BIOLOGICAL OVERVIEW

Regulatory elements termed enhancers, or locus control regions are capable of exerting their influence over long distances in an orientation-independent manner to orchestrate the complex gene expression patterns required for embryonic development. How are the effects of enhancers confined to the genes they regulate? In recent years the concept of domain boundaries or insulator elements has developed, based on the genetic properties of several eukaryotic genes. These elements serve to isolate chromosomal regions, confining regulatory effects of enhancers to the genes they are meant to serve. For example, specialized chromatin structures (scs and scs') have been found at outside the coding sequence of the Drosophila hsp70 gene. These sequences are associated with chromatin structures and serve as boundaries that can prevent activation by enhancer elements. Similarly the chicken ßglobin insulator can inhibit activation fo the gamma-globin gene promoter by the ß-globin locus control region. The Drosophila gypsy insulator functions similarly, confering position-independent transcription to genes and preventing activation of promoters by enhancers separated from proximal promoters by insulator elements (Gdula, 1996 and references).

The best characterized system is the gypsy chromatin insulator of Drosophila. Gypsy is an infectious retrovirus consisting of two long terminal repeat regions that function in viral replication and integration and three retroviral genes: gag, pol and env, coding for nucleocapsid, reverse transcriptase and envelope proteins respectively (Song, 1994). The gypsy insulator is located in the 5' transcribed untranslated region of the gypsy element. The unlinked gene, suppressor of Hairy-wing plays an essential role in functioning of the gypsy insulator. SU(HW) binds to the insulator element, interacting directly with an octamer motif present in the gypsy insulator. SU(HW) binds to approximately 200 other sites on Drosophila polytene chromosomes, possibly corresponding to endogenous insulators that play a role in the organization of the genome into higher order chromatin domains that allow for the normal regulation of gene expression.

yellow is one of the genes that has been used in studying the insulator properties of gypsy. yellow is a non-essential gene, lending pigmentation to various embryonic, larval and adult cuticular structures. yellow is regulated by a series of tissue-specific transcriptional enhancers (located in the 5' region of the gene). These enhancers regulate expression in the wings, body cuticle, and larval pigmented structures; a second series of intronic enhancers regulate expression in bristles and tarsal claws (Gdula, 1996 and references).

Insertion of gypsy in the 5' region of yellow inhibits the interaction between the upstream wing and body cuticle enhancers but does not affect the larval enhancer located proximal to the promoter. Mutation of su(Hw) results in a non-functional insulator. In flies with mutated su(Hw), upstream enhancers function even in the presence of insulator sequences between the enhancer and the proximal region of the gene. This suggests strongly that su(Hw) is required for the insulating properties of gypsy.

modifier of mdg4 was discovered during an analysis of mutations that alter the ability of su(Hw) to interfere with enhancer-promoter interactions. Mutations in mod(mdg4) cause an enhancement of the yellow phenotype, resulting in flies in which every cuticular structure of the larva and adult is unpigmented. This phenotype can be interpreted as a consequence of the inactivation of the yellow promoter itself, or inactivation of every enhancer controling yellow expression.

Cloning of mod(mdg4) revealed the structure of the protein for which it codes. MOD(MDG4) protein contains a BTB domain in the amino terminal region, a motif found in GAGA protein (known as Trithorax-like), a protein that plays a role in chromatin remodeling. MOD(MDG4) and SU(HW) demonstrate a direct physical interaction. MOD(MDG4) protein lacks a recognizable DNA binding domain and is unable to bind to insulator sequences of gypsy. Nevertheless MOD(MDG)4 is present on polytene chromosomes, in 2-3 times the number of bands containing SU(HW), suggesting that MOD(MDG4) might also interact with other proteins. In fact, mutation of mod(mdg4) is lethal, suggesting thaat MOD(MDG4) has a broad function. In the absence of MOD(MDG4), gypsy loses its uni-directional effects on distal enhancers and instead silences both upstream and downstream enhancers. Thus the function of MOD(MDG4) appears to be establishment of uni-directionality on the gypsy insulator. A mutated mod(mdg4) does not act as simply as a null allele, but instead causes a variegated phenotype similar to those caused by juxtaposing active genes to heterochromatic sequences. The resemblence of mod(mdg4) mutations to position-effect variegation, suggests that the effects of SU(HW) and MOD(MDG4) are mediated by changes in chromatin and not simply by classical transcription factor mediated changes in gene activation (Gdula, 1996).

There is a paradox with respect to SU(HW) function that begs for a resolution. On one hand the SU(HW)-MOD(MDG4) interaction suggests a chromatin based mechanism of gene activation and silencing similar to that mediated by Trithorax - and Polycomb-group genes. On the other hand, analysis of enhancers insulated from distal promoters by the gypsy element reveals that these enhancers can still function to regulate genes not protected by the insulator (Cai, 1995 and Scott, 1995). Paradoxically, the enhancer regions themselves are still functional, even if they are insulated form certain promoters by the gypsy element. The continued ability of enhancers to function, suggests that enhancers are not subject to chromatin mediated silencing. The resolution of this paradox will come from a biochemical study of the effects of SU(HW) and MOD(MDG4) chromatin structure. The generality of the role of chromatin in boundary element functioning is illustrated by the characterization of another boundary element associated factor, BEAF-32, which binds and contributes to boundary function at the scs' element of the hsp70 gene (Zhao, 1995).

How does the suHw insulator function to block enhancer-promoter interactions? Using divergently transcribed reporter genes in transgenic Drosophila embryos, it has been shown that an enhancer blocked from the downstream promoter by suHw is fully competent to activate an upstream promoter. To probe the insulator mechanism, the effect of suHw copy number on its insulator strength was tested in Drosophila embryos. The zerknullt enhancer VRE (ventral repression element) has been shown to be partially blocked by suHw. In blastoderm embryos, the V2 transgene containing VRE and E2, an evenskipped stripe 2 enhancer, directs reporter expression in a composite pattern of broad dorsal activation and dominant ventral repression of the E2 stripe. A single 340-bp suHw insulator element in the VS2 transgene partially blocks the upstream VRE enhancer. Two tandem suHw elements (arranged as direct repeats) were inserted between VRE and E2, resulting in VSS2. Instead of enhanced blockage, VSS2 embryos exhibit a loss of suHw insulator activity. This was observed in most VSS2 embryos and in all 10 independent VSS2 lines, indicating that it is unlikely to be caused by chromosomal position effects. Genomic polymerase chain reaction (PCR) analysis of independent VS2 and VSS2 lines further verifies the structural integrity of the transgenes in vivo (Cai, 2001).

To determine whether the loss of insulator function in VSS2 embryos is enhancer-specific, transgenes were constructed using a rhomboid neuroectodermal enhancer (NEE) and a hairy stripe 1 enhancer (H1). The NLH embryos containing NEE and H1 enhancers separated by a 1.4-kb neutral spacer (L) exhibit composite lacZ pattern directed by both enhancers. A single suHw element in the NSH transgene blocks the upstream NEE enhancer, whereas two tandem suHw elements (NSSH) do not block the NEE enhancer. A second group of transgenes uses a twist mesoderm enhancer (PE) and an evenskipped stripe 3 enhancer (E3). Both enhancers are active when separated by the L spacer (PL3). Insertion of a suHw element in the PS3 transgene blocks the upstream PE enhancer, whereas two tandem suHw elements do not block the PE enhancer. Replacing one of the two suHw elements in PSS3 with a spacer of comparable size (A) restores the enhancer-blocking activity of the remaining suHw in PSA3 embryos, indicating that loss of insulator activity with two suHw elements is not due to the spacing change but to the presence of the additional insulator. These results suggest that the loss of insulator activity with tandemly arranged suHw is independent of the enhancer tested (Cai, 2001).

The enhancer-blocking activity of suHw may require its interaction with other sites (or insulators) within the nucleus. A second suHw nearby may compete dominantly for the existing suHw and affect the neighboring enhancer-promoter interactions, depending on the cis arrangement of these elements. To test this hypothesis, the SVS2 transgene was constructed in which the VRE enhancer is flanked by two suHw elements. In contrast to the loss of insulator function seen in VSS2 embryos, the VRE enhancer is more effectively blocked in SVS2 embryos than in VS2 embryos. Thus, it is the tandem arrangement rather than physical proximity that causes the loss of insulator activity. VRE-mediated dorsal activation of the divergently transcribed miniwhite is also diminished in SVS2 embryos, indicating that VRE is blocked from promoters on either side. suHw-mediated blockage of VRE is significantly reduced in SVS2/mod(mdg4)^u1 embryos, indicating that a MOD(MDG4)-mediated complex is required for the enhanced insulator activity. VSS2, NSSH, and PSS3 transgenes were also examined in a mod(mdg4)^u1 background, and no change in the staining patterns was seen (Cai, 2001).

Insulators block enhancer-promoter interactions only when positioned between them. How can this occur without inactivating the enhancer or promoter? It has been hypothesized that insulators may interact with each other to form chromatin loop domains, restricting interactions among neighboring regulatory elements. The enhancer-blocking activity mediated by suHw is abolished when insulators are in tandem, and is enhanced when they flank the enhancer. Thus, suHw does not seem to block distal enhancers by locally capturing the enhancer complex or its associated proteins, because two tandem elements abolish rather than enhance the insulator function. Instead, a single intervening suHw insulator may interact with other insulators or chromosomal/nuclear sites, separating the enhancer and the promoter into topologically distinct chromatin domains. Two tandem suHw elements may preferentially interact with each other, excluding other interactions necessary to sequester the enhancer from the promoter, and may even augment the enhancer-promoter interaction by 'looping out' the intervening DNA. In contrast, suHw elements flanking an enhancer may readily interact as a result of their proximity, leading to better blockage of the enhancer. Loss of insulator function was seen when the distance between the two tandem suHw elements is 50, 150, and 170 bp (VSS2, PSS3, and NSSH, respectively). It has also been observed with spacers ranging from 200 bp to 5 kb in length. Therefore, it is unlikely to be caused by nonspecific steric hindrance due to the close juxtaposition of the insulators. DNA looping has been observed between interacting regulatory elements as close as 100 bp apart. Insulator assembly may induce alternative chromatin structure, resulting in DNA bending or nuclease-hypersensitive sites, which often indicate nucleosome-free DNA, to facilitate loop formation. Insulators or chromatin boundaries are frequently found in multiple copies, flanking enhancers or the genetic locus they regulate, such as the scs and scs' elements, the Mcp-1 and Fab boundaries, and the chicken ß-globin 5' and 3' boundaries. Selective interactions between neighboring insulators may regulate the access of tissue-specific enhancers to target promoters by forming alternative chromatin loop domains. It is conceivable that these domains not only block inappropriate enhancers but also facilitate interaction between distant enhancers and the target promoter (Cai, 2001).

The role of the Suppressor of Hairy-wing insulator protein in Drosophila oogenesis

The Drosophila Suppressor of Hairy wing [Su(Hw)] insulator protein has an essential role in the development of the female germline. This study investigated the function of Su(Hw) in the ovary. Su(Hw) is universally expressed in somatic cells, while germ cell expression is dynamic. Robust levels accumulate in post-mitotic germ cells, where Su(Hw) localization is limited to chromosomes within nurse cells, the specialized cells that support oocyte growth. Although loss of Su(Hw) causes global defects in nurse cell chromosome structure, it was demonstrated that these architectural changes are not responsible for the block in oogenesis. Connections between the fertility and insulator functions of Su(Hw) were investigated through studies of the two gypsy insulator proteins, Modifier of (mdg4)67.2 (Mod67.2) and Centrosomal Protein of 190kDa (CP190). Accumulation of these proteins is distinct from Su(Hw), with Mod67.2 and CP190 showing uniform expression in all cells during early stages of oogenesis that diminishes in later stages. Although Mod67.2 and CP190 extensively co-localize with Su(Hw) on nurse cell chromosomes, neither protein is required for nurse cell chromosome development or oocyte production. These data indicate that while the gypsy insulator function requires both Mod67.2 and CP190, these proteins are not essential for oogenesis. These studies represent the first molecular investigations of Su(Hw) function in the germline, which uncover distinct requirements for Su(Hw) insulator and ovary functions (Baxley, 2011).

Thus, during oogenesis somatic cells uniformly express Su(Hw). In contrast, germ cell accumulation is temporally and spatially regulated. Robust accumulation of Su(Hw) begins upon formation of egg chambers, where Su(Hw) localization is restricted to nurse cell (NC) nuclei and is absent from the oocyte nucleus, a distribution that is maintained throughout oogenesis. Other nuclear factors do not show dynamic expression in the germarium, suggesting that regulated accumulation of Su(Hw) may be important for its role in oogenesis (Baxley, 2011).

Phenotypic defects caused by loss of Su(Hw) were investigated through studies of ovaries obtained from females carrying different heteroallelic combinations of su(Hw) mutant alleles. These investigations demonstrate that NC chromosome development is delayed, but not blocked as previously reported. Structural defects in NC chromosomes appear to be independent of known genes involved in NC chromosome development, because transcription of these genes was largely maintained in su(Hw) null backgrounds. Finally, these studies addressed the long-standing hypothesis that the sterility in su(Hw) mutant females is caused by retention of the five-blob chromosome state that affects ribosome biogenesis, oocyte growth, and activates apoptosis. This study demonstrated that rRNA processing occurs normally in su(Hw) mutants, suggesting that ribosome biogenesis is not impaired. Further, it was discovered that NC chromosomes never disperse in ovaries obtained from fertile su(Hw)^f/v females, even though oocyte growth, rRNA processing and fecundity are wild type. These observations establish that decondensation and dispersal of NC chromosomes is not the cause of sterility in su(Hw) mutants (Baxley, 2011).

The uncoupling of the NC structural defects and Su(Hw) dependent fertility raises the question of which cells require Su(Hw) function for completion of oogenesis. At present, it is not known whether the essential function resides in NCs or the surrounding follicle cells, as these latter cells provide signals to the germline needed for egg chamber development. Additional investigations are required to resolve this issue (Baxley, 2011).

Mod67.2 and CP190 are BTB-domain proteins that are required for enhancer blocking by the gypsy insulator. The role of these proteins in oogenesis was investigated, to gain insights into the connections between the insulator and fertility functions of Su(Hw). It was found that during oogenesis Mod67.2 and CP190 show parallel accumulation, with these BTB-domain proteins found in all somatic and germ cells, including cells that lack Su(Hw) such as the oocyte nucleus. In early egg chambers, Mod67.2 and CP190 extensively co-localize with Su(Hw), while older egg chambers display diminished levels. NC chromosome association of Mod67.2 and CP190 is largely dependent on Su(Hw), although both proteins retain NC chromosome binding in su(Hw) mutants. These findings are consistent with genome-wide studies of protein binding in somatic cells that show that chromosome association of Mod67.2 and CP190 does not always overlap with Su(Hw) binding sites (SBSs) (Baxley, 2011).

Extensive co-localization of Su(Hw), Mod67.2 and CP190 is present in stages of oogenesis where the su(Hw) mutant phenotype becomes evident. Even so, null or nearly null mod(mdg4) and Cp190 single and double mutant females lay eggs of normal size. No evidence was observed for defects in NC chromosome development or increased apoptosis, implying that Mod67.2 and CP190 are not required for oogenesis. These observations imply that the fertility and insulator functions of Su(Hw) are different. Such findings may be explained if lower levels of BTB domain proteins are needed for oogenesis than are needed to establish an insulator. It is noted that while mod(mdg4) allele was used that fails to produce any of the Mod67.2 isoform, the Cp190 heteroallelic combinations studied were hypomorphic, because null alleles are pharate lethal. As such, Cp190 mutant ovaries may have enough CP190 activity to support Su(Hw) functions. However, CP190 null embryos have been generated from germline clones, implying that oogenesis is not blocked when CP190 is absent from germ cells. These data indicate that Mod67.2 and CP190 are not essential for oogenesis. It is predicted that Su(Hw) has Mod67.2 and CP190 independent functions. Support for this postulate comes from genome-wide studies that demonstrate that ~ 50% of SBSs do not bind Mod67.2 or CP190 (Baxley, 2011).

The different requirements for Su(Hw) in insulation and fertility raise the question of whether the essential role of Su(Hw) in oogenesis involves formation of chromatin insulators. It is predicted that if insulator function is involved, then novel interaction partners may be required for Su(Hw) to demarcate chromatin domains. Alternatively, Su(Hw) function may extend beyond that of an insulator protein, a possibility that is supported by recent genome wide studies of SBSs. The vast majority of non-gypsy SBSs contain a single motif, in contrast to the twelve Su(Hw) binding motifs found in the gypsy insulator. This observation is striking considering that enhancer blocking by the gypsy insulator requires at least four tightly spaced SBSs. Direct tests of the insulator activity of individual SBSs in transgene assays have shown that ~ 40% block enhancer action, suggesting that not all SBSs are insulators. If the formation of chromatin insulators by Su(Hw) is not required for fertility, then how does this protein contribute to nuclear functions during oogenesis? It is possible that Su(Hw) has the capacity to directly to modulate transcription of target genes. For example, studies of the function of one SBS revealed that this SBS was required for activation of transcription of the adjacent gene. Further, a repressor activity is suggested by genome-wide studies that correlate Su(Hw) localization with repressive chromatin and gene silencing. Interestingly, diverse regulatory functions have been documented for the major vertebrate insulator protein, CCCTC binding factor (CTCF). While CTCF is best known as an insulator protein, early studies of CTCF documented direct involvement in transcriptional activation and repression. More recent genetic studies in transgenic mice provide additional support for direct regulation of gene expression. These observations suggest that Su(Hw) may be similar to CTCF, functioning as a multi-faceted transcriptional regulator (Baxley, 2011).

Active chromatin and transcription play a key role in chromosome partitioning into topologically associating domains

Recent advances enabled by the Hi-C technique have unraveled many principles of chromosomal folding that were subsequently linked to disease and gene regulation. In particular, Hi-C revealed that chromosomes of animals are organized into Topologically Associating Domains (TADs), evolutionary conserved compact chromatin domains that influence gene expression. Mechanisms that underlie partitioning of the genome into TADs remain poorly understood. To explore principles of TAD folding in Drosophila melanogaster, Hi-C and PolyA+ RNA-seq was performed in four cell lines of various origins (S2, Kc167, DmBG3-c2, and OSC). Contrary to previous studies, this study found that regions between TADs (i.e. the inter-TADs and TAD boundaries) in Drosophila are only weakly enriched with the insulator protein dCTCF, while another insulator protein Su(Hw) is preferentially present within TADs. However, Drosophila inter-TADs harbor active chromatin and constitutively transcribed (housekeeping) genes. Accordingly, it was found that binding of insulator proteins dCTCF and Su(Hw) predicts TAD boundaries much worse than active chromatin marks do. Interestingly, inter-TADs correspond to decompacted interbands of polytene chromosomes, whereas TADs mostly correspond to densely packed bands. Collectively, these results suggest that TADs are condensed chromatin domains depleted in active chromatin marks, separated by regions of active chromatin. The mechanism of TAD self-assembly is proposed based on the ability of nucleosomes from inactive chromatin to aggregate, and lack of this ability is found in acetylated nucleosomal arrays. Finally, this hypothesis is tested by polymer simulations, and it was found that TAD partitioning may be explained by different modes of inter-nucleosomal interactions for active and inactive chromatin (Ulianov, 2015).

Recently developed 3C-based methods coupled with high-throughput sequencing have enabled genome-wide investigation of chromatin organization. Studies performed in human, mouse, Drosophila, yeasts, Arabidopsis and several other species have unraveled general principles of genome folding. Chromosomes in mammals and Drosophila are organized hierarchically. At the megabase scale, mammalian chromosomes are partitioned into active and inactive compartments. At the sub-megabase scale, these compartments are subdivided into a set of self-interacting domains called Topologically Associating Domains (TADs); TADs themselves are often hierarchical and are split into smaller domains. Similar to mammals, Drosophila chromosomes are partitioned into TADs that are interspaced with short boundaries or longer inter-TAD regions (inter-TADs) (Ulianov, 2015).

Partitioning of mammalian genomes into TADs appears to be largely cell-lineage independent and evolutionary conserved. Disruption of certain TAD boundaries leads to developmental defects in humans and mice. TADs correlate with units of replication timing regulation in mammals and colocalize with epigenetic domains (either active or repressed) in Drosophila. The internal structure of TADs was reported to change in response to environmental stress, during cell differentiation, and embryonic development. In addition, comparative Hi-C analysis has demonstrated that genomic rearrangements between related mammalian species occur predominantly at TAD boundaries. Consequently, TADs appear to evolve primarily as constant and unsplit units. Previous studies in Drosophila embryonic nuclei and embryo-derived Kc167 cells detected TADs of various sizes roughly corresponding to epigenetic domains. Additionally, long-range genomic contacts and clustering of pericentromeric regions were revealed, and TAD boundaries were found to be enriched with active chromatin marks and insulator proteins. Both active and inactive TADs were identified, and their spatial segregation was observed (Ulianov, 2015).

Despite extensive studies, mechanisms underlying TAD formation remain obscure. Architectural proteins, including cohesin and CTCF, are often found at TAD boundaries; thus, they have been proposed to play a key role in the demarcation of TADs. However, several studies suggest that other mechanisms may be responsible for partitioning and formation of TADs. Firstly, depletion of various insulator proteins did not affect the profile of chromosome partitioning into TADs, but rather decreased intra-TAD interactions. Secondly, CTCF may mediate loops that occur between the start and the end of the so-called 'loop domains'. However, domains of similar sizes but without a loop were observed as well (so-called 'ordinary domains'. Thirdly, polymer simulations of a permanent chromatin loop yield a noticeable interaction between the loop bases on a simulated Hi-C map, but without a characteristic square shape of a TAD. Loops of this kind are thought to occur between insulator proteins such as Su(Hw) in the 'topological insulation' model. Finally, chromosomal domains similar to TADs in the bacterium Caulobacter crescentus are demarcated by actively transcribed genes, and are not affected by the knockout of SMC, a homolog of cohesin subunits (Ulianov, 2015).

This study presents evidences that question the role of insulators in the organization of TAD boundaries in Drosophila . The results suggest that TADs are self-organized and potentially highly dynamic structures formed by numerous transient interactions between nucleosomes of inactive chromatin, while inter-TADs and TAD boundaries contain highly acetylated nucleosomes that are less prone to interactions. Finally, a polymer model of TAD formation is developed based on the two types of nucleosomes, and it was found that a polymer composed of active and inactive chromatin blocks forms TADs on a simulated Hi-C map (Ulianov, 2015).

This study and others (Hou, 2012; Sexton, 2012) revealed that boundaries and inter-TADs in Drosophila, as opposed to TADs, are strongly enriched with active chromatin and its individual marks, as well as with active transcription and with constitutively transcribed housekeeping genes. Consequently, active chromatin marks, in the simplest case only total transcription and H3K4me3 (a mark of active promoters), can relatively well predict a TAD/inter-TAD profile. The existence of long inter-TADs composed of active chromatin is per se an argument for the ability of this type of chromatin to separate TADs. Furthermore, the current observations demonstrate that the presence of active chromatin and transcribed regions within TAD undermines the TAD integrity making TAD less compact and generating weak boundaries inside TAD. Consequently, a bona fide TAD is inactive; TADs containing active chromatin become less dense, acquire weak internal boundaries and eventually split into smaller TADs that are composed of inactive chromatin. The observation that the majority of housekeeping genes are located within inter- TADs and TAD boundaries suggests that evolutionary conservation and cell-type independence of TAD/inter-TAD profiles may be explained by conservation of positions of housekeeping genes along the chromosomes (Ulianov, 2015).

It is noted that chromosomal interaction domains similar to TADs have been observed in the bacterium Caulobacter crescentus, where they are demarcated by sites of active transcription. Although the basic level of chromosomal folding is different in bacteria and eukaryotes, the model proposed in (Le, 2013) and the model stem from common principles. In Caulobacter, active transcription is thought to disrupt the fiber of supercoils (plectonemes) by creating a stretch of non-packaged DNA, free of plectonemes, which spatially separates chromosomal regions flanking it. In the model, transcription disrupts chromatin organization by introducing a 'non-sticky' region of chromatin, which is less compact and more unfolded in space, and thus spatially separates two flanking regions. Computer modeling shows that stickiness of non-acetylated (inactive) nucleosomes and the absence of stickiness for acetylated (active) nucleosomes are sufficient for chromatin partitioning into TADs and inter-TADs. Self-association of nucleosomes may be explained by the interaction of positively charged histone tails (in particular, the tail of histone H4) of one nucleosome with the acidic patch of histones H2A/H2B at an adjacent nucleosome. Acetylation of histone tails, which is typical of active chromatin, may interfere with inter-nucleosomal associations. In addition to a high level of histone acetylation, other features of active chromatin including lower nucleosome density in inter-TADs, manifested as the decreased histone H3 occupancy, might contribute to the generation of TAD profiles (Ulianov, 2015).

It should be mentioned that a significant difference between the polymer simulations and models previously suggested by the Cavalli and Vaillant groups (Jost, 2014) is the use of saturating interactions between inactive nucleosomes. In the case of volume interactions, all nucleosomes of the same type adjacent in 3D space will attract each other; in the case of saturating interactions, each molecule may attract only one neighbor. Using volume interactions leads to the formation of a single dense blob, and does not produce TADs in a simulated Hi-C map.It is noted that the saturating nature of interactions between nucleosomes is based on the known properties of nucleosomal particles. Previous studies considered a variety of mechanisms that may lead to the formation of TADs. In particular, Barbieri (2012) studied segregation of two TADs using cubic lattice simulations of a short 152-monomer chain consisting of two TADs, assuming that inter- monomer interactions could only form between monomers belonging to the same TAD. In the current model, this study shows that TADs emerge without requiring such specific interactions; any two regions of sticky monomers separated by a non-sticky linker would form TADs. Another study proposed that transcription-induced supercoiling may be responsible for the formation of TADs (Benedetti, 2014). Although this model is consistent with the current observation that sites of active transcription demarcate TAD boundaries, there is limited evidence that supercoiling of chromatinized DNA exists in Drosophila and other organisms. On the contrary, the current model is based on known biochemical properties of nucleosomes (Ulianov, 2015).

The fact that a minor fraction of TADs is built mostly from active chromatin apparently contradicts the current model, suggesting that additional ways of chromatin self-organization could exist. One possibility is the establishment of long-range contacts between enhancers and their cognate promoters, as well as loops between pairs of insulators. Such loops formed inside active unstructured chromatin linkers (i.e., inter-TADs) could probably be sufficient to compact them and thus to fold into TADs (Ulianov, 2015).

TAD profiles of X chromosomes are almost identical in the male and female cell lines, that is in agreement with recently published observations (Ramírez, 2015). Thus, it seems that hyperacetylation of male X-chromosomes due to dosage compensation does not generate new TAD boundaries. However, it should be noted that MOF histone acetyltransferase of the MSL complex introduces only the H4K16ac mark. Although this modification is important to prevent inter-nucleosomal interactions, acetylation at other histone positions and H2B ubiquitylation contribute as well. Additionally, H4K16 acetylation generated by the dosage compensation system occurs preferentially at regions enriched with transcribed genes and hence within inter-TADs (Ulianov, 2015).

The current analysis does not support the previously reported (Hou, 2012; Sexton, 2012) strong enrichments of insulator proteins Su(Hw) and dCTCF at TAD boundaries in Drosophila. To assess the possible reasons of this divergence, the dCTCF distribution was re-analyzed with respect to TAD positions in the current dataset using the raw ChIP-seq data. No strong difference was observed in the dCTCF coverage in TADs and inter-TADs. Interestingly, this study obtained the same result while analyzing dCTCF and Su(Hw) binding within TAD boundaries identified by Hou (2012). However, a strong enrichment of dCTCF at TAD boundaries was observed when the peak distribution was analyzed instead of read coverage. Additionally, the effect was much weaker when modENCODE peaks were used. Hence, the discrepancy may be caused by a different peak calling procedure in modENCODE and in Hou. (2012). The biological significance of these observations remains to be determined. It is noted that disruption of the cohesin/CTCF complex in mammals, as well as depletion of the Vtd (also known as Rad21) cohesin subunit in Drosophila, did not lead to disappearance of TAD boundaries, but rather only slightly decreased interactions inside TADs (in mammals) and reduced TAD boundary strength in the Drosophila genome. These observations favor a role for the cohesin/CTCF complex, which is known to form loops, in chromatin compaction inside the TADs (Ulianov, 2015).

Binding of insulator proteins might contribute to establishing TAD boundaries through introducing active chromatin marks. Indeed, when inserted into an ectopic position, a classical insulator triggers hyperacetylation of the local chromatin domain and recruits chromatin-remodeling complexes. However, absence of strong enrichment of dCTCF at TAD boundaries and preferential location of Su(Hw) inside TADs mean that at least dCTCF- and Su(Hw)-dependent insulators are not the major determinants of TAD boundaries and inter-TADs (Ulianov, 2015).

TADs are predicted based on the analysis of averaged data from a cell population. Although they are usually represented as large chromatin globules, direct experimental evidence for the existence of such globules in individual cells is controversial. Using confocal and 3D-SIM microscopy, ~1-Mb globular domains have been observed within chromosomal territories. However, using STORM microscopy, chromatin in individual mammalian cells has been found to be organized into 'clutches' composed of several nucleosomes, and that increased histone acetylation dramatically reduces size of these clutches. It is thus possible that sub-megabase TADs revealed by Hi-C represent a set of nucleosome clutches separated by relatively short spacers of various sizes. These short clutches may occupy various positions within TADs in different cells and stochastically assemble to form short-living aggregates. The stochastic nature of TADs is supported by computer simulations (Ulianov, 2015).

GENE STRUCTURE

Exons - 7

PROTEIN STRUCTURE

Amino Acids - 944

Structural Domains

SU(HW) contains 12 repeats of the C2H2 zinc finger motif. These zinc finger domains are located in the central portion of the protein. In addition, a highly acidic region, containing 48% Asp and Glu, is present between amino acids 154 and 202, immediately proceeding the first Zn finger motif (Parkhurst, 1988). A leucine zipper domain is found between the zinc finger domains and the C-terminal end of the protein (Harrison, 1993).

It is thought that su(Hw) forms discrete domains of gene activity by segregating promoters from enhancer elements through a change in chromatin organization. Functional domains of the su(Hw) protein were characterized that mediate the silencing effect of mod(mdg4) mutations. Two of three regions of su(Hw), regions B and C, that are located between the leucine zipper motif and the C-terminal acidic domain are conserved across Drosophila species and are necessary for both the unidirectional and bidirectional repression of transcription by su(Hw). These domains are implicated in an interaction with mod(mdg4) which is thought to mediate the unidirectional repression due to insulator function. In contrast, two acidic domains, the N-terminal acidic domain and the C-terminal acidic domain, dispensable for the unidirectional repression of enhancer elements, are critical for the bidirectional silencing of enhancer activity observed in mutants lacking functional mod(mdg4) protein. Bidirectional repression is thought to be due to changes in large blocks of chromatin structure (Gdula, 1997).

date revised: 23 June 2023

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.