Figures and data
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_fig1.tif/full/max/0/default.jpg)
Canonical nucleosomes in vitro and in situ
(A) Space-filling model of the reconstituted yeast nucleosome crystal structure (PDB 1ID3) (White et al., 2001), showing from left to right, the disc, side, and gyre views. The pseudo-dyad axis is indicated by the arrow. The DNA is rendered as light blue and the histones in the core are shaded blue (H3), green (H4), red (H2B), and yellow (H2A). (B) Subtomogram average of nucleosomes from wild-type (BY4741) yeast nuclear lysates. The linker DNA is indicated by the short arrows and the DNA gyre motifs are indicated by the arrowheads. (C) Subtomogram averages of nucleosomes in wild-type cryolamellae, oriented similarly to the nucleosomes in the other two panels. The upper (blue) class has more ordered linker DNA than the lower (magenta) class. Note that the subtomogram average in panel B looks different from those in panel C (and in Figures 3 and S17) because it is at higher resolution (18 Å vs 24 Å).
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_fig2.tif/full/max/0/default.jpg)
Canonical nucleosomes are rare in wild-type cells
(A) VPP tomographic slice (12 nm) of a BY4741 cell cryolamella. Large subcellular structures are labeled: nuclear pore complex (NPC), nuclear megacomplex (M), nuclear microtubule (MT), nuclear envelope (NE), and ribosome (R). The inset is a 4-fold enlargement of the boxed area, and a nucleosome-like particle (n) is indicated. (B) Remapped model of the two canonical nucleosome class averages in the tomogram from panel A. The approximate location of the nuclear envelope is indicated by the blue dashed line. The insets B1 and B2 show 4-fold enlargements of the corresponding boxed areas. Note that the remapped model projects the full 150 nm thickness of this cryolamella.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_fig3.tif/full/max/0/default.jpg)
Visualization of GFP-tagged nucleosomes in vitro
(A) Example 2-D class averages of nucleosome-like particles that were template-matched with a featureless cylinder reference. Out of 88,896 template-matched particles, 66,328 were retained after 2-D classification. (B) Class averages (3-D) of nucleosomes from nuclear lysates. Solid arrowheads indicate the extra (GFP) densities. Open arrowheads indicate the positions that lack this density. These class averages were obtained after classification directly from subtomogram averaging, without an intervening 2-D classification step. (C) The approximate positions of the H2A C-termini are rendered in yellow and indicated by arrows in the crystal structure of the yeast nucleosome (White et al., 2001). Note that in the crystal structure, fewer of the H2A C-terminal amino acids were modeled than for H2B, meaning that the H2A C-terminus is not perfectly ordered. To facilitate comparison, this structure is oriented like the class averages in panel B. The nucleosome densities in panel B are longer along the pseudo-dyad axis (horizontal) because they have linker DNA, which is absent in the nucleosome crystal structure.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_fig4.tif/full/max/0/default.jpg)
Canonical nucleosome class averages are not detected in LGY0016 (H2A-GFP) in situ
Class averages (3-D) of nucleosome-like particles in VPP cryotomograms of LGY0016 cryolamellae. The starred classes have two linear motifs. Movie S5 shows the progress of this classification job.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_fig5.tif/full/max/0/default.jpg)
Models of yeast nucleosome heterogeneity
Schematics of DNA (light blue) and histones (shaded pie slices) in the nucleosome disc view. The cartoons only illustrate the 147 bp of “core” DNA. (A) Canonical nucleosome, in which all the histones and 147 bp of DNA are part of an ordered complex. (B) Nucleosome with alternative histone H2A-H2B (yellow, red) conformations and partially dissociated DNA. (C) Nucleosome bound to non-histone proteins (gray). The blurred gray box represents different proteins that can bind, thereby contributing to constitutional heterogeneity. The blurred appearance represents a large range of positions and orientations that protein and DNA components adopt inside cells, which would result in the absence of a class average resembling a canonical nucleosome. (D) Canonical nucleosomes are a minority conformation in situ.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_figs1.tif/full/max/0/default.jpg)
Subtomogram classification workflow
Classification starts with subtomograms that contain nucleosome-like particles, template matched in PEET (Cai et al., 2018). (A) In the workflow recommended by Bharat (Bharat et al., 2015), the subtomograms are first subjected to sequential rounds of 2-D classification to remove particles that belong to ‘bad’ classes. Once the bad classes are removed, the remaining set is subjected to 3-D classification. (B) In the alternative “direct 3-D” classification workflow, the subtomograms are subjected directly to 3-D classification. The 2-D classification steps (greyed out) are bypassed.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_figs2.tif/full/max/0/default.jpg)
Overview of BY4741 (Wild type) nuclear lysate
Tomographic slice (12 nm) of BY4741 nuclear lysates. Some non-chromatin features are indicated: granule (G), coated vesicle (V), naked DNA (DNA), membrane fragments (membrane), carbon support film (carbon), virus-like particle (VLP), and gold fiducial (Au). The abundant granular densities in this field of view are nucleosome-like particles. One example is indicated in the 4-fold enlargement in the inset (n). The linear features at the lower left and upper right are back-projection artifacts from the image borders. These regions are excluded from analysis.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_figs3.tif/full/max/0/default.jpg)
Direct 3-D classification of BY4741 (Wild type) nuclear lysates
(A) Class averages (3-D) of two types of BY4741 nucleosomes. The canonical nucleosome class averages are shaded blue and the percent of particles belonging to the canonical classes are indicated under the density map. (B) Second round of classification, using the classes indicated in panel A. (C) The resolution of the refined canonical nucleosome is ∼18 Å by the FSC = 0.5 criterion. The refined density map is reproduced from Figure 1B.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_figs4.tif/full/max/0/default.jpg)
Overview of a BY4741 (Wild type) cryolamella
Tomographic slice (12 nm) of a BY4741 cryolamella. Some non-chromatin features are indicated: nuclear envelope (NE), nuclear microtubule (MT), nucleosome-like particle (n), megacomplex (M), ribosome (R), and a putative vacuole (Vac). The high-contrast linear features at the lower left are back-projection artifacts from image borders.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_figs5.tif/full/max/0/default.jpg)
Classification of BY4741 (Wild type) nucleosome-like particles in situ
(A) Class averages (2-D) of template-matched nucleosome-like particles in BY4741 (wild-type) cells. The class averages whose member particles were selected for additional rounds of classification are indicated by a green triangle in the box’s lower right corner. The starting dataset consisted of 62,428 template-matched candidate BY4741 nucleosomes from 3 tomograms, from which 16,608 particles were selected by 2-D classification. (B) Class averages (3-D) of nucleosome-like particles in BY4741 cells.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_figs6.tif/full/max/0/default.jpg)
Direct 3-D classification of BY4741 (Wild type) cryolamellae densities
Class averages (3-D) of nucleosome-like particles in cryotomograms of BY4741 cryolamellae. In this experiment, 2-D classification was bypassed.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_figs7.tif/full/max/0/default.jpg)
Classification using the nucleosome crystal structure reference
A) A featureless cylinder (left) and a simulated EM map from a crystal structure of the nucleosome (right). (B) Class averages (2-D) of BY4741 (wild type) nucleosome-like particles that were template matched using the nucleosome crystal structure. Classes that were pooled for 3-D classification are indicated by a green triangle in the lower right corner of its box. Out of 72,190 template-matched particles, 6,439 were selected by 2-D classification. (C) Class averages (3-D) of the most nucleosome-like particles selected by 2-D classification. The classification was performed using the same dataset as for Figures S5 and S6.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_figs8.tif/full/max/0/default.jpg)
Overview of a BY4741 (Wild type) cryolamella, VPP data
Tomographic slice (12 nm) of a BY4741 cryolamella imaged with a VPP. The nuclear envelope (NE), nuclear pore complex (NPC), ribosome (R), and megacomplex (M) are indicated. The inset shows a threefold enlargement of the boxed area. A nucleosome-like complex (n) is indicated.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_figs9.tif/full/max/0/default.jpg)
Classification of nucleosome-like particles in situ in BY4741 (Wild type) VPP data
(A) Class averages (2-D) of template-matched nucleosome-like particles in VPP cryotomograms of BY4741 (wild-type) cell cryolamellae. The class averages whose member particles were selected for additional rounds of classification are indicated by a green triangle in the box’s lower right corner. The starting dataset consisted of 129,473 template-matched candidate BY4741 nucleosomes from 3 tomograms. 12,551 particles were selected by 2-D classification. (B) Class averages (3-D) of nucleosome-like particles in BY4741 cells image with a VPP.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_figs10.tif/full/max/0/default.jpg)
Direct 3-D classification of BY4741 (Wild type) nuclei in VPP tomograms of cryolamellae
(A) Class averages (3-D) of BY4741 nucleosomes. The canonical nucleosome class average is shaded blue and the percent of particles belonging is indicated under the density map. Note that this percentage is relative to the 129,473 template-matched nucleosome-like particles, not the ∼25,000 expected number of nucleosomes in the 5 tomograms. (B) Second round of classification from the canonical nucleosome class identified in panel A. (C) The resolution of both class averages is ∼24 Å by the FSC = 0.5 criterion. The refined density maps are reproduced from Figure 1C and the number of particles per class is labeled in black. (D) Angular-distribution plot of the canonical nucleosome class from panel A. Movies S1 and S2 show the progress of this classification jobs in panels A and B, respectively.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_figs11.tif/full/max/0/default.jpg)
Overview of a BY4741 (Wild type) cryolamella imaged in the cytoplasm, VPP data
Tomographic slice (12 nm) of a BY4741 cryolamella imaged in the cytoplasm. The vacuoles (V), a ribosome (R), and cell wall (CW) are indicated.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_figs12.tif/full/max/0/default.jpg)
Classification of BY4741 (Wild type) cryolamellae VPP densities from the cytoplasm
Class averages (3-D) of “nucleosome-like” particles that were template matched from the cytoplasm. Both the template-matching and 3-D classification parameters were identical to the ones used to analyze the nucleus. 2-D classification was bypassed. One class has no contributing particles and was removed.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_figs13.tif/full/max/0/default.jpg)
Strategy to tag H2A with GFP
(A) Crystal structure of the yeast nucleosome, PDB 1ID3 (White et al., 2001), in the disc view (left) and side view (right). DNA, dark purple; histone octamer, light gray. One set of histone C-termini are labeled and rendered as space-filling models. Two key points of reference, pseudo-dyad and opposite pseudo-dyad, are indicated by the arrowhead and arrow tail, respectively. The other set of C-termini is related by a 180° rotation around the pseudo-dyad axis. (B) Two views of the GFP crystal structure, PDB 1GFL (Yang et al., 1996), rendered at the same scale as panel A. The N-terminus is rendered as a black space-filling model. (C) Details of the sequence at the H2A-GFP fusion, with (7aa) and without (0aa) the 7 amino-acid linker. The H2A C-terminal sequence is shaded yellow, GFP N-terminal sequence is shaded green, and the flexible linker is shaded gray. LGY0016 has a 0aa linker. (D) Histone genotype of the parent strain BY4741. (E) Homologous recombination strategy to C-terminally tag the sole copy of the H2A gene with GFP. Upper – replacement of the HTA2-HTB2 locus with KanR. Lower – Tagging of HTA1 with GFP. The junction between the HTA1 (H2A) and GFP is boxed. (F) Histone genotype of LGY0016.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_figs14.tif/full/max/0/default.jpg)
Experimental verification of H2A-GFP and H2B-GFP tagging
(A) Maps of the two H2A-H2B loci in the parent (wild-type) strain BY4741 and the H2A-GFP strain LGY0016. Primers used for PCR verification are indicated with the half arrow symbols. (B) Agarose gel of PCR amplicons expected (or not) from LGY0016 genomic DNA, in which the HTB2-HTA2 locus is deleted and the HTA1 locus is tagged with GFP. (C) Immunoblot analysis of strains LGY0016 (expresses H2A-GFP only) and LGY0012 (expresses H2A only, control). The α-GFP antibody correctly detected the large H2A-GFP fusion protein in LGY0016 (lane 4) but not in LGY0012 (lane 3, negative control). The α-H2A antibody detected H2A-GFP in LGY0016 and H2A in LGY0012. (D) DIC and GFP fluorescence confocal microscopy for LGY0016 cells. In the left panel, the GFP signals are overlaid in green. In the right panel, GFP signals are rendered with inverted contrast. (E) Map of the HTA1-HTB1 locus before (upper) and after (lower) the GFP insertion, with validation primers indicated. The color scheme is the same as in panel A. Note that LGY0015 still has the HTA2-HTB2 gene pair and therefore expresses untagged H2B. (F) Validation PCR for LG0015. (G) Validation immunoblots for strains BY4741 (wild type) and LGY0015 (expresses H2B and H2B-GFP). The α-GFP detected a ∼40 kDa protein (H2B-GFP) in LGY0015 but not BY4741. The α-H2B antibody detects the 40 kDa (H2B-GFP) in LGY0015 but not BY4741. Note that the α-H2B antibody has poorer specificity and does not generate a strong signal for the untagged H2B. (H) In the left panel, the GFP signals are overlaid in green. In the right panel, GFP signals are rendered with inverted contrast.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_figs15.tif/full/max/0/default.jpg)
Overview of LGY0016 (H2A-GFP) nuclear lysate
Tomographic slice (12 nm) of LGY0016 nuclear lysates. Some non-chromatin features are indicated: carbon support film (carbon), membrane fragments (membrane), gold fiducial (Au), and virus-like particle (VLP). The abundant granular densities in this field of view are nucleosomes. One small subarea (boxed) is enlarged 3-fold in the inset. A nucleosome-like particle (n) is indicated.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_figs16.tif/full/max/0/default.jpg)
Overview of LGY0015 (H2B, H2B-GFP) nuclear lysate
Tomographic slice (12 nm) of LGY0015 nuclear lysates. Some non-chromatin features are indicated: carbon support film (carbon), membrane fragments (membrane), gold fiducial (Au), and virus-like particle (VLP). The abundant granular densities in this field of view are nucleosomes. One small subarea (boxed) is enlarged 3-fold in the inset. A nucleosome-like particle (n) is indicated.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_figs17.tif/full/max/0/default.jpg)
Classification of LGY0015 (H2B, H2B-GFP) nuclear lysates
(A) Class averages (2-D) of nucleosome-like particles. Out of 96,979 template-matched particles from 6 tomograms, 56,872 particles were selected by 2-D classification. (B) Class averages (3-D) of two types of LGY0015 nucleosomes. The solid arrowhead indicates the position of the GFP tag whereas the open arrowheads indicate the positions that lack this density. These are the same class averages shown in the overall workflow in Figure S19. (C) Two views of the nucleosome crystal structure, indicating the location of the H2B C-terminus (salmon).
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_figs18.tif/full/max/0/default.jpg)
Direct 3-D classification of LGY0016 (H2A-GFP) lysates
(A) Class averages (3-D) of nucleosome-like particles from nuclear lysates of LGY0016 cells. The canonical nucleosome-like class averages are shaded blue while the non-canonical nucleosome averages are shaded gray. Movie S3 shows the convergence of the classification and shows more views of these class averages. (B) The second round of 3-D classification, using the canonical nucleosomes from panel A. See Movie S4 for more details of this classification job. (C) Refined densities of three types of LGY0016 nucleosomes isolated from nuclear lysates; reproduced from Figure 3B. The class numbers correspond to those in panel B (blue text) while the number of particles per class are labeled in black. The Fourier shell correlation (FSC) plot of the three refined class averages is labeled with the same numbering scheme. The resolution is ∼26 Å by the FSC = 0.5 criterion.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_figs19.tif/full/max/0/default.jpg)
Direct 3-D classification of LGY0015 (H2B, H2B-GFP) lysates
(A) Class averages of nucleosome-like particles from nuclear lysates of LGY0015 cells. The canonical nucleosome class averages are shaded blue while the non-canonical nucleosome averages are shaded gray. (B) The second round of 3-D classification, using the canonical nucleosomes from panel A. Note that some class averages, such as the gray one, have a density that is not connected to the nucleosome. This density is from a nearby particle that protruded into the mask. (C) The two types of LGY0015 nucleosomes are reproduced from Figure S17B. The numbers of particles per class are indicated below the density maps in black. The FSC plot of the LGY0015 nuclear lysates nucleosome class averages in panel D uses the same numbering scheme. The resolution is ∼26 Å by the FSC = 0.5 criterion.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_figs20.tif/full/max/0/default.jpg)
Overview of a LGY0016 (H2A-GFP) cryolamella
Tomographic slice (12 nm) of a LGY0016 cryolamella. Some non-chromatin features are indicated: nuclear envelope (NE), nuclear microtubule (MT), megacomplex (M), and ribosome (R). The inset is a threefold enlargement of the boxed area. A nucleosome-like particle (n) is indicated.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_figs21.tif/full/max/0/default.jpg)
Overview of a LGY0016 (H2A-GFP) cryolamella, VPP data
Tomographic slice (12 nm) of a LGY0016 cryolamella. The nuclear envelope (NE), nuclear microtubule (MT), and a megacomplex (M) are indicated. The inset is a threefold enlargement of the boxed area. A nucleosome-like particle (n) is indicated.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_figs22.tif/full/max/0/default.jpg)
Overview of a LGY0016 (H2A-GFP) cryosection, VPP data
Tomographic slice (12 nm) of a LGY0016 cryosection. Some non-chromatin features are indicated: nuclear envelope (NE), megacomplex (M), and ribosome (R). The inset is a threefold enlargement of the boxed area. A nucleosome-like particle (n) is indicated. The smeared features at the lower left are back-projection artifacts from image borders.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_figs23.tif/full/max/0/default.jpg)
Direct 3-D classification of LGY0016 (H2A-GFP) cryolamellae densities
Class averages (3-D) of nucleosome-like particles in cryotomograms of LGY0016 cryolamellae. The starred classes have two linear motifs.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_figs24.tif/full/max/0/default.jpg)
Direct 3-D classification of LGY0016 (H2A-GFP) cryosection densities in VPP data
Class averages (3-D) of nucleosome-like particles in VPP cryotomograms of LGY0016 cryolamellae. The starred classes have two linear motifs.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_figs25.tif/full/max/0/default.jpg)
Control VPP subtomogram analysis of ribosomes in situ
(A) Candidate ribosomes were template matched from a single VPP tomogram of the cytoplasm from Figure S11, using a 25-nm-diameter sphere as a reference. Direct 3-D classification into ten classes led to six non-empty classes, of which two were of ribosomes. Note that template matching is insufficient to uniquely identify a macromolecular complex, particularly when the reference is a featureless body like a sphere or cylinder. A simple reference was used to reduce the model bias. Template matching done with featureless references generates substantial numbers of false positives, which are confidently removed by classification. Complexes are only considered as identified after the 3-D classification step produces recognizable class averages. Therefore, the term “candidate ribosomes” is used to describe the 3,816 template matching hits. (B) The 1,150 ribosome subtomograms were pooled and 3-D refined, producing an average at 28 Å resolution by the “Gold-standard” FSC (0.143 cutoff) criterion and 33 Å resolution with an FSC = 0.5 cutoff. (C) Comparison of the in situ ribosome subtomogram average (STA) with density maps simulated from the yeast ribosome crystal structure (Ben-Shem et al., 2011) at three different resolutions.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_figs26.tif/full/max/0/default.jpg)
Verification of (H3, H3-GFP) strains
(A) Map of the HHT1-HHF1 locus in the parent (Wild type) strain BY4741 and the H3-GFP strains LGY0002 (H3, H3-RIPGLIN-GFP), LGY0007 (H3, H3-0aa-GFP) and LGY0070 (H3, H3-GGSGGS-GFP). Primers used for PCR verification are indicated with the half arrow symbols. (B) Agarose gel of PCR amplicons expected from LGY0002, LGY0007 and LGY0070 genomic DNA, in which the HHT1 locus is tagged with GFP. The only difference between the strains is the linker separating HHT1 and the GFP, which is too small to observe a difference in PCR amplicon sizes. (C) Immunoblot analysis of strains LGY0002, LGY0007 and LGY0070. The α-GFP antibody correctly detected the large H3-GFP fusion protein in LGY0002 (lane 3), LGY0007 (lane 4) and LGY0070 (lane 5), but not in BY4741 (lane 2, negative control). The α-H3 antibody detected H3 in all lanes, but failed to detect H3-GFP in LGY0002, LGY0007 and LGY0070. (D) DIC and GFP fluorescence confocal microscopy for LGY0002, LGY0007 and LGY0070 cells. In the left panel, the GFP signals are overlaid in green. In the right panel, GFP signals are rendered with inverted contrast.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_figs27.tif/full/max/0/default.jpg)
Verification of (H4, H4-GFP) strain
(A) Map of the HHT1-HHF1 locus in the parent (wild-type) strain BY4741 and the H4-GFP strain LGY0071. Primers used for PCR verification are indicated with the half arrow symbols. (B) Agarose gel of PCR amplicons expected from LGY0071 genomic DNA, in which the HHF1 locus is tagged with GFP. (C) Immunoblot analysis of strain LGY0071. The α-GFP antibody correctly detected the large H3-GFP fusion protein in LGY0071 (lane 3), but not in BY4741 (lane 2, negative control). The α-H4 antibody detected H4-GFP in LGY0071 and H4 in both lanes. (D) DIC and GFP fluorescence confocal microscopy for LGY0002, LGY0007 and LGY0070 cells. In the left panel, the GFP signals are overlaid in green. In the right panel, GFP signals are rendered with inverted contrast.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_figs28.tif/full/max/0/default.jpg)
Direct 3-D classification of LGY0007 (H3, H3-0aa-GFP) nuclear lysates
(A) Class averages (3-D) of nucleosome-like particles from nuclear lysates of LGY0007 cells, which express a wild-type copy of H3 and an H3-GFP that has no linker sequence. The canonical nucleosome-like class averages with an additional GFP density are shaded blue while the non-canonical nucleosome averages and canonical nucleosome-like class averages without any additional density are shaded gray. (B) The second round of 3-D classification, using the canonical nucleosomes with an additional GFP density from panel A. The remove duplicates function of RELION was used to remove particles (defined as particles within a radius of 180 Å from another particle) from the indicated class average for the 3-D refinement. Twenty-seven duplicate particles were removed before refining. (C) Refined density of LGY0007 nucleosomes isolated from nuclear lysates. The number of particles in the chosen class is labeled in black. The resolution is ∼23 Å by the FSC = 0.5 criterion.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_figs29.tif/full/max/0/default.jpg)
Direct 3-D classification of LGY0002 (H3, H3-RIPGLIN-GFP) nuclear lysates
Class averages (3-D) of nucleosome-like particles from nuclear lysates of LGY0002 cells, which express a wild-type copy of H3 and an H3-GFP that has a RIPGLIN linker sequence. In this experiment, 2-D classification was bypassed.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_figs30.tif/full/max/0/default.jpg)
Direct 3-D classification of LGY0070 (H3, H3-GGSGGS-GFP) nuclear lysates
Class averages (3-D) of nucleosome-like particles from nuclear lysates of LGY0070 cells, which express a wild-type copy of H3 and an H3-GFP that has a GGSGGS linker sequence. In this experiment, 2-D classification was bypassed.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_figs31.tif/full/max/0/default.jpg)
Direct 3-D classification of LGY0071 (H4, H4-GGSGGS-GFP) nuclear lysates
Class averages (3-D) of nucleosome-like particles from nuclear lysates of LGY0071 cells, which express a wild-type copy of H4 and an H4-GFP that has a GGSGGS linker sequence. In this experiment, 2-D classification was bypassed.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_figs32.tif/full/max/0/default.jpg)
Overview of a LGY0007 (H3, H3-GFP) cryolamella
Tomographic slice (12 nm) of a LGY0007 cryolamella. Some non-chromatin features are indicated: nuclear envelope (NE), megacomplex (M), and ribosome (R). The inset is a threefold enlargement of the boxed area. A nucleosome-like particle (n) is indicated.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_figs33.tif/full/max/0/default.jpg)
Direct 3-D classification of LGY0007 (H3, H3-GFP) cryolamellae densities
Class averages (3-D) of nucleosome-like particles in cryotomograms of LGY0007 cryolamellae. In this experiment, 2-D classification was bypassed.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_figs34.tif/full/max/0/default.jpg)
Simulated tomographic slices of single and stacked nucleosomes
Atomic models (upper, from PDB 1KX5) were used to simulate tomographic slices (12 nm, lower). (A and B) Mononucleosomes in the gyre and side views, respectively. (C and D) Two examples of stacked nucleosomes. In panel C, both nucleosomes have the same orientation. In panel D, the upper nucleosome is rotated 90° along the Y axis relative to the lower nucleosome. The stacked nucleosomes were separated by 55 Å center-to-center along the Y axis to emulate a worst-case (for image processing) scenario. Note that these simulated tomographic slices are not intended to accurately model the image-formation process. They serve to show how different side/gyre views of stacked nucleosomes appear.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_figs35.tif/full/max/0/default.jpg)
Fourier power spectra analysis of tilt series images
An image corresponding to a ∼12° pre-tilt (to make the lamella perpendicular to the electron-optical axis to maximize the signal-to-noise ratio) was extracted from tilt series 20211117_004 and then Fourier transformed using (A) CTFFIND and (B) IMOD ctfplotter. Thon rings could not be detected in the CTFFIND 2-D diagnostic image or the IMOD 1-D radial average of the Fourier power spectrum.
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_tbls1.tif/full/max/0/default.jpg)
Genotypes of strains used in this paper
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_tbls2.tif/full/max/0/default.jpg)
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_tbls2a.tif/full/max/0/default.jpg)
PCR primers, 5’ → 3’
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_tbls3.tif/full/max/0/default.jpg)
Antibodies for immunoblots
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_tbls4.tif/full/max/0/default.jpg)
Confocal microscopy details
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_tbls5.tif/full/max/0/default.jpg)
. Cryo-ET details
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_tbls6.tif/full/max/0/default.jpg)
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_tbls6a.tif/full/max/0/default.jpg)
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_tbls6b.tif/full/max/0/default.jpg)
Cryotomogram details
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_tbls7.tif/full/max/0/default.jpg)
Total nucleosome-like particles analyzed
![](https://prod--epp.elifesciences.org/iiif/2/87672%2Fv1%2Fcontent%2F438362v2_tbls8.tif/full/max/0/default.jpg)
Nucleus volume sampled for subtomogram analysis