGTGTCC.. ..TCTACGTGTGG.. ..GATACGTGTCT.. Figure 5. Illustration of ChIP-Seq method for highlighting its problems and improvements. (A) DNA Figure 5. Illustration of ChIP-Seq method for highlighting its problems and improvements. (A) DNA genome with gene1, gene2, gene 4, and gene7 sharing the same TFBS bound to the same TF of genome with gene1, gene2, gene 4, and gene7 sharing the same TFBS bound to the same TF of interest interest (red-colored squares). (B) Cross-linking fixes TF onto TFBS, and sonication is optimized (red-colored squares). (B) Cross-linking fixes TF onto TFBS, and sonication is optimized to minimize to minimize the sequences flanking TFBS. (C) A specific anti-TF antibody is used to bind to the TF of interest. (D) A general-purpose protein A/G coupled to agarose beads is used to pull down the TFBS- TF- antibody complex. (E) Reversal of cross-linking frees TFBS-containing fragments for sequencing. (F) High-throughput sequence generates millions or even billions of sequence reads that may contain a TFBS.

the sequences flanking TFBS. (C) A specific anti-TF antibody is used to bind to the TF of interest. (D) A general-purpose protein A/G coupled to agarose beads is used to pull down the TFBS- TF- antibody complex. (E) Reversal of cross-linking frees TFBS-containing fragments for sequencing. (F) High- throughput sequence generates millions or even billions of sequence reads that may contain a TFBS.

The ChIP-Seq approach will sequence all sequence fragments (Figure 5F), each of which may contain no or multiple TFBSs. Because TFBS length is typically in the order of 10, the sequence fragments from ChIP-Seq ideally should be just the plain TFBS without flanking sequences. However, Genes 2020, 11, 995 10 of 17 ChIP-seq with sonication often generates sequences of several hundred bases. A common approach is to sequence both ends, knowing that the putative TFBS is located between the two ends (Figure two6A). endsIf the overlap, two ends then overlap, the putative then the TFBS putative is included TFBS is in included the sequence. in the If sequence. the two ends If the do two not ends overlap do (e.g.,not overlap when the(e.g., fragment when the is fragment 500 base andis 500 the base paired-end and the readpaired-end length read is 100 length bases), is then100 bases), potentially then thepotentially two paired the two ends paired can be ends mapped can be onto mapped the genome onto the to genome obtain theto obtain complete the complete sequence sequence in between. in Figurebetween.6B showsFigure two 6B paired-endshows two reads paired-end from a Cut&Run reads from experiment a Cut&Run on Mus experiment musculus (BioProjecton Mus musculus ACCN PRJNA544746)(BioProject ACCN that PRJNA544746) do not overlap. that Bydo mappingnot overlap. them By ontomapping mouse them chromosome onto mouse 2 chromosome (NC_000068), 2 we(NC_000068), learn that theywe learn are separatedthat they are by justseparated a single by nucleotide just a single C nucleotide (Figure6C). C However, (Figure 6C). if we However, consider if TFBSwe consider as signal TFBS and as the signal flanking and sequences the flanking as noise, sequences then longas noise, flanking then sequences long flanking are associated sequences with are lowassociated signal/ withnoise low ratio. signal/noise TFBS detection ratio. isTFBS more detect sensitiveion iswith more short sensitive flanking with sequences short flanking as in sequences Figure6C. Oneas in ofFigure the advantages 6C. One of ofthe ChIP-exo advantages [17 ,of18 ]ChIP-e ChEC-seqxo [17,18] [19] andChEC-seq Cut&Run [19] [and20] isCut&Run the use of[20] MNase is the oruse exonuclease of MNase or which exonuclease cuts unbound which DNAcuts unboun (e.g., thed linkerDNA (e.g., region the nucleosomes) linker region and nucleosomes) digests the freeand endsdigests until the itfree encounters ends until the it encounters bound (protected) the bound region. (protected) The fragment region. The length fragment in Figure length6D in is Figure likely the6D is minimum likely the achievable minimum throughachievable sonication. through sonication. Most sequence Most reads sequence by ChIP-seq reads by methodsChIP-seq are methods much longerare much than longer that than in Figure that in6D. Figure For example,6D. For example, in an e ffinort an [effort3] to obtain[3] to obtain TFBS TFBS in two in paralogoustwo paralogous TFs (MdMYB88TFs (MdMYB88 and MdMYB124)and MdMYB124) in apple in (appleMalus ( domesticaMalus domestica), sonication), sonication was used was to generateused to sequencegenerate fragmentssequence fragments of 300–700 of bp, 300 but−700 the bp, single-read but the single-read has a length has ofa length only 50 of bases. only 50 This bases. implies This that,implies if TFBS that, isif inTFBS the is middle in the ofmiddle each sequenceof each sequence fragment, fragment, then the readthen ofthe 50 read bases of from50 bases the sequencefrom the fragmentsequence willfragment not include will not TFBS, include so the TFBS, signal /sonoise the ratio signal/noise is 0. However, ratio withis 0. MNaseHowever, or otherwith exonucleases,MNase or other it is theoreticallyexonucleases, possible it is theoretically to generate possible TFBS-containing to generate sequence TFBS-containing fragments ofsequence only 50 fragments bases or even of only shorter. 50 bases or even shorter. (B) 10 20 30 40 50 60 70 80 ...... CTCTACGTGTGGC...... ----|----|----|----|----|----|----|----|----|----|----|----|----|----|----|----| (A) SRR11577050.21.1 ------AAA SRR11577050.21.2 GGAACTGGCAGCCGAGAAACAGCTTCAGACCAAGCCCCCACTAGTACCTACTACCAAAAGTGACAAGAATGACTGG---- NC_000068 GGAACTGGCAGCCGAGAAACAGCTTCAGACCAAGCCCCCACTAGTACCTACTACCAAAAGTGACAAGAATGACTGGCAAA MNase cuts yield a 153nt DNA fragment 90 100 110 120 130 140 150 with 76 nt sequenced at each end . ----|----|----|----|----|----|----|----|----|----|----|----|----|----|--- SRR11577050.21.1 GCCTGTGCCCCTTTCTCACCCCTTTCTAAGAAGCATAGACCAGGGCCTAGGGGGAGTTTGGACACAGGAACCT SRR11577050.21.2 ------MNase cuts yield a 79nt DNA fragment with NC_000068 GCCTGTGCCCCTTTCTCACCCCTTTCTAAGAAGCATAGACCAGGGCCTAGGGGGAGTTTGGACACAGGAACCT 76 nt sequenced at each end . (C) 10 20 30 40 50 60 70 ----|----|----|----|----|----|----|----|----|----|----|----|----|----|----|---- SRR11577050.5001.1 ---CCTCAGCCTGACCATAGGCCTCTGACATCTGACCACAGGTCACTGTTCCAGCCATCTATCTATTGTTCCTTCAAGA Sonication yields a 120nt DNA fragment SRR11577050.5001.2 TCTCCTCAGCCTGACCATAGGCCTCTGACATCTGACCACAGGTCACTGTTCCAGCCATCTATCTATTGTTCCTTCA--- with 101 nt sequenced at each end . *************************************************************************

(D) 10 20 30 40 50 60 70 80 90 100 110 ----|----|----|----|----|----|----|----|----|----|----|----|----|----|----|----|----|----|----|----|----|----|----|----| SRR11806558.1.1 CCTCGCACAGGTCGCGGATATGGTCGGTGACGCCGAAAATCTCCTGGACCAGCACGAGCCCGCCCTTGCGTCGCCCGGTCGGCTCGGCGTGATAGACGCCG------SRR11806558.1.2 ------ATGGTCGGTGACGCCGAAAATCTCCTGGACCAGCACGAGCCCGCCCTTGCGTCGCCCGGTCGGCTCGGCGTGATAGACGCCGACCTCCGCACCATCCGACA **********************************************************************************

Figure 6.6.Reducing Reducing background background sequences sequences flanking flanking TFBS. TFBS. (A) Paired-end (A) Paired-end reading ofreading a TFBS- of containing a TFBS- sequencecontaining fragment. sequence (B fragment.) Paired-end (B) reads Paired-end that do reads not overlap, that do from not anoverlap, SRA file from (ACCN: an SRA SRR11577050 file (ACCN: in BioProjectSRR11577050 PRJNA544746) in BioProject generated PRJNA544746) by Cut&Run. generated (C) by Paired-end Cut&Run. reads (C) Paired-end overlap, from reads the sameoverlap, file from as in (theB). (sameD) Paired-end file as in reads(B). ( fromD) Paired-end an SRA file reads (ACCN: from SRR11806558 an SRA file in (ACCN: BioProject SRR11806558 PRJNA633509) in BioProject generated byPRJNA633509) ChIP-seq. The generated forward orby reverse ChIP-seq. reads The is reverse-complemented forward or reverse reads to generate is reverse-complemented the sequence alignment. to generate the sequence alignment. With a large number of sequence reads containing the putative TFBS for a particular TF, there are twoWith categories a large number of bioinformatics of sequence approaches reads containing to extract the TFBS. putative The first, TFBS represented for a particular by MACS TF, there [88], isare to two map categories the reads of ontobioinformatics the genome approaches to identify to peakextract regions TFBS. presumablyThe first, represented corresponding by MACS to TFBS [88], ((Figureis to map7A). the The reads sites onto in TFBS the aregenome expected to identify to be represented peak regions more presumably frequently than corresponding those in the flankingto TFBS sequences((Figure 7A). (Figure The 7sitesB). This in TFBS leads are to anexpected approximate to be represented location ofTFBS. more Iffrequently multiple non-homologousthan those in the genesflanking share sequences the seven (Figure sites in 7B). red (FigureThis leads7A), to but an not approximate the flanking location sequences, of TFBS. then we If maymultiple infer non- that the seven sites constitute a TFBS for the particular TF of interest. ChIP-seq also has the shortcoming of spurious cross-linking and large sample size which is needed for us to see the peak in Figure7. We would not need a large sample of cells (i.e., genomes) if we could trim off all flanking sequences from the TFBS-containing DNA fragments. This is one of the reasons that newer methods such as ChIP-exo [17,18], ChEC-seq [19] and Cut&Run [20] require fewer cells. These newer methods also do not need sonication. Another promising method is DamID-seq [21] Genes 2020, 11, 995 11 of 17 Genes 2020, 11, x FOR PEER REVIEW 11 of 17 whichhomologous only requires genes theshare construction the seven sites of a fusionin red (Figure protein 7A), and but methylation-specific not the flanking sequences, genome sequencing, then we andmay theoretically infer that the could seven work sites with constitute single a cells. TFBS for the particular TF of interest.

Gene1 Gene2 ....CGACTACGACCGATACTCTACGTGTGGCCACCTGCGGACTCGA...... CACGACGTAGATACGTGTTCCGAACTGC.. CTCTACGTGTGGCCACCTGCGGACTCGA CACGACGTAGATACGTGTTCCGAA ACCGATACTCTACGTGTGGCCACCTGC CGTAGATACGTGTTCCGAACTGC CGACTACGACCGATACTCTACGTGTGGCCACCTG (A) GACGTAGATACGTGTTCCGAACTG ACTCTACGTGTGGCCACCTGCG ACGACGTAGATACGTGTTCCGAACTGC CGACCGCCACTCTACGTGTGGCC TAGATACGTGTTCCGAACTGC ......

(B) Count Count Site Site FigureFigure 7. 7.Identifying Identifying TFBS TFBS byby mappingmapping ChIP-SeqChIP-Seq reads reads to to genome. genome. (A (A) Two) Two genes genes on on the the genome genome (colored(colored yellow) yellow) are are not not homologous homologous but but share share a a TFBS TFBS (colored (colored red),red), eacheach withwith manymany reads mapped to them.to them. (B) The(B) The count count of eachof each site site represented represented in in reads, reads, with with eacheach site site in in TFBS TFBS expected expected to tohave have higher higher countscounts than than ﬂanking flanking sequences. sequences.

TheChIP-seq second methodalso has of the extracting shortcoming TFBS of is byspurious Gibbs samplercross-linking which and has beenlarge usedsample in thesize identiﬁcation which is ofneeded functional for us motifs to see in the proteins peak in [89 Figure–91] and7. We regulatory would not sequences need a large in DNAsample [92 of– 100cells]. (i.e., The genomes) algorithm if has beenwe numericallycould trim off illustrated all flanking in sequences detail [62 from,101] the and TFBS-containing implemented in DNA DAMBE fragments. [66,67 This]. It is one particularly of the usefulreasons for that extracting newer methods TFBS when such we as haveChIP-exo few reads[17,18], mapped ChEC-seq to [19] each and gene Cut&Run or when [20] there require are fewfewer cells tocells. start with.These Figurenewer8 methodsA shows also six reads do not each need mapped sonication. to a separateAnother gene.promising We know method that is a DamID-seq motif may be hidden[21] which in these only reads, requires but dothe not construction know what of the a fusion motif looksprotein like and or wheremethylation-specific it is located in genome the reads. sequencing, and theoretically could work with single cells. Gibbs sampler is exactly suitable for identifying this motif and pinpointing its location. One may The second method of extracting TFBS is by Gibbs sampler which has been used in the compile the sequences into a ﬁle (Figure8B) and feed it to Gibbs sampler. The typical output includes identification of functional motifs in proteins [89–91]and regulatory sequences in DNA [92–100]. The a PWM (Figure8C) that one can use to scan the genome for new locations of the TFBS, and a motif that algorithm has been numerically illustrated in detail [62,101] and implemented in DAMBE [66,67]. It is over-represented in the input sequences (Figure8D). With new technologies such as Cut&Run [ 20] is particularly useful for extracting TFBS when we have few reads mapped to each gene or when that is touted to work with few cells, I expect Gibbs sampler to become an essential tool embedded in there are few cells to start with. Figure 8A shows six reads each mapped to a separate gene. We know allthat bioinformatics a motifGenes may2020, 11 software be, x FORhidden PEER packages REVIEW in these usedreads, for but protein-DNA do not know interactions.what the motif looks like or12 ofwhere 17 it is located in the reads. Gibbs sampler is exactly suitable for identifying this motif and pinpointing its location. One5' mayCTCTACGTGTGGCCACCTGCGGACT compile the sequences into(A) a file (Figure 8B)TAGCTACGCGTGCCGAACTGC and feed it to Gibbs sampler. The typical output includes a PWM (Figure 8C) that one can use to scan the genome for new locations of the TFBS, and a motif that is over-represented in the input sequences (Figure 8D). With new AACTGCTAGCTACGCGTTCCG technologies suchCGCCACCTGCGGACTCGATACGTGTCGTC as Cut&Run [20] that is touted...... to work with few cells, I expect Gibbs sampler to become an essential tool embedded in all bioinformatics software packages used for protein-DNA interactions. CACCTGCGGACTCGATATGTGTCGCCGTC 3' CCAACTGCTAGCTGCGCGTTG (C) A C G T 1 -2.93032 -3.04203 -2.73555 2.47865 >S1 (B) 2 2.13108 -3.04203 -0.52324 -2.33338 CTCTACGTGTGGCCACCTGCGGACT 3 -2.93032 0.89597 -2.73555 0.13002 >S2 4 -2.93032 -3.04203 1.77572 -2.33338 TAGCTACGCGTGCCGAACTGC 5 -2.93032 0.22043 -2.73555 1.52912 >S3 6 -2.93032 -3.04203 1.77572 -2.33338 CGCCACCTGCGGACTCGATACGTGTCGTC 7 -2.93032 -3.04203 -2.73555 2.47865 >S4 Gibbs sampler S1 CTCTACGTGTGGCCACCTGCGGACT AACTGCTAGCTACGCGTTCCG S2 TAGCTACGCGTGCCGAACTGC >S5 S3CGCCACCTGCGGACTCGATACGTGTCGTC CCAACTGCTAGCTGCGCGTTG S4 AACTGCTAGCTACGCGTTCCG (D) >S6 S5 CCAACTGCTAGCTGCGCGTTG CACCTGCGGACTCGATATGTGTCGCCGTC S6 CACCTGCGGACTCGATATGTGTCGCCGTC

Figure 8. UsingFigure Gibbs8. Using sampler Gibbs sampler to extract to extract TFBS. TFBS. ( A(A)) Six reads reads mapped mapped eacheach mapped mapped to a different to a digenefferent gene on the genomeon the (represented genome (represented by the by blue the blue line. line. (B (B)) FASTAFASTA file file for for the the six reads six reads as input as to input Gibbs tosampler. Gibbs sampler. (C) An optimal(C) An position optimal position weight weight matrix matrix (PWM) (PWM) that that oneone can can use use to scan to scan the genome the genome for new forlocations new locations of the TFBS. (D) The aligned motif that generates the most informative PWM in (C). (C) and (D) are of the TFBS. (D) The aligned motif that generates the most informative PWM in (C). (C) and (D) are generated from DAMBE [61] based on the FASTA input file in (B). generated from DAMBE [61] based on the FASTA input file in (B). Some researchers (e.g., [53]) thought that ChIP-seq experiments do not inherently measure the

relative motif preference of a TF to individual sequences, but they do just that, as is illustrated in Figure 8. Given the PWM in Figure 8C, the preference for TACGTGT is measured by the PWM score (PWMS) and is equal to 13.06491 (the summation of the bolded numbers). Another 7mer, TACGCGT, resulting from a transition at the 5th site, would have a PWMS of 11.75622. Thus, the relative motif preference of a TF can be readily quantified.

4. Conclusions Regulon research is a rapidly progressing field with many new experimental protocols, computational algorithms, and databases having been developed in recent years. I present conceptual rationales of these protocols and algorithms and highlight their advantages and disadvantages. Existing literature on regulon research, especially in tree species, were critically reviewed to highlight how better results could have been obtained.

Funding: This study is supported by a Discovery Grant from the Natural Science and Engineering Research Council (NSERC, RGPIN/2018-03878) of Canada.

Acknowledgments: I thank members of the Xia Lab for discussion. Three anonymous reviewers provided excellent comments that led to significant improvement of the paper.

Conflicts of Interest: The authors declare no conflict of interest.

References

1. Romero; Fuertes; Benito; Malpica; Leyva; Paz-Ares, J. More than 80 R2R3-MYB regulatory genes in the genome of Arabidopsis thaliana. Plant J. 1998, 14, 273–284, doi:10.1046/j.1365-313x.1998.00113.x. 2. Stracke, R.; Werber, M.; Weisshaar, B. The R2R3-MYB gene family in Arabidopsis thaliana. Curr. Opin. Plant Biol. 2001, 4, 447–456, doi:10.1016/s1369-5266(00)00199-0.

Genes 2020, 11, 995 12 of 17

Some researchers (e.g., [53]) thought that ChIP-seq experiments do not inherently measure the relative motif preference of a TF to individual sequences, but they do just that, as is illustrated in Figure8. Given the PWM in Figure8C, the preference for TACGTGT is measured by the PWM score (PWMS) and is equal to 13.06491 (the summation of the bolded numbers). Another 7mer, TACGCGT, resulting from a transition at the 5th site, would have a PWMS of 11.75622. Thus, the relative motif preference of a TF can be readily quantiﬁed.

4. Conclusions Regulon research is a rapidly progressing ﬁeld with many new experimental protocols, computational algorithms, and databases having been developed in recent years. I present conceptual rationales of these protocols and algorithms and highlight their advantages and disadvantages. Existing literature on regulon research, especially in tree species, were critically reviewed to highlight how better results could have been obtained.

Funding: This study is supported by a Discovery Grant from the Natural Science and Engineering Research Council (NSERC, RGPIN/2018-03878) of Canada. Acknowledgments: I thank members of the Xia Lab for discussion. Three anonymous reviewers provided excellent comments that led to significant improvement of the paper. Conflicts of Interest: The authors declare no conflict of interest.

References

1. Romero, I.; Fuertes, A.; Benito, M.J.; Malpica, J.M.; Leyva, A.; Paz-Ares, J. More than 80 R2R3-MYB regulatory genes in the genome of Arabidopsis thaliana. Plant J. 1998, 14, 273–284. [CrossRef][PubMed] 2. Stracke, R.; Werber, M.; Weisshaar, B. The R2R3-MYB gene family in Arabidopsis thaliana. Curr. Opin. Plant Biol. 2001, 4, 447–456. [CrossRef] 3. Xie, Y.; Chen, P.; Yan, Y.; Bao, C.; Li, X.; Wang, L.; Shen, X.; Li, H.; Liu, X.; Niu, C.; et al. An atypical R2R3 MYB transcription factor increases cold hardiness by CBF-dependent and CBF-independent pathways in apple. New Phytol. 2017, 218, 201–218. [CrossRef] 4. Maas, W.K.; Clark, A. Studies on the mechanism of repression of arginine biosynthesis in Escherichia coli. J. Mol. Biol. 1964, 8, 365–370. [CrossRef] 5. Koornneef, M.; Hanhart, C.J.; Van Der Veen, J.H. A genetic and physiological analysis of late flowering mutants in Arabidopsis thaliana. Mol. Genet. Genom. 1991, 229, 57–66. [CrossRef][PubMed] 6. Böhlenius, H.; Huang, T.; Charbonnel-Campaa, L.; Brunner, A.M.; Jansson, S.; Strauss, S.H.; Nilsson, O. CO/FT Regulatory Module Controls Timing of Flowering and Seasonal Growth Cessation in Trees. Science 2006, 312, 1040–1043. [CrossRef] 7. Mizoi, J.; Shinozaki, K.; Yamaguchi-Shinozaki, K. AP2/ERF family transcription factors in plant abiotic stress responses. Biochim. Biophys. Acta 2012, 1819, 86–96. [CrossRef] 8. Novillo, F.; Medina, J.; Salinas, J. Arabidopsis CBF1 and CBF3 have a different function than CBF2 in cold acclimation and define different gene classes in the CBF regulon. Proc. Natl. Acad. Sci. USA 2007, 104, 21002–21007. [CrossRef] 9. Fowler, S.G.; Cook, D.; Thomashow, M.F. Low Temperature Induction of Arabidopsis CBF1, 2, and 3 Is Gated by the Circadian Clock1. Plant Physiol. 2005, 137, 961–968. [CrossRef] 10. Harmer, S.L.; HogenEsch, J.B.; Straume, M.; Chang, H.-S.; Han, B.; Zhu, T.; Wang, X.; Kreps, J.A.; Kay, S.A. Orchestrated Transcription of Key Pathways in Arabidopsis by the Circadian Clock. Science 2000, 290, 2110–2113. [CrossRef] 11. Artlip, T.S.; Wisniewski, M.; Bassett, C.L.; Norelli, J.L. CBF gene expression in peach leaf and bark tissues is gated by a circadian clock. Tree Physiol. 2013, 33, 866–877. [CrossRef][PubMed] 12. Ding, J.; Böhlenius, H.; Rühl, M.G.; Chen, P.; Sane, S.; Zambrano, J.A.; Zheng, B.; Eriksson, M.E.; Nilsson, O. GIGANTEA- like genes control seasonal growth cessation in Populus. New Phytol. 2018, 218, 1491–1503. [CrossRef][PubMed] Genes 2020, 11, 995 13 of 17

13. Jia, J.; Zhou, J.; Shi, W.; Cao, X.; Luo, J.; Polle, A.; Luo, Z.B. Comparative transcriptomic analysis reveals the roles of overlapping heat-/drought-responsive genes in poplars exposed to high temperature and drought. Sci. Rep. 2017, 7, 43215. [CrossRef][PubMed] 14. Pucholt, P.; Sjödin, P.; Weih, M.; Rönnberg-Wästljung, A.-C.; Berlin, S. Genome-wide transcriptional and physiological responses to drought stress in leaves and roots of two willow genotypes. BMC Plant Biol. 2015, 15, 244. [CrossRef] 15. Usadel, B.; Kuschinsky, A.; Steinhauser, D.; Pauly, M. Transcriptional co-response analysis as a tool to identify new components of the wall biosynthetic machinery. Plant Biosyst. Int. J. Deal. All Asp. Plant Biol. 2005, 139, 69–73. [CrossRef] 16. Persson, S.; Wei, H.; Milne, J.; Page, G.P.; Somerville, C.R. Identification of genes required for cellulose synthesis by regression analysis of public microarray data sets. Proc. Natl. Acad. Sci. USA 2005, 102, 8633–8638. [CrossRef] 17. Rhee, H.S.; Pugh, B.F. Comprehensive Genome-wide Protein-DNA Interactions Detected at Single-Nucleotide Resolution. Cell 2011, 147, 1408–1419. [CrossRef] 18. He, Q.; Johnston, J.; Zeitlinger, J. ChIP-nexus enables improved detection of in vivo transcription factor binding footprints. Nat. Biotechnol. 2015, 33, 395–401. [CrossRef] 19. Zentner, G.E.; Kasinathan, S.; Xin, B.; Rohs, R.; Henikoff, S. ChEC-seq kinetics discriminates transcription factor binding sites by DNA sequence and shape in vivo. Nat. Commun. 2015, 6, 8733. [CrossRef] 20. Skene, P.J.; Henikoff, S. An efficient targeted nuclease strategy for high-resolution mapping of DNA binding sites. eLife 2017, 6, 576. [CrossRef] 21. Wu, F.; Olson, B.G.; Yao, J. DamID-seq: Genome-wide Mapping of Protein-DNA Interactions by High Throughput Sequencing of Adenine-methylated DNA Fragments. J. Vis. Exp. 2016, 53620. [CrossRef] [PubMed] 22. Chapal, M.; Mintzer, S.; Brodsky, S.; Carmi, M.; Barkai, N. Resolving noise-control conflict by gene duplication. PLoS Biol. 2019, 17, e3000289. [CrossRef][PubMed] 23. Deng, Y.; Humbert, S.; Liu, J.-X.; Srivastava, R.; Rothstein, S.J.; Howell, S.H. Heat induces the splicing by IRE1 of a mRNA encoding a transcription factor involved in the unfolded protein response in Arabidopsis. Proc. Natl. Acad. Sci. USA 2011, 108, 7247–7252. [CrossRef][PubMed] 24. Zhang, L.; Chen, H.; Brandizzi, F.; Verchot, J.; Wang, A. The UPR Branch IRE1-bZIP60 in Plants Plays an Essential Role in Viral Infection and Is Complementary to the Only UPR Pathway in Yeast. PLoS Genet. 2015, 11, e1005164. [CrossRef] 25. O’Malley, R.C.; Huang, S.C.; Song, L.; Lewsey, M.G.; Bartlett, A.; Nery, J.R.; Galli, M.; Gallavotti, A.; Ecker, J.R. Cistrome and Epicistrome Features Shape the Regulatory DNA Landscape. Cell 2016, 165, 1280–1292. [CrossRef] 26. Chapman, R.; Sidrauski, C.; Walter, P. INTRACELLULAR SIGNALING FROM THE ENDOPLASMIC RETICULUM TO THE NUCLEUS. Annu. Rev. Cell Dev. Biol. 1998, 14, 459–485. [CrossRef] 27. Chapman, R.E.; Walter, P. Translational attenuation mediated by an mRNA intron. Curr. Biol. 1997, 7, 850–859. [CrossRef] 28. Xia, X. Translation Control of HAC1 by Regulation of Splicing in Saccharomyces cerevisiae. Int. J. Mol. Sci. 2019, 20, 2860. [CrossRef] 29. Ogawa, N.; Mori, K. Autoregulation of the HAC1 gene is required for sustained activation of the yeast unfolded protein response. Genes Cells 2004, 9, 95–104. [CrossRef] 30. Agarwal, M.; Hao, Y.; Kapoor, A.; Dong, C.-H.; Fujii, H.; Zheng, X.; Zhu, J.-K. A R2R3 Type MYB Transcription Factor Is Involved in the Cold Regulation of CBF Genes and in Acquired Freezing Tolerance. J. Biol. Chem. 2006, 281, 37636–37645. [CrossRef] 31. Chinnusamy, V.; Zhu, J.; Zhu, J.-K. Cold stress regulation of gene expression in plants. Trends Plant Sci. 2007, 12, 444–451. [CrossRef][PubMed] 32. Miura, K.; Jin, J.B.; Lee, J.; Yoo, C.Y.; Stirm, V.; Miura, T.; Ashworth, E.N.; Bressan, R.A.; Yun, D.J.; Hasegawa, P.M. SIZ1-mediated sumoylation of ICE1 controls CBF3/DREB1A expression and freezing tolerance in Arabidopsis. Plant Cell 2007, 19, 1403–1414. [CrossRef][PubMed] 33. Mori, K. The unfolded protein response: The dawn of a new field. Proc. Jpn. Acad. Ser. B. 2015, 91, 469–480. [CrossRef] 34. Walter, P. Walking Along the Serendipitous Path of Discovery. Mol. Biol. Cell 2010, 21, 15–17. [CrossRef] Genes 2020, 11, 995 14 of 17

35. Mori, K.; Kawahara, T.; Yoshida, H.; Yanagi, H.; Yura, T. Signalling from endoplasmic reticulum to nucleus: Transcription factor with a basic-leucine zipper motif is required for the unfolded protein-response pathway. Genes Cells 1996, 1, 803–817. [CrossRef][PubMed] 36. Cox, J.S.; Walter, P. A Novel Mechanism for Regulating Activity of a Transcription Factor That Controls the Unfolded Protein Response. Cell 1996, 87, 391–404. [CrossRef] 37. Nikawa, J.; Akiyoshi, M.; Hirata, S.; Fukuda, T. Saccharomyces cerevisiae IRE2/HAC1 is involved in IRE1-mediated KAR2 expression. Nucleic Acids Res. 1996, 24, 4222–4226. [CrossRef][PubMed] 38. Travers, K.J.; Patil, C.K.; Wodicka, L.; Lockhart, D.J.; Weissman, J.S.J.S.; Walter, P. Functional and Genomic Analyses Reveal an Essential Coordination between the Unfolded Protein Response and ER-Associated Degradation. Cell 2000, 101, 249–258. [CrossRef] 39. Fordyce, P.M.; Pincus, D.; Kimmig, P.; Nelson, C.S.; El-Samad, H.; Walter, P.; DeRisi, J.L. Basic leucine zipper transcription factor Hac1 binds DNA in two distinct modes as revealed by microfluidic analyses. Proc. Natl. Acad. Sci. USA 2012, 109, E3084–E3093. [CrossRef] 40. Casagrande, R.; Stern, P.; Diehn, M.; Shamu, C.; Osario, M.; Zúñiga, M.; O Brown, P.; Ploegh, H.L. Degradation of Proteins from the ER of S. cerevisiae Requires an Intact Unfolded Protein Response Pathway. Mol. Cell 2000, 5, 729–735. [CrossRef] 41. Miller, K.A.; DiDone, L.; Krysan, D.J. Extracellular secretion of overexpressed glycosylphosphati dylinositol-linked cell wall protein Utr2/Crh2p as a novel protein quality control mechanism in Saccharomyces cerevisiae. Eukaryot Cell 2010, 9, 1669–1679. [CrossRef][PubMed] 42. Steffen, K.K.; McCormick, M.A.; Pham, K.M.; Mackay, V.L.; Delaney, J.R.; Murakami, C.J.; Kaeberlein, M.; Kennedy, B.K. Ribosome Deficiency Protects Against ER Stress in Saccharomyces cerevisiae. Genet. 2012, 191, 107–118. [CrossRef][PubMed] 43. Mori, K.; Sant, A.; Kohno, K.; Normington, K.; Gething, M.; Sambrook, J. A 22 bp cis-acting element is necessary and sufficient for the induction of the yeast KAR2 (BiP) gene by unfolded proteins. EMBO J. 1992, 11, 2583–2593. [CrossRef][PubMed] 44. Kohno, K.; Normington, K.; Sambrook, J.; Gething, M.J.; Mori, K. The promoter region of the yeast KAR2 (BiP) gene contains a regulatory domain that responds to the presence of unfolded proteins in the endoplasmic reticulum. Mol. Cell. Biol. 1993, 13, 877–890. [CrossRef] 45. Patil, C.K.; Li, H.; Walter, P. Gcn4p and Novel Upstream Activating Sequences Regulate Targets of the Unfolded Protein Response. PLoS Biol. 2004, 2, e246. [CrossRef] 46. Badis, G.; Chan, E.T.; Van Bakel, H.; Peña-Castillo, L.; Tillo, D.; Tsui, K.; Carlson, C.D.; Gossett, A.J.; Hasinoff, M.J.; Warren, C.L.; et al. A Library of Yeast Transcription Factor Motifs Reveals a Widespread Function for Rsc3 in Targeting Nucleosome Exclusion at Promoters. Mol. Cell 2008, 32, 878–887. [CrossRef] 47. Ruberti, I.; Sessa, G.; Lucchetti, S.; Morelli, G. A novel class of plant proteins containing a homeodomain with a closely linked leucine zipper motif. EMBO J. 1991, 10, 1787–1791. [CrossRef] 48. Sessa, G.; Carabelli, M.; Possenti, M.; Morelli, G.; Ruberti, I. Multiple Links between HD-Zip Proteins and Hormone Networks. Int. J. Mol. Sci. 2018, 19, 4047. [CrossRef] 49. Sessa, G.; Morelli, G.; Ruberti, I. The Athb-1 and 2 HD-Zip domains homodimerize forming complexes of − different DNA binding specificities. EMBO J. 1993, 12, 3507–3517. [CrossRef] 50. Van Dalfsen, K.M.; Hodapp, S.; Keskin, A.; Otto, G.M.; Berdan, C.A.; Higdon, A.; Cheunkarndee, T.; Nomura, D.K.; Jovanovic, M.; Brar, G.A. Global Proteome Remodeling during ER Stress Involves Hac1-Driven Expression of Long Undecoded Transcript Isoforms. Dev. Cell 2018, 46, 219–235. [CrossRef] 51. Wingender, E.; Dietze, P.; Karas, H.; Knüppel, R. TRANSFAC: A database on transcription factors and their DNA binding sites. Nucleic Acids Res. 1996, 24, 238–241. [CrossRef] 52. Fornes, O.; A Castro-Mondragon, J.; Khan, A.; Van Der Lee, R.; Zhang, X.; A Richmond, P.; Modi, B.P.; Correard, S.; Gheorghe, M.; Baranašić,D.; et al. JASPAR 2020: Update of the open-access database of transcription factor binding profiles. Nucleic Acids Res. 2019, 48, D87–D92. [CrossRef][PubMed] 53. Weirauch, M.T.; Yang, A.; Albu, M.; Cote, A.G.; Montenegro-Montero, A.; Drewe, P.; Najafabadi, H.S.; Lambert, S.A.; Mann, I.; Cook, K.; et al. Determination and Inference of Eukaryotic Transcription Factor Sequence Specificity. Cell 2014, 158, 1431–1443. [CrossRef] 54. Mei, S.; Qin, Q.; Wu, Q.; Sun, H.; Zheng, R.; Zang, C.; Zhu, M.; Wu, J.; Shi, X.; Taing, L.; et al. Cistrome Data Browser: A data portal for ChIP-Seq and chromatin accessibility data in human and mouse. Nucleic Acids Res. 2016, 45, D658–D662. [CrossRef][PubMed] Genes 2020, 11, 995 15 of 17

55. Zhou, K.-R.; Liu, S.; Sun, W.-J.; Zheng, L.-L.; Zhou, H.; Yang, J.-H.; Qu, L.-H. ChIPBase v2.0: Decoding transcriptional regulatory networks of non-coding RNAs and protein-coding genes from ChIP-seq data. Nucleic Acids Res. 2016, 45, D43–D50. [CrossRef] 56. Yevshin, I.; Sharipov, R.N.; Valeev, T.; Kel, A.; Kolpakov, F. GTRD: A database of transcription factor binding sites identified by ChIP-seq experiments. Nucleic Acids Res. 2016, 45, D61–D67. [CrossRef] 57. Chow, C.-N.; Lee, T.-Y.; Hung, Y.-C.; Li, G.-Z.; Tseng, K.-C.; Liu, Y.-H.; Kuo, P.-L.; Zheng, H.-Q.; Chang, W.-C. PlantPAN3.0: A new and updated resource for reconstructing transcriptional regulatory networks from ChIP-seq experiments in plants. Nucleic Acids Res. 2018, 47, D1155–D1163. [CrossRef][PubMed] 58. Jin, J.; Tian, F.; Yang, D.-C.; Meng, Y.-Q.; Kong, L.; Luo, J.; Gao, G. PlantTFDB 4.0: Toward a central hub for transcription factors and regulatory interactions in plants. Nucleic Acids Res. 2016, 45, D1040–D1045. [CrossRef][PubMed] 59. Ambrosini, G. PWMTools. Available online: http://ccg.vital-it.ch/pwmtools (accessed on 24 August 2020). 60. Grant, C.E.; Bailey, T.L.; Noble, W.S. FIMO: Scanning for occurrences of a given motif. Bioinformatics 2011, 27, 1017–1018. [CrossRef][PubMed] 61. Heinz, S.; Benner, C.; Spann, N.; Bertolino, E.; Lin, Y.C.; Laslo, P.; Cheng, J.X.; Murre, C.; Singh, H.; Glass, C.K. Simple Combinations of Lineage-Determining Transcription Factors Prime cis-Regulatory Elements Required for Macrophage and B Cell Identities. Mol. Cell 2010, 38, 576–589. [CrossRef][PubMed] 62. Xia, X. Position Weight Matrix, Gibbs Sampler, and the Associated Significance Tests in Motif Characterization and Prediction. Science 2012, 2012, 1–15. [CrossRef][PubMed] 63. Xia, X. Position weight matrix and Perceptron. In Bioinformatics and the Cell; Springer Science and Business Media LLC: Berlin, Germany, 2018; pp. 77–98. 64. Miguel, V.N.; Manavella, P.A.; Chan, R.L.; Capella, M. The AtHB1 Transcription Factor Controls the miR164-CUC2 Regulatory Node to Modulate Leaf Development. Plant Cell Physiol. 2019, 61, 659–670. [CrossRef][PubMed] 65. Kulkarni, S.R.; Jones, D.M.; Vandepoele, K. Enhanced Maps of Transcription Factor Binding Sites Improve Regulatory Networks Learned from Accessible Chromatin Data. Plant Physiol. 2019, 181, 412–425. [CrossRef] [PubMed] 66. Xia, X. DAMBE6: New Tools for Microbial Genomics, Phylogenetics, and Molecular Evolution. J. Hered. 2017, 108, 431–437. [CrossRef] 67. Xia, X. DAMBE7: New and Improved Tools for Data Analysis in Molecular Biology and Evolution. Mol. Biol. Evol. 2018, 35, 1550–1552. [CrossRef] 68. Capella, M.; Ribone, P.A.; Arce, A.L.; Chan, R.L. Arabidopsis thaliana HomeoBox 1 (At HB 1), a Homedomain-Leucine Zipper I (HD-Zip I) transcription factor, is regulated by PHYTOCHROME- INTERACTING FACTOR 1 to promote hypocotyl elongation. New Phytol. 2015, 207, 669–682. [CrossRef] 69. Ezer, D.; Shepherd, S.J.; Brestovitsky, A.; Dickinson, P.; Cortijo, S.; Charoensawan, V.; Box, M.S.; Biswas, S.; Jaeger, K.E.; Wigge, P.A. The G-Box Transcriptional Regulatory Code in Arabidopsis. Plant Physiol. 2017, 175, 628–640. [CrossRef][PubMed] 70. Stegmaier, P.; E Kel, A.; Wingender, E. Systematic DNA-binding domain classification of transcription factors. Genome Inform. Int. Conf. Genome Inform. 2004, 15, 276–286. 71. Kalderon, D.; Roberts, B.L.; Richardson, W.D.; Smith, A.E. A short amino acid sequence able to specify nuclear location. Cell 1984, 39, 499–509. [CrossRef] 72. Landschulz, W.; Johnson, P.F.; McKnight, S. The leucine zipper: A hypothetical structure common to a new class of DNA binding proteins. Science 1988, 240, 1759–1764. [CrossRef] 73. Pal, B.; Chan, N.C.; Helfenbaum, L.; Tan, K.; Tansey, W.P.; Gething, M.-J. SCFCdc4-mediated Degradation of the Hac1p Transcription Factor Regulates the Unfolded Protein Response inSaccharomyces cerevisiae. Mol. Biol. Cell 2007, 18, 426–440. [CrossRef][PubMed] 74. Yoshida, H.; Oku, M.; Suzuki, M.; Mori, K. pXBP1(U) encoded in XBP1 pre-mRNA negatively regulates unfolded protein response activator pXBP1(S) in mammalian ER stress response. J. Cell Biol. 2006, 172, 565–575. [CrossRef] 75. Goffin, L.; Vodala, S.; Fraser, C.; Ryan, J.; Timms, M.; Meusburger, S.; Catimel, B.; Nice, E.C.; Silver, P.A.; Xiao, C.-Y.; et al. The unfolded protein response transducer Ire1p contains a nuclear localization sequence recognized by multiple β importins. Mol. Biol. Cell 2006, 17, 5309–5323. [CrossRef] Genes 2020, 11, 995 16 of 17

76. Fontes, M.R.M.; Teh, T.; Kobe, B. Structural basis of recognition of monopartite and bipartite nuclear localization sequences by mammalian importin-α11Edited by K. Nagai. J. Mol. Biol. 2000, 297, 1183–1194. [CrossRef][PubMed] 77. Zeng, X.; Herndon, A.M.; Hu, J.C. Buried asparagines determine the dimerization specificities of leucine zipper mutants. Proc. Natl. Acad. Sci. USA 1997, 94, 3673–3678. [CrossRef][PubMed] 78. Clauss, I. The basic domain/leucine zipper protein hXBP-1 preferentially binds to and transactivates CRE-like sequences containing an ACGT core. Nucleic Acids Res. 1996, 24, 1855–1864. [CrossRef][PubMed] 79. He, Y.; Sun, S.; Sha, H.; Liu, Z.; Yang, L.; Xue, Z.; Chen, H.; Qi, L.; Qi, A.L. Emerging Roles for XBP1, a sUPeR Transcription Factor. Gene Expr. 2010, 15, 13–25. [CrossRef][PubMed] 80. Iwata, Y.; Fedoroff, N.V.; Koizumi, N. Arabidopsis bZIP60 Is a Proteolysis-Activated Transcription Factor Involved in the Endoplasmic Reticulum Stress Response. Plant Cell 2008, 20, 3107–3121. [CrossRef] 81. Tjong, H.; Zhou, H.-X. DISPLAR: An accurate method for predicting DNA-binding sites on protein surfaces. Nucleic Acids Res. 2007, 35, 1465–1477. [CrossRef] 82. Siggers, T.; Gordân, R. Protein-DNA binding: Complexities and multi-protein codes. Nucleic Acids Res. 2013, 42, 2099–2111. [CrossRef] 83. Siggers, T.W.; Honig, B. Structure-based prediction of C2H2 zinc-finger binding specificity: Sensitivity to docking geometry. Nucleic Acids Res. 2007, 35, 1085–1097. [CrossRef][PubMed] 84. Farrel, A.; Guo, J.-T. An efficient algorithm for improving structure-based prediction of transcription factor binding sites. BMC Bioinform. 2017, 18, 342. [CrossRef] 85. Farrel, A.; Murphy, J.; Guo, J.-T. Structure-based prediction of transcription factor binding specificity using an integrative energy function. Bioinformatics 2016, 32, i306–i313. [CrossRef][PubMed] 86. Ofran, Y.; Mysore, V.; Rost, B. Prediction of DNA-binding residues from sequence. Bioinformatics 2007, 23, i347–i353. [CrossRef][PubMed] 87. Foat, B.C.; Morozov, A.V.; Bussemaker, H.J. Statistical mechanical modeling of genome-wide transcription factor occupancy data by MatrixREDUCE. Bioinformatics 2006, 22, e141–e149. [CrossRef] 88. Feng, J.; Liu, T.; Qin, B.; Zhang, Y.; Liu, X.S. Identifying ChIP-seq enrichment using MACS. Nat. Protoc. 2012, 7, 1728–1740. [CrossRef] 89. Neuwald, A.F.; Liu, J.S.; Lawrence, C.E. Gibbs motif sampling: Detection of bacterial outer membrane protein repeats. Protein Sci. 1995, 4, 1618–1632. [CrossRef] 90. A Mannella, C.; Neuwald, A.F.; E Lawrence, C. Detection of likely transmembrane β strand regions in sequences of mitochondrial pore proteins using the Gibbs sampler. J. Bioenerg. Biomembr. 1996, 28, 163–169. [CrossRef] 91. Qu, K.; A McCue, L.; Lawrence, C.E. Bayesian protein family classifier. Proc. Int. Conf. Intell. Syst. Mol. Biol. 1998, 6, 131–139. 92. E Lawrence, C.; Altschul, S.F.; Boguski, M.S.; Liu, J.S.; Neuwald, A.F.; Wootton, J.C. Detecting subtle sequence signals: A Gibbs sampling strategy for multiple alignment. Science 1993, 262, 208–214. [CrossRef] 93. Aerts, S.; Van Loo, P.; Thijs, G.; Mayer, H.; De Martin, R.; Moreau, Y.; De Moor, B. TOUCAN 2: The all-inclusive open source workbench for regulatory sequence analysis. Nucleic Acids Res. 2005, 33, W393–W396. [CrossRef] [PubMed] 94. Coessens, B.; Thijs, G.; Aerts, S.; Marchal, K.; De Smet, F.; Engelen, K.; Glenisson, P.; Moreau, Y.; Mathys, J.; De Moor, B. INCLUSive: A web portal and service registry for microarray and regulatory sequence analysis. Nucleic Acids Res. 2003, 31, 3468–3470. [CrossRef][PubMed] 95. Thijs, G.; Marchal, K.; Lescot, M.; Rombauts, S.; De Moor, B.; Rouzé, P.; Moreau, Y. A Gibbs Sampling Method to Detect Overrepresented Motifs in the Upstream Regions of Coexpressed Genes. J. Comput. Biol. 2002, 9, 447–464. [CrossRef] 96. Thijs, G.; Moreau, Y.; De Smet, F.; Mathys, J.; Lescot, M.; Rombauts, S.; Rouze, P.; De Moor, B.; Marchal, K. INCLUSive: Integrated clustering, upstream sequence retrieval and motif sampling. Bioinformatics 2002, 18, 331–332. [CrossRef][PubMed] 97. Thijs, G.; Lescot, M.; Marchal, K.; Rombauts, S.; De Moor, B.; Rouzé, P.; Moreau, Y. A higher-order background model improves the detection of promoter regulatory elements by Gibbs sampling. Bioinformatics 2001, 17, 1113–1122. [CrossRef] Genes 2020, 11, 995 17 of 17

98. Qin, Z.S.; McCue, L.A.; Thompson, W.; Mayerhofer, L.; Lawrence, C.E.; Liu, J.S. Identiﬁcation of co-regulated genes through Bayesian clustering of predicted regulatory binding sites. Nat. Biotechnol. 2003, 21, 435–439. [CrossRef] 99. Thompson, W.; Palumbo, M.J.; Wasserman, W.W.; Liu, J.S.; Lawrence, C.E. Decoding Human Regulatory Circuits. Genome Res. 2004, 14, 1967–1974. [CrossRef] 100. Thompson, W. Gibbs Recursive Sampler: Finding transcription factor binding sites. Nucleic Acids Res. 2003, 31, 3580–3585. [CrossRef] 101. Xia, X. (Ed.) Gibbs sampler. In Bioinformatics and the Cell: Modern Computational Approaches in Genomics, Proteomics and Transcriptomics; Springer: Cham, Switzerland, 2018; pp. 99–111.

© 2020 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).