| Abstract: | Cryoelectron tomography of the cell nucleus using scanning transmission electron microscopy and deconvolution processing technology has highlighted a large-scale, 100- to 300-nm interphase chromosome structure, which is present throughout the nucleus. This study further documents and analyzes these chromosome structures. The paper is divided into four parts: 1) evidence (preliminary) for a unified interphase chromosome structure; 2) a proposed unified interphase chromosome architecture; 3) organization as chromosome territories (e.g., fitting the 46 human chromosomes into a 10-μm-diameter nucleus); and 4) structure unification into a polytene chromosome architecture and lampbrush chromosomes. Finally, the paper concludes with a living light microscopy cell study showing that the G1 nucleus contains very similar structures throughout. The main finding is that this chromosome structure appears to coil the 11-nm nucleosome fiber into a defined hollow structure, analogous to a Slinky helical spring [https://en.wikipedia.org/wiki/Slinky; motif used in Bowerman et al., eLife 10, e65587 (2021)]. This Slinky architecture can be used to build chromosome territories, extended to the polytene chromosome structure, as well as to the structure of lampbrush chromosomes.A recent publication introduced iterative deconvolution for scanning transmission electron microscopy (TEM) tomograms of cryopreserved cellular structures (ref. 1 and references therein). Micron-thick areas of the vitrified cells were accessible without prior cryosectioning or lamella preparation. The deconvolution computation simplified interpretation of the tomograms by substantially filling the missing wedges of information that result from incomplete tilts. The effect was a substantial improvement in resolution along the depth (Z) direction. This technology made it possible to assess a tomogram from an area of the nucleus intact, in which large-scale interphase chromosome structures were noted (1).Here, the chromosome structures observed in these nuclear tomograms are further documented and analyzed. This paper is divided into four parts. The first part presents the evidence, preliminary but compelling, for a unified interphase chromosome structure. The second part presents the proposed unified interphase chromosome architecture. The third part shows that this interphase chromosome structure could be further organized as chromosome territories: for example, by fitting the 46 human chromosomes into a 10-μm-diameter nucleus. The fourth part unifies this structure into a polytene chromosome architecture and lampbrush chromosomes. The paper concludes with a living light microscopy cell study showing that the G1 nucleus has very similar structures throughout this organelle.The interphase nucleus encloses the genomic DNA, as well as the machinery for regulation of gene expression, RNA synthesis, and DNA replication (2). DNA is packaged into chromatin, the in vivo structure of which remains unclear. While mitotic chromosomes are highly condensed, interphase chromosomes decondense but remain in distinct territories with little overlap. Interphase chromatin is organized in a number of ways, including immutable gene-rich and -poor domains in the primary sequence and expression-promoting or -suppressing regions that may vary during the cell cycle or reflect cell differentiation (2). A classic distinction is drawn between euchromatin and heterochromatin, with the former more “open” and prone to expression, while the latter is more “closed” and prone to silencing (but see ref. 3). However, different methods, such as fluorescence and electron microscopy (EM), or posttranslational histone modifications, are sensitive to different parameters and do not necessarily agree in their identification.The predominant model to describe the path of the DNA strand in the interphase nucleus is the constrained random walk (4) or fractal globule (5, 6). At prophase, the dispersed polymers must recondense without entangling. During mitosis, the space-filling interphase chromatin condenses into a compact micrometer-sized structure. The degree of order in these structures remains undefined.The double-stranded DNA polymer itself, which in isolation appears as a semiflexible, right-hand helix 2 nm in diameter, winds tightly around core histones to form nucleosomes. Each nucleosome has a DNA footprint of 146 base pairs and a geometric diameter of ∼11 nm (7–9). Nucleosomes appear as beads on a string, but the density and spacing of nucleosomes along the DNA sequence may be highly variable (10). In the next stage, the nucleosomes are supposed to coil up into a 30-nm filament, possibly as a tight solenoid or alternatively with a zig-zag structure (11, 12). Today, the 30-nm filament is largely considered an artifact (13, 14). It has been observed in vitro, in isolated or ruptured nuclei, and in cases of deliberate manipulation of divalent cation concentration, but not in intact nuclei (13–15).Current insights into chromatin structure arise primarily from methods based on sequencing. With a number of significant variations, chromatin is cross-linked, cleaved, captured, and sequenced in order to determine which sequences lie in close proximity (5, 16). These methods have revealed a genetic structure of chromosome territories at the largest scale, active and inactive compartments at the multimega base level, followed by topologically associated domains (also known as TADs) whose regulation is controlled concomitantly even if they appear to be distant in sequence (16, 17). An overall, three-dimensional (3D) spatial map of the genome can be generated from the proximity constraints. Extension of the methods to analysis of individual cells revealed a strong heterogeneity, however, making it difficult to connect proximity data to local structure.Microscopy offers the most direct observations of structure, but specimen preparation may be disruptive. Classic EM requires fixation, followed by solvent-based dehydration and impregnation with a hardening polymer. Heavy metal salts are added to generate image contrast based on electron scattering; the indirect nature makes it difficult to interpret apparent density in terms of molecular composition (18). This limitation was circumvented by electron spectroscopic imaging, which distinguishes protein from nucleic acid on the basis of nitrogen and phosphorus concentrations (19). A recent advance used a DNA-binding dye to induce a localized polymerization of diaminobenzidine, which in turn binds an osmium stain (15). A modest density difference between euchromatin and heterochromatin was found, but no evidence was seen for long-range order (15). Fluorescence microscopy has made great advances with the introduction of superresolution methods. The combination with in situ hybridization permits even a degree of sequencing in situ. Still, long exposures and biochemical manipulations require strong cross-linking, which necessarily influences local structure. Cryoelectron tomography offers the most direct and pristine view of cellular structure, including chromatin, but conventional TEM requires thin sectioning or lamella fabrication using the focused ion beam microscope.Cryoscanning transmission electron tomography is a new addition to the toolkit of cellular imaging techniques. The most obvious advantages in relation to conventional defocus phase-contrast TEM are the ability to accommodate thicker specimens and the quantitative contrast based on electron scattering cross-sections. As implemented for cellular tomography, it provides: 1) a unipolar optical transfer function with the specimen in focus, 2) a long depth of field, and most importantly, 3) strong contrast for low spatial frequencies (20). We have recently demonstrated the application of cryoscanning transmission electron tomography in combination with 3D iterative deconvolution processing to whole-cell tomography and obtained a view of the cell nucleus that revealed unexpected large-scale structures (1).Data Considerations and Their Complexity.The primary data for this paper have considerably different attributes, such as the close-spaced 3D pixels with subtle gray level differences and textures, requiring new ways to display its details and architecture. This is a common problem for various imaging technologies (e.g., cellular cryoelectron tomography and MRI). The usual methods to visualize 3D data, such as moving up and down in the Z-dimension through two-dimensional (2D) slices, do not adequately show 3D relationships. Therefore, we propose the extensive use of stereo with extensions to circumvent this problem. Evidence for the chromosome structure is presented primarily in 3D stereo movies of various kinds (Movie S1 and rocking angular stereo pairs A to C′ presented in Movies S2–S8). Visualization guidance and challenges for the stereo movies are also discussed in depth in SI Appendix. |
