To fit inside the nucleus of a eukaryotic cell, the genome must be highly compacted. This compaction is a barrier to processes that require access to the DNA, such as replication and transcription, and therefore is not uniform across the genome. Broadly, DNA is compartmentalized into regions known as A and B compartments, comprised of active and inactive chromatin, respectively. Further chromatin organization occurs within compartments in the form of topologically associated domains (TADs). TADs are segments of a chromosome, ranging from 50 kb to 2 Mb, that have an increased propensity for self-interaction and are defined by flanking boundary regions. Within TADs, genes and their associated enhancers are often nested into sub-TADs, fostering the interactions between enhancers and promoters that drive gene expression. Together this hierarchical organization of chromatin within three-dimensional (3D) space functions as a mechanism to regulate gene expression and replication (extensively reviewed in Ref. [1]).

While multiple technologies have helped to determine the 3D organization of chromatin within cells, the mechanisms underlying the establishment of this chromatin architecture are less well understood. Recent advances in single-nucleus genomic analyses and high-resolution microscopy have begun to shed light on our understanding of the dynamics of chromatin organization during development [2, 3, 4]. Because of the dramatic genomic reprogramming that occurs during early embryonic development, application of these technologies to the early embryo provided valuable insights into the dynamics of 3D chromatin structure and the biological processes driving organization in vivo. In this review, we summarize the dynamics of hierarchical chromatin structures during early embryogenesis in various animal models, highlight recent findings describing the molecular mechanisms driving 3D chromatin organization, and discuss the functional relevance of such architecture on the regulation of gene expression.

During early embryonic development, the genomes of the two highly specialized germ cells must come together and be reprogrammed to generate a totipotent embryo. This genomic reprogramming occurs in the absence of transcription and is controlled by maternally deposited factors. Transcription from the zygotic genome is gradually activated, allowing control of embryonic development to be progressively passed from mother-to-zygote in a process known as the maternal-to-zygotic transition (MZT). This developmental transition is widely conserved across the animal kingdom, although the timing of key events is variable.

In many externally fertilized embryos, early development is characterized by a series of incredibly rapid division cycles. Drosophila melanogaster embryogenesis begins with 13 rapid, synchronous nuclear divisions without cellular division. Transcriptional activation of the zygotic genome begins approximately 1-h post-fertilization (hpf) at about the eighth nuclear division (NC8) with widespread zygotic genome activation (ZGA) occurring at ∼2.5 hpf (NC14) (Figure 1a). Other externally fertilized organisms undergo ZGA on a similar timescale: 2–4 hpf in Danio rerio and 3–5 hpf in Xenopus species. By contrast, in mammals where the early division cycle is slower, ZGA occurs later but after fewer divisions: 10–24 hpf at the one- and two-cell stages in Mus musculus and 48–72 hpf at the four- and eight-cell stages in humans (Figure 1a). Because of the relative ease of experimentation, molecular studies of externally fertilized model systems have provided a foundation for our understanding of chromatin dynamics during this conserved period of genomic reprogramming. Nonetheless, recent technological advances have enabled studies in mammals, which have highlighted the conserved and divergent mechanisms regulating chromatin architecture during embryogenesis.

Because of the dramatic reprogramming that occurs during early development to establish the totipotent embryo, studies of this developmental period provide fundamental insights into how 3D chromatin architecture is established. Gametes have well-defined 3D chromatin organization. By contrast, after fertilization, 3D structure is largely absent and is gradually reestablished during the MZT. Hi-C studies in many organisms (flies, frogs, zebrafish, mice and humans) showed that the establishment of TADs coincides with ZGA (Figure 1b) [5, 6, 7, 8, 9∗]. Thus, across organisms, TAD organization is established concomitantly with transcriptional activation of the zygotic genome.

Whereas TAD formation is concurrent with widespread ZGA, the timing of euchromatic and heterochromatic A/B compartmentalization is more variable, ranging from pre-ZGA to post-ZGA (Figure 1c and d). In mice, separation into A and B compartments is evident before ZGA, although the timing of this organization is dependent on the parental origin of the chromosome [10,11]. Distinct active and inactive compartments emerge during NC14 in Drosophila embryos [5,6,12]. Chromatin in other organisms, such as Xenopus, zebrafish, and humans, does not appear to undergo compartmentalization until after ZGA: 12hpf in Xenopus tropicalis, 24hpf in zebrafish, and at the morula stage (three days post-fertilization) in humans [7, 8, 9∗]. By leveraging the reprogramming of the early embryo, these studies clearly distinguish between chromatin structure at the level of TADs and compartmentalization.

Identification of the dynamics of sub-TAD structures were constrained by the resolution of Hi-C experiments. Recent development of Micro-C, improvements to the algorithms used to analyze Hi-C data, and the advent of microscopy-based approaches to visualize intra-chromosomal contacts, such as Hi-M, have provided the resolution to detect these small–scale interactions [3,4,13]. In Drosophila embryos, Hi-M revealed that multi-way interactions between cis-regulatory elements inside a single TAD are evident prior to emergence of global TAD structure [13,14]. The increased resolution provided by Micro-C as compared to Hi-C enabled the genome-wide identification of a novel type of cis-regulatory elements – tethering elements – that are located within TAD boundaries and promote enhancer-promoter contacts (Figure 2) [15,16]. These were identified in Drosophila embryos at NC14, although the timing of when tethering element interactions initiate is unknown. As technologies continue to advance, their application to early embryogenesis will contribute to our understanding of how sub-TAD chromatin contacts are established and how they contribute to overall 3D chromatin organization.

The seemingly synchronized timing of transcriptional activation and TAD establishment suggest a possible coordination between these two events. However, studies in Drosophila and mice demonstrate that transcriptional state does not correlate with TAD organization nor does inhibition of transcription during ZGA prevent TAD formation [2,5,10,11,17]. However, suppression of transcription in Drosophila cell culture models derived from later developmental stages results in modest-to-considerable changes in TAD organization and border strength, suggesting developmental timing is an additional influence on the relation between transcription and TAD structure [18,19]. In contrast to what has been shown in Drosophila and mouse embryos, preventing transcription in human embryos prior to the eight-cell stage leads to reduced TAD formation [7]. In addition to transcription, it is also possible that DNA replication may play a role in driving TAD formation. Inhibiting replication in two-cell mouse embryos hinders TAD formation, although additional studies in other organisms are needed to clarify how DNA replication during ZGA contributes to higher-order chromatin organization [10].

Because transcription does not appear to be required for TAD formation in the early embryo, it is likely that other events promote TAD formation. In all organisms studied to date, ZGA is driven by pioneer transcription factors, which are specialized transcription factors that promote chromatin accessibility. Thus, these factors may contribute to the establishment of the 3D genome. In Drosophila embryos, Zelda (Zld) and GAGA factor (GAF) are pioneer factors that drive reprogramming during ZGA [20, 21, 22]. Both factors are enriched at TAD boundaries as the 3D architecture becomes established during ZGA (Figure 2) [5,6]. While depletion of Zld during the MZT does not cause a global loss of TAD organization, individual TAD structures are disrupted, suggesting Zld may be important for the establishment of specific TADs [5]. Both pioneer factors are also bound to tethering elements and may therefore function in promoting enhancer-promoter interactions (Figure 2) [15]. Indeed, high resolution analysis of Zld-depleted embryos by Hi-M revealed that Zld is required for multi-way interactions between enhancers and promoters within a TAD [14]. Maternal depletion of GAF in Drosophila embryos does not significantly impact TAD formation, but Micro-C revealed a role for GAF in tethering enhancers and promoters [23,24]. Because of the conserved role of pioneer factors in driving ZGA, more detailed analyses of their role in promoting 3D chromatin organization in other organisms will help to determine the broader role of these essential proteins in reprogramming the 3D genome.

TAD boundaries are demarcated by insulator-binding proteins. In most organisms, convergently oriented CCTCF-binding factor (CTCF) binding sites flank the TAD (Figure 2). This CTCF orientation facilitates extrusion of the intervening DNA by cohesin ring complexes [25, 26, 27, 28]. Depletion of CTCF during early embryogenesis in zebrafish, frogs, mice, and humans causes a reduction in TAD formation [7, 8, 9∗,29,30]. Chromatin architecture is also disrupted in zebrafish embryos and mouse embryonic stem cells lacking cohesin complex components, supporting the cooperative function of CTCF/cohesin in establishing 3D chromatin organization in many organisms [31,32]. This does not appear to be conserved in Drosophila whose TAD borders lack convergent orientation binding by CTCF. Knockdown of maternal CTCF in flies does not perturb embryonic development [33,34]. This difference between Drosophila and other species may be due, in part, to the existence of additional insulator-binding proteins, including BEAF-32, Chromator, and CP190. In 2–6 h embryos depleted of CP190, a subset of TAD boundaries is lost [35]. In contrast to the majority of boundaries, which overlap with promoters, these CP190-dependent boundaries are not at promoters [35,36]. Instead, transcription influences the formation of boundaries at promoters [34]. Together these studies demonstrate contributions for both insulator-binding proteins and transcription to TAD formation [34,35]. By contrast, depletion of these factors in Drosophila cell culture models alters genome-wide TAD structure, positing a function of insulator-binding proteins in the maintenance of TAD organization [37]. Thus, much remains to be determined about the mechanisms that establish TAD formation during early embryonic reprogramming.

While much also remains unknown regarding the forces that drive formation of the A and B compartments, the early embryo is a powerful system to investigate these mechanisms. The formation of the silenced heterochromatic B compartment is not evident at the very initial stages following fertilization and is gradually established over the MZT (Figure 1c and d) [38, 39, 40, 41, 42, 43]. One compelling hypothesis is that these compartments are partitioned by phase separation [44]. Phase separation is driven by weak multi-valent interactions between macromolecules that can lead to partitioning within a shared volume. While chromatin itself is capable of phase separation, proteins and transcription factors also are likely to play important roles [45]. For example, formation of the inactive B compartment in Drosophila embryos requires heterochromatin protein 1a (HP1a) (Figure 1d) [12]. HP1a from both Drosophila and humans can phase separate, and this property has been implicated in driving heterochromatin formation [46,47]. By contrast, mouse HP1a shows conflicting phase separation abilities when surveyed in vitro and in vivo, highlighting the need for additional mechanistic studies [48].

In addition to phase-separation of heterochromatin, active euchromatin is also organized in 3D space. Transcription often occurs in localized regions that are visualized as hubs of high transcription factor concentration (Figure 2b) [49, 50, 51, 52, 53]. In early Drosophila embryos, transcription factors, including Zld, Bicoid (Bcd), and Ultrabithorax (Ubx), form dynamic subnuclear condensates that are enriched for other activating factors important for facilitating temporal gene expression during development [51,54, 55, 56, 57]. Mediator and RNA polymerase II complexes colocalize to these hubs, both of which have also been shown to exist in droplets [50,58,59]. Similarly, the transcription factor Nanog initiates the formation of transcription bodies at the mir-430 locus in zebrafish embryos [60]. In addition to these proteins, transcription itself may function to organize this active nuclear compartment. Studies in early zebrafish embryos, prior to the formation of heterochromatin, demonstrated that the RNA resulting from active transcription displaced transcriptionally inactive chromatin and in so doing could promote the formation of distinct active domains [61]. Together these data from early embryos and cell culture implicate phase separation and the regulation of these multivalent interactions in promoting the 3D structure of the active compartment, but much remains to be investigated regarding the functional relevance.

Through decades of study, it is clear that the organization of chromatin into a 3D architecture is a feature shared amongst the animal kingdom. Despite the ubiquity of this observed organization, the functional contribution to gene expression remains ambiguous. During development, tissue-specific networks of enhancer-promoter interactions drive gene expression to promote temporal and spatial expression of tissue-specific genes. Early studies proposed that TADs facilitate these interactions by bringing cis-regulatory regions in proximity with target promoters. However, TADs are relatively similar across developmental stages and tissues. This suggests that TADs are not a major regulator of tissue-specific gene expression patterns, but instead may have locus-specific requirements [3,62]. Nonetheless, single-cell, imaging-based studies identified cell-type specific domain structures at distinct loci in the Drosophila embryo, leaving open the possibility that developmental regulation may control 3D genome architecture to different degrees depending on the genomic region [63]. Studies in early Drosophila embryos demonstrated that tissue-specific gene expression does not coincide with differences in local or global chromatin organization, suggesting that 3D chromatin organization at this level is not a major regulator of gene expression [14,64]. Supporting this disconnect between TAD structure and gene expression, alterations in TAD organization caused by chromosome rearrangements in Drosophila or by targeted mutation in developing mouse limb buds did not produce significant differences in gene expression [65,66]. Disruption of individual insulators or insulator binding proteins in Drosophila embryos cause TAD fusions, but minimal gene expression changes [15,34,35]. These data support a model whereby TADs function to limit promiscuous enhancer-promoter interactions as opposed to actively promoting interactions within a TAD.

In contrast to the relatively minor effects of TADs on tissue-specific gene expression, it is possible that 3D organization at the level of transcriptional hubs and enhancer-promoter contacts may have larger effects on the regulation of gene expression. Disruption of transcription bodies during ZGA in zebrafish results in the increase in expression of genes outside the hub, suggesting that these transcription bodies broadly regulate gene expression by sequestering components of the transcriptional machinery [67] Deletions in tethering elements, which function to bring enhancers and promoters together, resulted in delayed transcriptional activation [15,16]. In the early Drosophila embryo, these tethering elements are bound by pioneer factors that are known to localize in concentrated hubs, suggesting that multivalent interactions may function to facilitate coordinated transcription from these hubs [15,16].