Revealing chromatin organization in metaphase chromosomes

Abstract

The local structural organization of chromatin in mitotic chromosomes is not well understood. A new cryo‐electron tomography study from the Daban laboratory (Chicano et al, 2019) reveals that mitotic chromatin isolated from human cells can assume a plate‐like fine structure, containing layers of interdigitated nucleosomes. Such a multilayered organization suggests a possible model for mitotic chromatin compaction.

Genome sequences on eukaryotic chromosomes are packaged into chromatin, a nucleoprotein complex in which arrays of nucleosomes organize the genomic DNA. During interphase, chromosomes occupy distinct territories within eukaryotic cell nuclei. Chromosome conformation capture (3C) methods have been instrumental at revealing principles of long‐range organization within chromatin in vivo. At the DNA megabase (Mb) scale, the chromatin polymer exhibits a fractal organization into clusters of self‐interacting domains (Lieberman‐Aiden et al, 2009). At the 100 kilobase (kb) scale, distinct domains called topologically associating domains (TADs) can be identified. These TADs are demarcated by insulator complexes, exhibit high degrees of self‐interaction, and often contain co‐regulated genes (Dixon et al, 2012). Finally, at even shorter‐range, single‐kilobase scales, individual nucleosomes self‐associate through histone–protein contacts (Hsieh et al, 2015; Risca et al, 2017). This results in the formation of chromatin secondary structure, such as segments of two‐start (or zig‐zag) fiber structure (Dorigo et al, 2004).

During mitosis, chromosome structure is drastically remodeled. Sister chromatids dissociate and individual chromosomes form elongated, highly compacted structures. Transcription is strongly repressed and, beyond a subset of “bookmarking” transcription factors and chromatin architectural proteins, most chromatin binding proteins dissociate. Genomic studies based on 3C methodologies have revealed that most long‐range structure, such as TADs, is dissolved (Gibcus et al, 2018). A model compatible with the experimental data proposes extrusion of chromatin loops from a central core of condensin and topoisomerase IIα proteins, giving rise to a “bottle‐brush”‐like structure of mitotic chromosomes (Fig 1A). Chromatin fibers within these loops are assumed to adopt a polymer melt‐like state, characterized by a high degree of interdigitation and self‐association. Conversely, cross‐linking studies of mitotic chromatin showed that nucleosome–nucleosome contacts corresponding to two‐start secondary structure are preserved during mitosis, but in addition also detected a higher prevalence of more long‐range contacts (Grigoryev et al, 2016). Therefore, local chromatin organization in mitotic chromosomes is still enigmatic and direct structural information is lacking.

Figure 1. Models of mitotic chromatin organization.

(A) In the scaffold/radial loop model, a core scaffold of condensin I and II complexes together with topoisomerase IIα compacts chromosomes by extrusion of nested chromatin loops, including long 400‐kb outer and 80‐kb inner loops (Gibcus et al, 2018). The chromatin fibers are locally disordered, forming a polymer melt‐like state containing interdigitated fibers of stacked nucleosomes. (B) In the multilayered plate model of mitotic chromatin organization, individual chromatin loops are organized into dense plates of nucleosomes (Chicano et al, 2019). Plates can interdigitate at contact sites and interact through nucleosome stacking interactions.

In their present study, Chicano et al (2019) employed cryo‐electron tomography (cryo‐ET) and small‐angle X‐ray scattering (SAXS) analyses to investigate mitotic chromatin structure. To this end, they carefully purified individual mitotic chromosomes from human cells. They then denatured the chromosomes under mild conditions (incubating in low ionic‐strength buffer and applying shear stress), while maintaining them at native Mg²⁺ concentrations. Under such conditions, thin and flexible plates emanated from the mitotic chromosomes, whose suspension in vitreous ice and cryo‐ET analysis allowed the authors to accurately assign their dimension and nanostructure. Most plates exhibited a large surface area, with dimensions of >1 μm in the x‐y plane and 0.2–0.4 μm along the z‐axis (limited by ice thickness). In contrast, their diameter was very thin, at around 7.5 nm. Considering the known dimensions of individual nucleosomes (with a diameter of 11 nm and height of 5.6 nm), these results imply that the observed plate structures may be composed of interacting, but tilted nucleosomes. In agreement with this notion, earlier results from the Daban group had shown that isolated chromatin fibers, obtained from micrococcal nuclease‐digested metaphase chromosomes, can self‐associate to form chromatin plates in vitro (Milla & Daban, 2012). Moreover, a subset of plates was found to be decorated with small particles. In‐depth tomographic analysis of those particles revealed that their dimensions and overall structure were compatible with the structure of individual nucleosomes. Isolated chromatin segments could also be recognized, exhibiting solenoidal structure over short length spans. Finally, while individual plates were discernable, certain regions of the tomogram contained multilayered assemblies resembling stacked plates. Intriguingly, plates in direct contact formed structures that appeared denser, as two stacked single layers—indicating that nucleosomes may be able to interdigitate to form interwoven structures (Fig 1B).

Due to the dense architecture of the observed plate structures, cryo‐ET could not reveal internal chromatin structure. Thus, the authors performed synchrotron SAXS measurements on whole, purified mitotic chromosomes under conditions that favor internucleosomal interactions and preserve chromatin compaction. In all measured SAXS profiles, a defined peak corresponding to structural features of 6 nm was detectable. This 6‐nm peak corresponds to the distance between nucleosomes that are engaged in a face‐to‐face stacked conformation. This finding indicates the prevalence of internucleosomal stacking interactions, compatible with both local two‐start structural elements and plate‐like conformations. Note that the content of classical long 30‐nm chromatin fibers is low in these structures, as indicated by a lack of SAXS signal at 30 nm. Together, the cryo‐ET and SAXS measurements thus support a view of mitotic chromosomes being organized as layered stacks of plate‐like arrangements of nucleosomes. Individual nucleosomes are laterally associated and engage in interdigitation and face‐to‐face interactions, in particular at contact regions between plates (Fig 1B). The proposed structure of mitotic chromosomes as stacks of connected plates is attractive, as it can explain previously observed chromosome banding patterns. Moreover, nucleosome density values obtained from this work are comparable to expected nucleosome densities from loop‐based polymer models (Gibcus et al, 2018).

Still, the current cryo‐ET study has been performed on isolated mitotic chromosomes following denaturation. The question therefore remains whether the observed structures are also prevalent in vivo. Recently, the Gan group applied cryo‐ET of macromolecular structures in their native cellular environment to mitotic chromatin of Schizosaccharomyces pombe (Cai et al, 2018). In that case, the authors were not able to detect higher‐order organization, but instead found that many nucleosomes were partially unwrapped. Moreover, they observed high but uneven crowding and compaction in mitotic chromatin, indicating a dynamic state. Another recent study by the O'Shea laboratory developed a method (chromEMT) combining labeling of chromatin with heavy metals and electron microscopy tomography to visualize chromatin in situ in cells (Ou et al, 2017). When imaging mitotic chromatin using chromEMT, they detected higher chromatin density compared to interphase chromosomes. In addition, individual chromatin fibers were found to adopt a disordered but compact state, as could, e.g., be caused by folding back onto themselves.

At this point, it is not clear why plate‐like chromosome organization was observed in isolation but not detected in the in situ studies. Several options are conceivable: First, it is possible that ordered multilayered states co‐exist with more disordered regions within mitotic chromosomes, due to spatial heterogeneity. Indeed, the presence of mitotic bookmarking factors and other chromatin binding proteins within mitotic chromosomes are expected to induce local structural disorder. In contrast, a crystalline plate‐like state might not allow chromatin access and seems incompatible with chromatin factor binding. Second, the appearance of highly structured states in metaphase chromosomes might be limited to a relatively short time window when chromosome compaction is maximal. Third, the in vitro investigation by Chicano et al (2019) requires sample isolation and preparation. It is possible that these procedures promote the formation of semi‐crystalline sheets from high‐density polymer melt‐like chromatin states. At this point, more research is thus required to differentiate between these different hypotheses.

In summary, the study of isolated chromatin by Chicano et al (2019) strengthens a model of organization of mitotic chromosomes as stacks of dense plate‐like structures, composed of interacting, intercalated nucleosomes. Importantly, diffraction measurements highlight the prevalence of nucleosome stacking interactions within mitotic chromatin. Taken together, the chromatin layer model is able to explain key features of mitotic chromatin organization, including the nucleosome density as well as defined banding pattern. Whether such states persist in living cells within the dynamic environment of a dividing cell is, however, still not known. A future challenge in this field is thus to improve resolution and chromatin tracing abilities in tomography or super‐resolution microscopy experiments to directly reveal the structure of defined chromatin region within the cellular environment.

The EMBO Journal (2019) 38: e101699