Methods xxx (2014) xxx–xxx

Contents lists available at ScienceDirect

Methods journal homepage: www.elsevier.com/locate/ymeth

Chromatin structure revealed by X-ray scattering analysis and computational modeling Kazuhiro Maeshima a,b,c,⇑, Ryosuke Imai a,b, Takaaki Hikima c, Yasumasa Joti c,d a

Biological Macromolecules Laboratory, Structural Biology Center, National Institute of Genetics, Mishima, Shizuoka 411-8540, Japan Department of Genetics, School of Life Science, Graduate University for Advanced Studies (Sokendai), Mishima, Shizuoka 411-8540, Japan c RIKEN SPring-8 Center, 1-1-1 Kouto, Sayo-cho, Sayo-gun, Hyogo 679-5148, Japan d XFEL Utilization Division, Japan Synchrotron Radiation Research Institute (JASRI), 1-1-1 Kouto, Sayo-cho, Sayo-gun, Hyogo 679-5198, Japan b

a r t i c l e

i n f o

Article history: Received 3 May 2014 Revised 23 July 2014 Accepted 18 August 2014 Available online xxxx Keywords: Chromatin Chromosomes 30-nm chromatin fiber Small-angle X-ray scattering Cryo-electron microscopy Computational modeling

a b s t r a c t It remains unclear how the 2 m of human genomic DNA is organized in each cell. The textbook model has long assumed that the 11-nm-diameter nucleosome fiber (beads-on-a-string), in which DNA is wrapped around core histones, is folded into a 30-nm chromatin fiber. One of the classical models assumes that the 30-nm chromatin fiber is further folded helically to form a larger fiber. Small-angle X-ray scattering (SAXS) is a powerful method for investigating the bulk structure of interphase chromatin and mitotic chromosomes. SAXS can detect periodic structures in biological materials in solution. In our SAXS results, no structural feature larger than 11 nm was detected. Combining this with a computational analysis of ‘‘in silico condensed chromatin’’ made it possible to understand more about the X-ray scattering profiles and suggested that the chromatin in interphase nuclei and mitotic chromosomes essentially consists of irregularly folded nucleosome fibers lacking the 30-nm chromatin structure. In this article, we describe the experimental details of our SAXS and modeling systems. We also discuss other methods for investigating the chromatin structure in cells. Ó 2014 Elsevier Inc. All rights reserved.

1. Introduction How the 2 m of genomic DNA is organized in cells remains one of the basic questions in cell biology. In the typical textbook view of chromatin, the 2-nm-diameter DNA molecule is wrapped around histones and forms a ‘‘nucleosome’’ (10-nm fiber) structure [1–3]. This nucleosome has been assumed to become folded into a regular ‘‘30-nm chromatin fiber’’ (see Fig. 3A) [4,5] and further higher-order structures. In terms of higher-order structures, the well-established ‘‘hierarchical helical folding model’’ assumes that the 30-nm chromatin fiber is folded progressively into larger fibers, including 100-nm and then 200-nm fibers, to form the final mitotic chromosomes or large chromatin fibers (chromonema fibers) in the interphase cell [6–8]. Another model, the ‘‘radial loop model’’, suggests that the 30-nm chromatin fiber folds into radially oriented loops [9– 11]. What does chromatin actually look like in a cell? Small-angle X-ray scattering (SAXS; Fig. 1A) is a suitable technique for investigating bulky chromatin structures. When X-rays are exposed to non-crystalline materials, the small-angle scattering

patterns generally reveal local periodic structures (Fig. 1B) (e.g., [12]). SAXS has been used widely to analyze the structures of isolated nucleosomes and chromatin in solution (e.g., [13–18]). In the case of interphase nuclei and mitotic chromosomes, using SAXS analysis, Langmore and Paulson detected a 30-nm structure [19,20], which has long been regarded as evidence for the existence of the 30-nm fiber in the interphase nuclei and mitotic chromosomes. In this method review paper, we describe in detail a SAXS method for isolated nuclei and chromosomes [21–23]. To further understand the X-ray scattering profiles, we introduced computational modeling of condensed chromatin using three-dimensional atomic coordinates (i.e., ‘‘in silico condensed chromatin’’). Our findings support a model in which chromatin consists essentially of irregularly folded nucleosome fibers lacking the 30-nm chromatin structure [21–23]. We also discuss other strategies for studying chromatin structures in cells, including X-ray- and electron-microscope (EM)-based methods. 2. Methods 2.1. Isolation of human nuclei and chromosomes

⇑ Corresponding author at: Biological Macromolecules Laboratory, Structural Biology Center, National Institute of Genetics, Mishima, Shizuoka 411-8540, Japan. E-mail address: [email protected] (K. Maeshima).

In SAXS analysis, membranous structures, such as small vesicles, can induce very strong scattering that can hide or mask

http://dx.doi.org/10.1016/j.ymeth.2014.08.008 1046-2023/Ó 2014 Elsevier Inc. All rights reserved.

Please cite this article in press as: K. Maeshima et al., Methods (2014), http://dx.doi.org/10.1016/j.ymeth.2014.08.008

2

K. Maeshima et al. / Methods xxx (2014) xxx–xxx

chromosomes were suspended briefly in Buffer A [15 mM Tris– HCl (pH 7.5), 80 mM KCl, 2 mM EDTA, 2 mM spermine, 5 mM spermidine, 0.1 mM PMSF, and 0.05% (w/v) digitonin]. After centrifugation at 1000g for 5 min, nuclei or chromosomes were recovered as pellets and resuspended in IB for analysis. As a control, chicken erythrocyte nuclei were prepared and evaluated, as described by Langmore and Schutt [26]: first, 1 mL of fresh chicken blood was lysed with10 mL of MLB [60 mM KCl, 15 mM NaCl, 15 mM HEPES, pH 7.3, 2 mM MgCl2, 0.1% Nonidet P-40, and 1 mM PMSF] for 10 min on ice. After centrifugation at 4 °C and 1000g for 5 min, the supernatant was removed and resuspended in 10 mL of MLB. This step was repeated twice. The nuclei were centrifuged into glass capillaries at 1000g and 4 °C for 5 min. 2.2. SAXS measurements at the SPring-8 BL45XU beamline (Figs. 1 and 2)

Fig. 1. Small-angle X-ray scattering (SAXS). (A) Nuclei or chromosomal pellets in quartz capillary tubes were exposed to a synchrotron X-ray beam and scattering patterns were recorded on imaging plates. (B) When X-rays irradiate noncrystalline materials, small-angle scattering generally reflects the periodicities of internal structures. (C) A typical scattering pattern features concentric rings and is reminiscent of a doughnut. Signals at smaller scattering angles (evidencing smaller scattering vectors [S values] with respect to the center) are indicative of larger periodic structures, and vice versa. The image was reproduced from [22], with modifications.

scattering from chromatin. Therefore, human nuclei and chromosomes were purified from HeLa cells, essentially as described previously [24,25]: HeLa cells (2  107 cells) were gently resuspended in 50 mL of isotonic buffer IB [10 mM HEPES–KOH (pH 7.5), 50 mM NaCl, 5 mM MgCl2, 0.1 mM PMSF, 0.1% Trasylol], and then incubated for 10 min on ice. After centrifugation, the cell pellet was suspended in 12 mL of buffer IB containing 0.5 M sucrose. After a 2-min incubation on ice, a solution of 10% NP40 and 5% sodium deoxycholate (DOC) was diluted 100-fold into the cell suspension, and the cells were disrupted immediately with 15 vigorous strokes of a small type A pestle in a Dounce homogenizer. The lysates were layered immediately onto 30 mL of a sucrose cushion containing IB and 40% sucrose, 0.1% NP40, and 0.05% DOC. Centrifugation was carried out for 30 min at 2500g and 4 °C. The chromosome pellets were resuspended in 0.2 mL of the same buffer. When preparing mitotic chromosomes, we sought to avoid prolonged mitotic arrest by briefly (2 h) treating mitotic HeLa cells with 0.1 lg/mL nocodazole followed by mitotic shake-off. First, the experimental conditions of Langmore and Paulson were reproduced precisely [19,20]. Isolated nuclei and chromosomes were suspended in IB with 0.1% (v/v) NP40 and examined using SAXS. Isolated nuclei and chromosomes maintained under more physiological conditions, i.e., in IB2 and IB3 [10 mM HEPES–KOH (pH 7.5), 50 mM (IB2), or 100 mM (IB3) NaCl, 5 mM MgCl2, 0.1 mM PMSF, and 0.1% (v/v) NP40], yielded similar scattering profiles, each with the same three peaks [21]. To remove ribosomal aggregates from the surfaces of nuclei or chromosomes, isolated nuclei and

SAXS can be performed using a laboratory X-ray generator or the beamline at a synchrotron radiation facility (e.g., [12]). A beamline has the advantages of greater parallelism and a higher photon flux, compared with the beam afforded by an X-ray generator. A beam that is more parallel can be used to measure smaller scattering angles, facilitating structural analysis of larger objects. A higher photon flux allows detailed scattering data to be collected in a shorter time. The SAXS experiments were performed at the RIKEN Structure Biology Beamline I (BL45XU) at SPring-8; this is a third-generation synchrotron facility in Japan [27]. The BL45XU beam was configured for the SAXS experiments as follows (Fig. 1A). The X-ray wavelength and sample-to-detector distance were 1.0 Å (12.4 keV) and 2.1 m, respectively. Scattering data from chromosomal samples and buffer were collected at room temperature using an imaging plate system (R-AXIS IV++; Rigaku, Tokyo, Japan). A cooled charge-coupled device (CCD) equipped with a twodimensional (2D) X-ray image intensifier can also be used to this end. The precise distance between the sample and detector was determined using diffraction data from powdered silver behenate. Our set-up was capable of observing up to 90- or 120-nm range (lowest scattering vector). A typical SAXS scattering pattern composed of concentric rings is shown in Fig. 1C. The signals at smaller angles (i.e., closer to the center) reflect larger structures, and vice versa [12]. The radial average of the scattering pattern, the so-called ‘‘SAXS profile’’, is used for the analysis to increase the signal-to-noise ratio. In Fig. 2, the scattering signals on the concentric rings were averaged using the program Fit2d (http://www.esrf.eu/computing/scientific/ FIT2D/) and are drawn as one-dimensional (1D) plots. For SAXS measurements, concentrated nuclear or chromosomal suspensions were loaded into the top portions of quartz glass X-ray capillaries (2 mm in diameter; Tohso, Tokyo, Japan) pre-filled with a buffer such as IB. Nuclei or chromosomes in glass capillaries were next sedimented, to form pellets, at 4 °C, in a swinging bucket centrifuge running at 1000g for 5 min. The nuclei or chromosomes typically formed a 0.5–1.0 mm-deep pellet in a capillary, which was otherwise filled with buffer. A sample must be carefully injected to avoid bubbles and impurities that cause scattering noise. Nuclear and chromosomal samples were maintained at 0–4 °C at all times. Notably, we did not fix samples in aldehydes or osmium tetroxide, and eschewed alcohol dehydration. The use of such reagents is common during conventional electron microscopy sample processing, and might produce artifacts [28]. The pellets were exposed to the synchrotron radiation beam for 0.1–60 s, depending on the signal intensity obtained. As five sequential exposures to the X-ray beam did not change the scattering profiles (data not shown), we believe that the radiation delivered did not significantly damage the chromatin structure.

Please cite this article in press as: K. Maeshima et al., Methods (2014), http://dx.doi.org/10.1016/j.ymeth.2014.08.008

K. Maeshima et al. / Methods xxx (2014) xxx–xxx

3

Fig. 2. SAXS profiles of HeLa interphase and mitotic chromatin. (A) SAXS profiles of HeLa interphase nuclei (upper) and chromosomes (lower). Three peaks at 6, 11 (faint) and 30 nm were evident (arrows). Profiles before (left) and after (right) removal of ribosomes. Note that only the 30-nm peak disappeared after removal of ribosomal aggregates; the other peaks remained. (B) Two types of nucleosome positioning: face-to-face, at 6-nm spacing, and edge-to-edge, at 11-nm spacing. (C) Model: the SAXS 30-nm peak reflects the presence of regularly spaced ribosomal aggregates, and does not yield information on chromosomes or nuclei per se. (D) A typical SAXS pattern of chicken erythrocyte nuclei. A sharp 30-nm peak (arrow) is evident. Additionally, two peaks, at 11 nm and 6 nm, are prominent. The data are reproduced from [21,22], with modifications.

Measurements were performed three to five times on different segments of the pellet. Buffer scattering was measured by delivering the beam to the buffer-only region of the same capillary. Following the approach of Langmore and Paulson [19,20], we show

most of the SAXS data in this Methods paper as plots of log(I(S)  S2) vs. S, obtained after subtracting buffer scattering, where I(S) is the average intensity along the concentric rings (SAXS profile) and S is the size of the scattering vector. The calculation of

Please cite this article in press as: K. Maeshima et al., Methods (2014), http://dx.doi.org/10.1016/j.ymeth.2014.08.008

4

K. Maeshima et al. / Methods xxx (2014) xxx–xxx

I(S)  S2 yields the compensated relative strength (power) of any sample structural periodicity.

model nucleosome structures were calculated using Eq. (4). Summation was over non-hydrogen atoms.

2.2.1. Computer simulation of small-angle X-ray scattering profiles (Figs. 3 and 4) The scattering intensity of a material is proportional to the following equation;

2.2.3. Computational details of SAXS simulations (Figs. 3 and 4) Our in-house software, parallelized using the Message Passing Interface (MPI) library, was adapted to efficiently compute SAXS profiles for a few tens of millions of atoms. When the SAXS profiles of Eq. (4) are computed directly, the required computational time is proportional to the product of the number of sampling points of S and the square of the number of atoms in the model. To reduce computational time, Eq. (4) was approximated as:

Z 2   IðSÞ ¼  qðrÞ expð2piS  rÞdr

ð1Þ

V

where qðrÞ and S are the sample electron density distribution and the scattering vector, respectively. Here, I(S) is defined in terms of the 3D vector S. Integration is performed over the interaction volume (V) between X-rays and the sample. The magnitude of the scattering vector is related to the scattering angle, 2h, and the X-ray wavelength, k, by;

S ¼ jSj ¼

2 sin h k

IðSÞ ¼

X sinð2pSr AB Þ f A ðSÞf B ðSÞHðr AB Þ 2pSr AB r

ð5Þ

AB

where H(rAB) is the number of pairs of atoms of types A and B such that the distance between them ranges from rAB  Dr/2 to rAB + Dr/

ð2Þ

When we expand the electron density distribution with atomic contributions, Eq. (1) can be rewritten as:

2   X   IðSÞ ¼  f a ðSÞ expð2piS  ra Þ :   a

ð3Þ

Here, f a ðSÞ and ra denote the atomic scattering factor of atom a, i.e., the Fourier transform of the electron density of atom a at the origin, and the positional vector of atom a, respectively. In SAXS experiments, scattering intensity is usually measured in randomly oriented samples. This renders the intensity isotropic. Accordingly, the 1D function of the SAXS profile, I(S), is calculated as the spherical average of the 3D function of I(S) in Eq. (3), and is:

IðSÞ ¼ hIðSÞiX ¼

X sinð2pSr ab Þ f a ðSÞf b ðSÞ 2pSrab a;b

ð4Þ

where rab is the distance between atoms a and b. The right-hand side of Eq. (4) is called the Debye formula [29]. Therefore, SAXS profiles can be computed from atomic coordinates with the aid of Eq. (4). The atomic scatting factor of a, fa(S), is computed as a combination of Gaussian functions (International Tables for Crystallography, Vol. C., edited by E. Prince [2004]). 2.2.2. Computational modeling of condensed chromatin from 30-nm chromatin fibers (‘‘in silico condensed chromatin’’, Fig. 4A) We modeled condensed chromatin structurally using atomic coordinates to derive a one-start helix (solenoid) model and a two-start helix (zigzag) model. Atomic coordinate models of the one-start and two-start helices were kindly provided by Dr. D Rhodes, LMB, UK [30,31]. Condensed chromatin structures were modeled as follows. (i) The position and orientation of the first 22-mer (of either a one- or a two-start helix) were randomly generated within a sphere of radius R. (ii) The position and orientation of the second 22-mer were also generated randomly within the same sphere, so that the two 22-mers were in contact (two 22mers were defined to be in contact if the distance between nucleosomes of the 22-mers was less than 12 nm.). Model selection (one-start or two-start) was performed using random numbers. (iii) The positions and orientations of the third and later 22-mers were generated randomly within a sphere of radius R, so that the ‘‘incoming’’ 22-mer had at least two points of contact with prior 22-mers. In computation of SAXS profiles, five models were generated for each radius R and the average of normalized SAXS profiles over 15 models was calculated. Table 2 summarizes the parameters used in modeling of complex structures. SAXS profiles of

Fig. 3. Two structural models of 30-nm chromatin fibers. (A) The one-start helix (the solenoid model; left) and the two-start helix (the zigzag ribbon model; right) are shown, with their atomic coordinates (kindly provided by Dr. D. Rhodes) [31,32]. The models were constructed using MolScript [71]. (B) The atomic coordinates were used to derive scattering profiles computationally simulated using one-start helix (blue line) and two-start helix (red line) models; details are provided in Section 2.

Please cite this article in press as: K. Maeshima et al., Methods (2014), http://dx.doi.org/10.1016/j.ymeth.2014.08.008

5

K. Maeshima et al. / Methods xxx (2014) xxx–xxx

2. Here, Dr was 0.3 nm, chosen so that the difference between Eqs. (4) and (5) was negligible in the region of interest for S. Thus, the computational time was proportional to the sum of the number of sampling points for S and the square of the number of atoms in the model. The computational times for the SAXS simulations of single one-s and two-start 22-mers (Fig. 3B) were 60 s using a 64-CPU core. The average computational times for SAXS simulations of 25, 50, and 100 of 22-mers were 6400 s using a 250 CPU core, 12,000 s using a 500 CPU core, and 45,000 s using a 500 CPU core, respectively. The computations were performed using the ‘‘Thin compute nodes’’ in the supercomputer system of the National Institute of Genetics (NIG). 3. Experimental and computational results

Table 1 Summary of the structural peaks revealed by X-ray-scattering and computational modeling.

Erythrocyte nuclei Single 30-nm chromatin fiber simulation (one-start) Single 30-nm chromatin fiber simulation (two-start) In silico condensed chromatin Human nuclei (with ribosomes) Human nuclei (without ribosomes) Human mitotic chromosomes (with ribosomes) Human mitotic chromosomes (without ribosomes)

30-nm peak

11-nm peak

6-nm peak

+++ +++

+++ ++++

+++ +++

+++

++++

+++

+++ +++  +++

++++ + + +

+++ +++ +++ +++



+

+++

3.1. SAXS measurements At SPring-8 (Harima, Japan), isolated human interphase nuclei and mitotic chromosomes were exposed to the synchrotron Xray beam (Fig. 1A). A typical scattering profile of interphase nuclei and mitotic chromosomes exhibited three peaks, at 30 nm, 11 nm (weak), and 6 nm (Fig. 2A, left) [21,22]. These results are completely consistent with those of Langmore and Paulson [19,20]. The cited authors suggested that the 6- and 11-nm peaks reflected nucleosome face-to-face and edge-to-edge positioning, respectively (Fig. 2B). It was assumed that the 30-nm peak reflected side-by-side packing of 30-nm chromatin fibers. From this SAXS observation, could we conclude that the interphase nuclei and mitotic chromosomes have the 30-nm chromatin fiber? The answer is no because SAXS can identify inverse-space (regularity or periodicity within the sample), but not a real image (how the sample looks). To see the origin of the 30-nm peak in SAXS, the samples were observed using cryo-EM; it was found that the chromosome surface was coated with electron-dense granules. Subsequently, the electron-dense granules on the chromosome surface were found to be ribosomes by immunostaining using a specific antibody for ribosome components. The ribosomes on the surface were stacked regularly at a spacing of 30 nm, which might produce the 30 nm SAXS peak. Consistently, when ribosomes were removed from the chromosome surfaces by washing with A buffer, containing a polyamine and EDTA [19,20,24,32], the 30-nm SAXS peaks of both the nuclei and mitotic chromosomes disappeared (Fig 2A, right). However, other peaks characteristic of internal nucleosomal structures remained. This treatment did not change the size or shape of the chromosomes [21]. These results suggested that no 30-nm fiber was present in either interphase chromatin or mitotic chromosomes (for a structure model, see Fig. 5). As a control, chicken erythrocyte nuclei were exposed to the beam and a sharp 30-nm peak and two prominent peaks at 11 and 6 nm were observed (Fig. 2D and Table 1), consistent with previous reports [26,33]. We therefore concluded that SAXS should reveal 30-nm structures if such structures in fact existed (see also Section 4.4). 3.2. Computational modeling As we mentioned, SAXS can identify regularity or periodicity within a sample, but not the real structure. To further understand the SAXS profiles obtained, an ‘‘in silico SAXS study’’ was conducted using computational modeling. If the 3D coordinates of a structure are available, the expected SAXS pattern of that structure can be calculated via Fourier transformation (see Section 2.2.1). Comparison between real and in silico SAXS results can yield quantitative structural information, including the origins of the peaks in SAXS profiles.

In silico SAXS models of two 30-nm chromatin structures [30,31] featured prominent peaks at 30–40, 10–13, and 6 nm (Fig. 3B; Table 1). The 6- and 11-nm peaks were derived from face-to-face and edge-to-edge nucleosomal positioning, respectively (Fig. 2B). Here, edge-to-edge positioning likely corresponds to the helical pitch of the fiber. Considering the structural model in Fig. 3 [30,31], formation of a 30-nm chromatin fiber seems to require similar frequencies of face-to-face (6-nm peak) and edge-to-edge (11-nm peak) positioning (Table 1). Consistent with this notion, a SAXS analysis of erythrocyte nuclei revealed two prominent peaks of 11 and 6 nm in addition to the 30 nm peak (Fig. 2D and Table 1). Fig. 2A shows that face-to-face positioning is apparently dominant over edge-to-edge positioning. This predominance of edge-to-edge stacking also strongly suggests the absence of a 30-nm fiber in human interphase and mitotic chromatin. Next, ‘‘in silico condensed chromatin’’ containing tightly packed 30-nm chromatin fibers was reconstructed (Fig. 4A). The nucleosome concentration of in silico condensed chromatin was about 0.5 mM, comparable to that of dense heterochromatin or mitotic chromosomes (for a review, see [34]). The calculated scattering profile of in silico condensed chromatin contained a prominent signal in the large-sized region and also a prominent peak at 30 nm (arrow in Fig. 4B), in addition to two prominent peaks at 11 and 6 nm (Fig. 4B; Table 1). This profile shows that in silico condensed chromatins retain a characteristic 30-nm fiber. As the SAXS profiles of interphase and mitotic chromatin did not have 30-nm peaks, we concluded that chromatin was not assembled from 30-nm chromatin fibers, but rather from interdigitated nucleosomal fibers, which we term a polymer-melt.

4. Discussion 4.1. Comparison with cryo-EM and other EM-based procedures SAXS analysis showed that chromatin did not consist of 30-nm fibers, but rather of interdigitated and melted nucleosomal fibers. This finding is consistent with what was observed by cryo-EM, which allows observation of biological samples in near-native states. In this approach, mammalian cells are frozen rapidly; thin frozen sections are produced at low temperature; and these are observed. Strikingly, cryo-EM did not reveal any 30-nm chromatin fibers in cells, but rather a uniform disordered 10-nm structure [35–41]. In addition to the cryo-EM studies of mammalian cells, using cryo-EM tomography to study the picoplankton Ostreococcus tauri, the smallest known free-living eukaryote, Gan et al., found that O. tauri chromatin resembles a disordered assembly of nucleosomes

Please cite this article in press as: K. Maeshima et al., Methods (2014), http://dx.doi.org/10.1016/j.ymeth.2014.08.008

6

K. Maeshima et al. / Methods xxx (2014) xxx–xxx

Table 2 Parameters for in silico modeling of condensed chromatin (Fig. 4). Model name

R [nm]

Number of 22mers

Ratio of one-start to two-start helices

mix100 mix50 mix25

122.4 97.1 77.1

100 50 25

1:1 1:1 1:1

[43,44]). The cited authors found that pluripotent mouse cells contained highly dispersed meshes of 10-nm fibers, but no 30-nm fibers [45,46]. ESI tomographic data showed that condensed heterochromatin domains, such as chromocenters, were formed of 10- but not 30-nm fibers [46] (also, see [47,48]). 4.2. Ultra-small angle X-ray scattering of human nuclei and chromosomes Above, we focused on local chromatin structure. What is the higher-order chromatin structure? As mentioned above, EM-based methods examine only limited regions of cells in thin sections, thus not whole cells. It may be difficult to observe higher-order chromatin structure. We earlier explored the large-scale chromatin structures of interphase nuclei and mitotic chromosomes via ultrasmall-angle X-ray scattering (USAXS), as developed by Nishino et al. [49]. No regular periodic structure, for example no 100– 150- or 200–250-nm fibers, in either interphase or mitotic chromatin was observed. Such observations are not in line with the classical hierarchical helical folding model [21,22]. The scattering properties suggested that a scale-free structure of up to 275 nm was present in interphase chromatin and that the extent of scale-free structure attained 1000 nm in mitotic chromosomes. These data implied that interphase chromatin was condensed, to an extent of at least 275 nm, but that this reflected irregular folding of nucleosome fibers (10-nm fibers) in the absence of any 30nm chromatin structure (Fig. 5). 4.3. Chromatin domains observed by fluorescence microscopy imaging and chromosome capture methods Chromatin domains of 1 Mb in size were also observed using fluorescence microscopy imaging, as foci of DNA replication via pulse labeling [50–52]. Recently, several reports using Hi-C and 5C methods have proposed that the genomic DNA is packaged physically in the cells. The Chromosome Conformation Capture (3C) and its variants, including 4C, 5C, and Hi-C, constitute a basic method for investigating the 3D organization of genomic DNA. The supposed DNA packing structures have been called ‘‘topologically associating domains’’ (TADs) [53–55] (for a review see [56]) and can be hundreds of kilobases in size. TADs have been identified in fly, mouse, and human cells, suggesting that the chromatin domains are universal building blocks of chromosomes.

Fig. 4. Reconstruction of ‘‘in silico condensed chromatin.’’ ‘‘In silico condensed chromatin’’ models were constructed in environments containing 100, 50, and 25 tightly packed 30-nm chromatin fibers. The nucleosome concentration was 0.5 mM. The 100-fiber model was drawn using MolScript [71]. (B) The simulated scattering profiles, yielding average among-model values, have prominent peaks at 30, 11, and 6 nm. This shows that in silico condensed chromatin retains characteristics of 30-nm chromatin fibers.

without any 30-nm chromatin fiber, similar to a polymer-melt [42]. While SAXS analyses can examine bulk chromatin structure, they used nuclei and chromosomes isolated from cells. This might cause a structural preservation problem with the samples. On the other hand, cryo-EM can examine biological samples in close to their native states, but only a limited portion of the sample, because the section thickness is only 50 nm [40]. It can be difficult to observe the bulk structure of chromosomes using cryoEM. Therefore, these two strategies complement each other and their combination enables a decisive conclusion to be drawn. Bazett-Jones et al. obtained similar data in work on various types of mouse cells using another EM-based imaging method, electron-spectroscopic imaging (ESI). In this technique, mapping of phosphorus and nitrogen atoms affords levels of contrast and resolution sufficient to visualize 10-nm fibers (for a review, see

4.4. A novel chromatin model and absence of the 30-nm chromatin fiber in cells As described above, recent findings suggest that chromatin consists of irregularly folded nucleosome fibers without the 30-nm chromatin fiber, and forms numerous condensed domains (e.g., TADs) (Fig. 5). Although the near absence of 30-nm chromatin fibers in eukaryotic cells was suggested, why are these structures shown in EM images in many papers and molecular biology textbooks; e.g., [4,5,16–18,31,57–60]? We believe that most 30-nm chromatin fibers in EM images are in vitro artifacts caused by the low-salt buffer conditions. The formation of 30-nm chromatin fibers requires the selective binding of nucleosomes, which are close neighbors on the DNA strand, via intra-fiber nucleosomal association. In low-salt buffer conditions of