Plant and animal chromatin three-dimensional organization: similar structures but different functions

Chromatin is the main carrier of genetic information and is non-randomly distributed within the nucleus. Next-generation sequence-based chromatin conformation capture technologies have enabled us to directly examine its three-dimensional organization at an unprecedented scale and resolution. In the best-studied mammalian models, chromatin folding can be broken down into three hierarchical levels, compartment, domains, and loops, which play important roles in transcriptional regulation. Although similar structures have now been identified in plants, they might not possess exactly the same functions as the mammalian ones. Here, we review recent Hi-C studies in plants, compare plant chromatin structures with their mammalian counterparts, and discuss the differences between plants with different genome sizes.


Introduction
Chromosomes contain different types of genetic information, such as genes, cis-regulatory elements, and repeats, all of which are stored in distinct locations inside the nucleus and require specific packaging and folding to function. For example, a regulatory element hundreds of kilobases away from a gene would need to be folded back to the proximal promoter in order to regulate the transcription of the gene. It is now widely accepted that the spatial organization of chromatin plays important roles in biological processes such as DNA replication and repair, spatial-and temporal-specific gene expression, and in transposable element (TE) repression. Hence, it is of great interest to examine the three-dimensional (3D) organization of chromatin and associate its structural features to functions; this will enable us to better understand the transcriptional regulation of key biological processes.
In the past two decades, the development of techniques that combine chromosome conformation capture (3C) and high-throughput sequencing has dramatically improved our understanding of chromosome spatial organization (Denker and De Laat, 2016). Among these techniques, Hi-C is capable of capturing all chromatin interactions genome-wide at low resolution (Lieberman-Aiden et al., 2009), while chromatin interaction analysis by paired-end tag sequencing (ChIA-PET) (Fullwood et al., 2009) and Hi-C coupled to ChIP-seq (HiChIP or PLAC-Seq) (Fang et al., 2016;Mumbach et al., organization, but less is known about the roles of chromatin organization in plants. In plants, early cytological studies have shown that chromosomes at interphase are non-randomly organized; each chromosome occupies a distinct and exclusive space within the nucleus, known as the chromosome territory (Tiang et al., 2012). The plant genome varies remarkably in size and TE content due to whole-genome duplication, TE amplification, and DNA purging, and exhibits different chromosome territory patterns (Dong and Jiang, 1998;Fransz and Jong, 2011;Tiang et al., 2012). Chromosomes of large-genome plants, such as barley and rye, exhibit the "Rabl" configuration during interphase, in which the telomere and centromere residue at the opposite poles of the nuclei, reminiscent of the chromosome conformation at anaphase during mitosis (Dong and Jiang, 1998). By contrast, the small-genome plants often adopt various non-Rabl chromosome configurations. For example, in Arabidopsis, most repeat elements are located in the pericentromeric region and form a densely packed distinct chromocenter, while the euchromatin emanates outward, forming a "Rosette" configuration (Fransz et al., 2002). Despite the low resolution and focus on centromere and telomere positions, these early studies clearly illustrated the diversity of 3D chromatin organization in plants.
Recently, 3C technologies have been used to examine chromatin 3D organization in multiple crop species, in different tissues or in the same tissue under different growth conditions. In this review, we will discuss these new findings and compare plant 3C data side by side with the best-studied animal models at the levels of compartment, domain, and loops, in order to provide an overview of this emerging field.

Mammalian A/B compartments
The first Hi-C experiment showed that chromatin is nonrandomly arranged inside the mammalian nucleus, and the active and repressive regions are spatially segregated (Lieberman-Aiden et al., 2009). A key finding of this pilot study was that the active chromatin prefers to interact with the other active regions, and the repressive chromatin also interacts with itself. Thus the genome is partitioned into two different territories in space and exhibits a plaid pattern on the Hi-C interaction frequency matrix. The active and repressive territories are referred to as the A compartment and the B compartment, respectively.
The mammalian A compartment is actively transcribed and is enriched for open chromatin, active histone modifications such as H3K4me3 and H3K27ac, and high GC content. The B compartment is associated with the nuclear lamina, is enriched for repressive histone marks such as H3K9me3, and is AT rich (Lieberman-Aiden et al., 2009;Ryba et al., 2010). The A/B compartments can be further divided into sub-compartments at higher resolution, which also have unique histone modification patterns (Rao et al., 2014). The Hi-C based compartment definition is consistent with previous microscopic and nuclear lamina association studies, which found that active chromatin residues in the center of the nucleus and the heterochromatin is pushed toward the nuclear periphery (Bickmore and van Steensel, 2013). One could argue that the A/B compartments could be an artifact resulting from averaging the interaction signal from millions of cells analyzed in Hi-C. However, superresolution imaging has confirmed that compartments can exist at the single-cell level and are polarized along the chromosome .
Compartment partitioning is not static, and can switch frequently between different tissues or developmental stages. For example, 36% of the human genome switched compartments in at least one of the lineages analyzed; loci that switched from A to B often showed decreased gene expression, while those that switched from B to A often showed increased expression (Dixon et al., 2015). Intra-chromosome chromatin co-accessibility also positively correlates with the compartment dynamics, further suggesting that compartment organization could be linked to transcription (Gate et al., 2018).

Global and local A/B compartments in plants
Using Hi-C, compartment-like structures have been identified in plant species with genomes of different sizes, such as Arabidopsis (135 Mb), rice (430 Mb), and maize (2.4 Gb) (Feng et al., 2014;Grob et al., 2014;Dong et al., 2017;Liu et al., 2017). In plant chromosomes, the two actively transcribed euchromatin arms often form the A compartment and the pericentromeric heterochromatin forms the B compartment ( Fig. 1) (Feng et al., 2014;Grob et al., 2014). This partitioning is largely stable across tissues (Dong et al., 2020), and reduced compartment interaction has been found in Arabidopsis DNA methylation mutants (Feng et al., 2014) and in rice and maize endosperm tissues in which DNA demethylation occurred naturally (Dong et al., 2020). These findings support the hypothesis that the genome-wide compartment partitioning could be due to spatial separation of the loose euchromatin and the condensed heterochromatin.
Besides the A/B compartments, another compartment-like structure can be identified within the Arabidopsis A compartment (Fig. 1A). Based on the local Hi-C interaction matrix of individual Arabidopsis chromosome arms, regions tending to associate with the chromocenter were called compacted structural domains (CSDs), while the remaining regions, which are associated with active gene expression, were referred to as loose structural domains (LSDs) (Grob et al., 2014). It has been shown that the CSDs are associated with the nuclear periphery, which requires lamina-like proteins CRWN1 and CRWN4, as well as CHG and CHH DNA methylation (Bi et al., 2017;Grob and Grossniklaus, 2019;Hu et al., 2019).
In plants with medium-sized and large genomes, such as tomato and maize, compartment partitioning can be even more complicated, as their euchromatin and heterochromatin are much larger and less homogeneous than those of Arabidopsis. For example, in tomato chromosomes, most of the heterochromatin is categorized as the B compartment using the genomewide Hi-C interaction matrix at low resolution (bin size 0.5 Mb). When a higher resolution or the local Hi-C interaction matrix is used, gene islands inside the heterochromatin and TE/repeat regions in the euchromatin are identified as local A and B compartments, respectively (Fig. 1B). This phenomenon is most pronounced in plants with large TE-rich genomes, such as maize, tomato, and sorghum (Dong et al., 2017(Dong et al., , 2020.

Phase separation, nuclear lamina, and compartment formation
Several layers of force have been proposed that contribute to the observed partitioning into A/B compartments. The B compartment was observed to overlap well with nuclear lamina-associated domains, suggesting that the A/B compartments might be formed by separating the chromatin into two parts, in the nuclear periphery and the nuclear interior (van Steensel and Belmont, 2017). However, in the inverted nuclei of mouse rod photoreceptor cells, the heterochromatin is located at the nuclear center and the euchromatin is located around the nuclear periphery. The A/B compartment annotation of these nuclei remained unchanged, suggesting that localization at the periphery or the interior of the nucleus might not drive A/B compartmentation (Falk et al., 2019). Another hypothesis is that transcription activity drives formation of the A compartment, as the actively transcribed regions are located in the nuclear interior. It has been shown that knockdown of RNA PolII resulted in euchromatin being pushed toward the nuclear periphery (Krüger et al., 2015;Stevens et al., 2017).
More recently, it has been suggested that the attraction between heterochromatin is the key driving force for mammalian chromatin compartmentalization (Falk et al., 2019). This chromatin affinity theory is in line with the liquid-liquid phase separation model, in which the concentration of macromolecules, as well as other characteristics such as salt type, co-solutes, and pH, helps macromolecules to condense into dense phases, which exist alongside neighboring dilute phases (Alberti et al., 2019). In addition, chromatin bridges between H3K9me2/3 nucleosomes can promote the formation of collapsed chromatin globules that are separated from the surrounding chromatin, which means the loci within the globule prefer to contact each other and behave like one compartment (Hiragami-Hamada et al., 2016). In plants with large genomes, both global and local compartment partitions correlate well with heterochromatin/euchromatin status, and these compartments exhibit features like those typical of mammalian compartments (Dong et al., 2017), suggesting that the chromatin attraction and phase separation model might explain their formation (Fig. 1B). However, the Arabidopsis LSD/CSD-type compartments, which lack a clear distinction in euchromatin and heterochromatin status, might result from nuclear periphery positioning . In addition, electron microscopy confirmed that plant heterochromatin is not always pushed toward the nuclear periphery in the large-genome plants such as maize (Fig. 2), a finding that is consistent with the early cytology studies (Fransz and Jong, 2011). In addition, unlike the local compartments in the large-genome plants, the LSD/CSD compartments in Arabidopsis do not overlap with chromatin domains, which are another layer of 3D chromatin organization that will be discussed in the following sections.

Mammalian TADs and the loop extrusion model
Chromatin domains are the most prominent feature in the mammalian genome, and are often referred to as topologically associated domains (TADs). The interaction frequency of loci within a TAD is higher than that of loci between TADs and decays sharply at the domain boundaries (Dixon et al., 2012;Nora et al., 2012). This interaction pattern hints at a potential function of TADs, which is to confine chromatin interactions, such as those between distal enhancers and proximal promoters, within a single domain.
This hypothesis is further supported by observations that the expression of genes within a TAD are weakly coordinated (Dixon et al., 2012;Nora et al., 2012;Zhan et al., 2017) and the majority of promoter-enhancer interactions are restricted within TADs (Ji et al., 2016). By inserting reporter genes into TADs, it was found that the sensor expression profile correlates well with the TAD environment, suggesting that TADs partition the genome into isolating blocks within which the regulatory elements are confined (Symmons et al., 2014). It is worth noting that after the destruction of most TADs, the overall gene expression pattern is not changed (Nora et al., 2017;Rao et al., 2017;Ghavi-Helm et al., 2019), indicating that TADs are not the only mechanism for regulating gene expression. It has also been shown that loci within a TAD exhibit similar histone modification patterns (Le Dily et al., 2014;Rao et al., 2014) and that the boundaries of TADs overlap with the boundaries of DNA replication domains (Pope et al., 2014).
The borders of some TADs were found to have CCCTCbinding factor (CTCF) and cohesin binding, which could form a chromatin loop, and this type of domain is often called the "loop domain" (Fig. 3A) (Rao et al., 2014). CTCF is an 11-zinc-finger (ZF) transcription factor that has an orientation-specific binding motif and is famous for its role in insulator function (Heger et al., 2012). This finding later led to the loop extrusion model of TAD formation (Murayama et al., 2018;Davidson et al., 2019). The structural maintenance of chromosomes (SMC) subunits of the cohesin complex can form dimers through their hinge domain at one end, and interact with DNA at the other end (Wood et al., 2010). When the cohesin complex binds to two different chromatin loci, the two subunits progressively slide along the chromatin fiber to form a loop, which is eventually stalled by the directional CTCF binding (Fudenberg et al., 2016). During this process, the interaction between loci within two convergent CTCF binding sites is facilitated by the extrusion of the loop (Fudenberg et al., 2016).
A growing body of evidence supports the loop extrusion model. It has been found that cohesin binding is highly mobile in the human genome, and its binding sites often occur at the inner side of CTCF at the TAD border (Tang et al., 2015). Degradation of either CTCF or a cohesin subunit can disrupt the TAD structure (Nora et al., 2017;Rao et al., 2017). Flipping the CTCF binding site can also disrupt the TAD (Guo et al., 2015). Recent single molecular imaging studies also support that cohesin extrusion contributes to loop formation (Davidson et al., 2019;Kim et al., 2019).
It is important to keep in mind that all 3D structures, such as A/B compartments and domains identified by 3C experiments, are likely to represent the average interaction pattern of a large number of cells. For example, with highresolution Hi-C data, human TADs can be further partitioned into subdomains (sub-TADs or contact domains). These domains exhibit a similar self-association structure with a median size of ~200 kb, smaller than the TADs, which are at the megabase scale (Fig. 3A) (Phillips-Cremins et al., 2013;Rao et al., 2014). In the smaller genome of Drosophila (180 Mb), the TAD-like domains are similar in size to the human sub-TADs at 50-100 kb, depending on the Hi-C resolution (Bonev and Cavalli, 2016). Recent single-cell Hi-C studies have now confirmed that TADs exhibit substantial variation among cell populations (Stevens et al., 2017;Tan et al., 2018). In addition, domain border loci do not always co-localize, and TADs can break or merge into bulk TADs (Fig. 3A).

TAD-like chromatin domains in plants
TAD organization is not a prominent feature in Arabidopsis (Feng et al., 2014;Grob et al., 2014). Using kilobase-resolution Hi-C, few domain-like structures could be identified in Arabidopsis. Their boundaries are enriched for active genes and associated with active epigenetic marks such as open chromatin, H3K4me3, and H3K9ac, while the interior regions are not actively transcribed (Wang et al., 2015). Compared with the mammalian TADs, the Arabidopsis domains are smaller and the interaction strength is weaker (Fig. 3C). In addition, a few TAD-like structures have also been found in the Arabidopsis H3K27me3-rich and H3K9me2-rich chromocenter heterochromatic regions (Feng et al., 2014;Rowley et al., 2017).
In large-genome plants such as maize and tomato, a lot more TAD-like structures can be identified, and TAD-like domains could cover 50-90% of their genomes depending on the calling algorithm and cutoff (Fig. 3B). Their domain boundaries are enriched for active genes and active histone marks, and are stable in different tissues, developmental stages, and growth conditions (Dong et al., 2017(Dong et al., , 2018(Dong et al., , 2020Liu et al., 2017;Wang et al., 2018). Unlike the mammalian TADs, most of these plant domain-like structures overlap with the local A/B compartment, and should be referred to as "compartment domains" (Fig. 3A) (Dong et al., 2017).

Unique features of the plant compartment domains
Mammalian TADs and plant compartment domains are two different structures that should not be confused (Fig. 3). By definition, TADs and domains are structures marked by the strong interaction within themselves and also strong insulation at the border region, while compartment status is defined by a region's global interaction pattern. Cohesin and CTCF are required for the formation of the mammalian loop domains. In studies, when cohesin or CTCF was depleted, compartment switches were often observed between neighboring loci, and small compartment domains associated with uniform epigenetic marks were formed, which were similar to the compartment domains found in plants (Nora et al., 2017;Rao et al., 2017). CTCF is present only in bilateria and is absent from plant genomes (Heger et al., 2012;Heger and Wiehe, 2014). Although proteins such as AS1 and AS2 have been suggested to have a potential insulator function in the plant genome, the limited binding sites make them unlikely to serve as a general insulator protein (Guo et al., 2008;Iwasaki et al., 2013). Besides, one study in rice showed that the enriched motifs at domain borders belong to the TCP transcription factor and bZIP proteins rather than those of the Zn finger proteins like CTCF . A unique feature of the plant compartment domain is that domains of the same type interact with each other at both intra-and inter-chromosome levels, forming a checkerboard pattern like that of the A/B compartments (Dong et al., 2017). Mammalian TADs show no such pattern, and they often exhibit a nested structure (Fig. 3), while their loops usually occur at the inner side of the loop domain and are called "corner loops". In both maize and tomato, large chromatin loops between gene islands can be detected by Hi-C. However, these loops are enriched outside of the domain (Dong et al., 2017), which is similar to the small compartment domains of Drosophila (Rowley et al., 2017).
Mammalian loop domains are conserved between different species, while the conservation of loop domains is associated with conservation of CTCF binding sites. Plant compartment domains also have different chromatin states, similar to the ones in Drosophila, while their border regions are not conserved in related plant species (Dong et al., 2017;Rowley et al., 2017). Neighboring plant compartment domains often possess different transcription activities, unlike the TAD boundaries in human that are able to separate consecutive active or consecutive heterochromatin regions. It is most likely that without a CTCF-like factor to physically clamp two chromatin fibers together, the interaction of plant domains would be weaker than that of the mammalian ones. When comparing Hi-C data of different plant tissues, it was found that the changes of domain border are often associated with differential gene expression (Dong et al., 2020). In addition, the Arabidopsis heterochromatin domains are diminished in methylation mutants, and the domain insulation becomes much weaker in endosperm tissues (Feng et al., 2014;Dong et al., 2020). Together, these findings suggest that transcriptional activity and heterochromatin status could be the main factors determining domain formation in plants.

Early studies of promoter chromatin loops
It is well known that distal regulatory elements can make physical contact with genes through chromatin looping. The best-studied case is the β-globin genes, which revealed a causal relationship between looping and gene activation (Smallwood and Ren, 2013). With 3C coupled to sequencing, many chromatin loops have now been identified in the mammalian genome. As well as promoter-enhancer interactions, promoter-promoter and enhancer-enhancer interactions are also prevalent, and the interacting gene pairs tend to be expressed in a coordinated manner Ji et al., 2016). These transcription-related loops are often associated with a mediator complex and are cell type specific. The most prominent loops are those formed between loci bound by the CTCF and cohesin. These loops exhibit the most intense interaction frequency and are relatively conserved between cell types (Dowen et al., 2014). There are also some loops associated with repressive signals, such as polycomb repressive complex 2 (PRC2) mark H3K27me3 (Entrevan et al., 2016). The first identified long-range interaction in plant species is that formed between the maize b1 gene and upstream regulatory elements spanning ~100 kb. Active expression of the b1 gene requires a proper loop structure, which is mediated in a tissue-and epiallele-specific manner (Louwers et al., 2009).

Loops between plant heterochromatin regions
Long-range promoter-enhancer loops are not common in the model plant Arabidopsis (Feng et al., 2014;Grob et al., 2014). The most prominent long-range interaction in the Arabidopsis genome is formed between 10 heterochromatin islands from different chromosomes and is termed Interactive Heterochromatic Islands (IHI or KNOT) (Feng et al., 2014;Grob et al., 2014). These loci are located in both euchromatic arms and pericentromeric heterochromatic chromocenters, and interact with each other as well as with the telomeric regions, whereas they do not interact with other pericentromeric regions (Feng et al., 2014). They also form a separate compartment to the rest of the genome. Similar loops have been observed in the monocot rice genome, where the IHI is also enriched for TE, sRNAs, and the repressive histone mark H3K9me2 (Dong et al., 2018).
A recent study showed that IHI is able to silence transgenes through establishing physical interactions with transgene loci, suggesting a potential role in defense against invasive DNA (Grob and Grossniklaus, 2019). Counterintuitively, heterochromatin condensation might not be required for IHI loops, since they remained intact in suvh4/suvh5/suvh6 triple mutants (which show reduced non-CG methylation and H3K9me2) as well as ddm1 and met1 mutants (which show reduced DNA methylation in all sequence contents); in fact, more IHI loops were found in these mutants (Feng et al., 2014).
It is worth noting that large-genome plants also have similar heterochromatin notches in the euchromatic arms without forming IHI-like structures. A recent H3K9me2 ChIA-PET experiment in maize identified over 10 000 H3K9me2associated loops. Unlike the IHIs, most of them are formed between loci within chromocenters . However, changes in these heterochromatin loops are not associated with gene expression changes, unlike the mammalian ones.

A variety of loops in plant euchromatin
Despite the lack of long-range interaction, short-range loops have been found in Arabidopsis (Crevillén et al., 2013;Ariel et al., 2014;Cao et al., 2014;Liu et al., 2014Liu et al., , 2016. The actively transcribed FLC locus forms a loop between the 5′ and 3′ end of the gene. Upon cold treatment, FLC is repressed by the polycomb protein complex and the loop is disrupted (Crevillén et al., 2013). Ultra-high-resolution Hi-C at sub-kilobase resolution has also identified over 20 000 chromatin loops ranging from 2 kb to 25 kb in Arabidopsis . Such gene self-looping often correlates with gene expression, suggesting either a potential regulatory role or that it is the active transcription that causes the head-to-tail gene looping.
In large-genome plants, many long-range chromatin interactions are found between gene islands and can be identified as "loops" from the Hi-C interaction matrix (Dong et al., 2017(Dong et al., , 2020Wang et al., 2017). For example, in the tomato genome, the highly expressed genes located in the heterochromatin neighborhood tend to interact with each other, while this pattern is diminished in the gene-rich euchromatin (Fig. 4). Their loop anchor loci are enriched for active histone modifications (H3K4me3 and H3K27ac), while tissue-specific loop changes are often associated with changes in gene expression (Dong et al., 2017(Dong et al., , 2020Wang et al., 2017). One interesting observation is that if two gene loci are joined by a chromatin loop in maize, their syntenic gene pairs in related species (rice, sorghum, and millet) have a shorter genomic distance compared with those of the non-loop genes (Dong et al., 2020). This suggests that the loop genes are under selection and that they could be co-regulated. It should be noted that despite the large variation in genome size, most plants have similar numbers of genes and open chromatin regions. As a result, genes and their distal regulatory elements are likely to be separated by more TEs and repeats as the genome size increases.
However, it is challenging for Hi-C to resolve loops at the 10 kb range, in large and repetitive genomes, where most of the plant cis-regulatory elements reside. Recently, ChIA-PET and HiChIP, which have higher resolution than Hi-C, have been used to study the chromatin loops (Li et al., 2019;Peng et al., 2019;Ricci et al., 2019;Zhao et al., 2019). A plethora of geneto-gene and gene-to-distal open chromatin loops have been identified in rice and maize by ChIA-PET. These chromatin loop loci are often enriched for expression quantitative trait loci, and the loop genes exhibit a correlated expression pattern. Unlike the loops in animal species, the 10 kb range loops detected by ChIA-PET are not enriched within plant domains, supporting the argument that, in contrast to the mammalian TADs, the plant compartment domains are not used to confine enhancer-promoter interactions.

Concluding remarks
Plant interphase chromatin is non-randomly organized within the nucleus and occupies a distinct space (Tiang et al., 2012). In the past two decades, the development of 3C-based techniques has enabled us to directly examine the 3D organization of chromatin (Yu and Ren, 2017). Most of these techniques are based on the principle that spatially close DNA fragments are ligated more efficiently than distal ones. Therefore, the relative Fig. 4. Chromatin loops form between active gene islands in regions of low gene density. Tomato genes are ranked according to their expression level in leaf tissue and neighborhood gene density (1 Mb upstream and downstream), and are equally divided into 4×4 groups based on the dual-index ranking (Dong et al., 2017). The heatmap represents the average Hi-C interaction matrix for each group of genes centered on their gene transcriptional start sites (TSS). Highly expressed genes located in low-gene-density regions are more likely to be associated with chromatin loops. In other words, these Hi-C loops are often identified between two actively transcribed regions separated by TE-rich heterochromatin, consistent with the observation that these loops are often small local compartments. spatial distance between any two DNA loci could be inferred from their ligatability. However, we should keep in mind that all these so-called plant chromatin "3D structures" inferred from 3C data are based on the average ligatability of DNA in millions of cells, and often from a non-uniform cell population. Without single-cell-based assays, it is impossible to be certain whether these structures actually exist in individual cells.
Multiple Hi-C studies have now confirmed that plant cells have similar compartment, domain, and loop structures to those in mammalian cells (Feng et al., 2014;Grob et al., 2014;Liu et al., 2016Liu et al., , 2017Dong et al., 2017Dong et al., , 2018Wang et al., 2017;Li et al., 2019). Many of the observed differences between plant and animal species could be due to the lack of a CTCF-like insulator protein in plants. As we have passed the early discovery phase, the next big question is whether these structures have biological functions, rather than being the result of the spatial separation of euchromatin and heterochromatin.
The other equally important but often overlooked question is why plants have not evolved a CTCF equivalent and lack the mammalian TAD-like loop domains. As TADs and chromatin loops are vital for mammalian transcriptional regulation, would the lack of this important tool fundamentally change the way gene expression is regulated in plants? After all, plants and animals have similar numbers of protein-coding genes and of open chromatin regions per cell/tissue type. However, open chromatin regions of plants show significantly less tissue-specific dynamics compared with the mammalian ones (Mascher et al., 2017;Corces et al., 2018;Lü et al., 2018;Li et al., 2019;Peng et al., 2019;Yoshida et al., 2019). One possibility is that without TADs and CTCF loops it would be difficult for a plant gene to "loop" with a different open chromatin region when its expression level needs to change in different tissues. Alternatively, the lack of tissue-specific open chromatin dynamics could be the reason why plants did not evolve TADs or CTCF. Hence, alternative mechanisms such as tissue-specific epigenetic marks would be used to regulate gene expression in plants. Perhaps that is the penalty for not evolving TADs and CTCF loops.
The quest to understand plant 3D chromatin organization and its functions is by no means complete. The importance of crop models and the long divergence time of plants and animals, and even within plant kingdoms, ensure that there will be plenty of novel discoveries in the years to come.