Chromatin in a marine picoeukaryote is a disordered assemblage of nucleosomes
Chromatin organization is central to many conserved biological processes, but it is generally unknown how the underlying nucleosomes are arranged in situ. Here, we have used electron cryotomography to study chromatin in the picoplankton Ostreococcus tauri, the smallest known free-living eukaryote. By visualizing the nucleosome densities directly, we find that O. tauri chromosomes do not arrange into discrete, compact bodies or any other higher level of order. In contrast to the textbook 30-nm fiber model, O. tauri chromatin resembles a disordered assemblage of nucleosomes akin to the polymer melt model. This disorganized nucleosome arrangement has important implications for potentially conserved functions in tiny eukaryotes such as the clustering of nonhomologous chromosomes at the kinetochore during mitosis and the independent regulation of closely positioned adjacent genes.
The nucleosome hypothesis (Kornberg 1974) defined a fundamental unit of eukaryotic chromosome organization. While the structure of the nucleosome core particle is now known, it is still unclear how nucleosomes pack into the higher order chromatin structures that influence transcription, replication, and mitosis (Luger et al. 1997; van Steensel 2011). An electron microscopy study of purified chromatin lead to the 30-nm fiber model of chromatin organization, which proposed that nucleosomes pack into ordered helical fibers (Finch and Klug 1976). Two later electron cryomicroscopy (cryo-EM) studies of frozen-hydrated sections (cryosections) showed that chromatin organization can differ substantially depending on the cell type. One study did not find 30-nm fibers in mitotic Chinese hamster ovary cells (McDowall et al. 1986), but the other study did observe 30-nm chromatin fibers in starfish sperm, sea cucumber sperm, and purified chicken erythrocyte nuclei (Woodcock 1994). Recently, two groups searched for 30-nm fibers using 2D Fourier analysis of HeLa cell cryosections and 3D electron cryotomography (cryo-ET) of cryosections of purified chicken erythrocyte nuclei (Eltsov et al. 2008; Scheffer et al. 2011). They showed that HeLa cells do not have 30-nm fibers, but chicken erythrocytes do (though they appear short). The absence of 30-nm fibers has further been supported by groups using small-angle X-ray scattering and electron spectroscopic imaging of mammalian chromosomes and cells, respectively (Fussner et al. 2011, 2012; Joti et al. 2012; Maeshima et al. 2010; Nishino et al. 2012).
We study the smallest known free-living eukaryote, the picoplankton Ostreococcus tauri, as a model cell-biology system. O. tauri is a unicellular organism of the green lineage and has just one chloroplast, one mitochondrion, and a tiny nucleus that contains 20 linear interphase chromosomes (Courties et al. 1994; Derelle et al. 2006). Using cryo-ET of intact plunge-frozen cells, we found that each cell typically contained just one cytoplasmic microtubule (Henderson et al. 2007). This minimalistic ultrastructure suggested that further studies of O. tauri might reveal new principles of conserved cell-biological processes. For instance, when we imaged mitotic O. tauri cells by both cryo-ET of cryosections and room-temperature electron tomography of serial plastic sections, we found that each cell had only ~10 spindle microtubules, which was significantly fewer than the minimum 40 expected from textbook models (Gan et al. 2011). We therefore proposed that O. tauri might cluster kinetochores together to allow spindle microtubules to segregate more than one chromosome at a time.
To gain insights into how O. tauri chromatin is organized, here we have further analyzed our tomograms of interphase and mitotic O. tauri cryosections. Thirty-nanometer fibers were not seen: instead, in both interphase and mitotic cells, the nucleosome packing was patternless. Using a template-matching approach, we found that there is no large-scale reorganization indicative of condensation. O. tauri chromatin is therefore organized as a “polymer melt”—a disordered configuration with great flexibility (Eltsov et al. 2008; Maeshima et al. 2010). This chromatin model could explain how centromeres from multiple nonhomologous chromosomes, for instance, could cluster kinetochores and enable segregation by a smaller number of spindle microtubules, or how closely positioned adjacent genes could be independently regulated.
O. tauri chromatin is not organized as 30-nm fibers
O. tauri chromatin is disorganized
The 30-nm fiber model has been central to our understanding of eukaryotic chromosome biology (Alberts 2008; Lodish 2013). There are many variants of the 30-nm fiber model (Grigoryev and Woodcock 2012), but they all suggest that nucleosomes are packed in ordered arrays with fibers 25–40 nm wide (Dorigo et al. 2004; Robinson et al. 2006). Under special experimental conditions, chromatin fibers can be assembled with well-defined dimensions. These fibers have often been used in studies of chromatin structure and function at the first level beyond the nucleosome (Robinson and Rhodes 2006). Larger-scale chromatin organization is typically studied in the context of chromosomes or cells. A recent ultra-low angle X-ray scattering study of purified HeLa chromosomes found evidence of a “fractal” organization (Nishino et al. 2012), in agreement with earlier chromatin conformation capture and light microscopy studies of mammalian cells (Bancaud et al. 2009; Lieberman-Aiden et al. 2009). Fractal structures do not have a characteristic length scale; examples of fractal structures include self-similar motifs that span different length scales such as the textbook 10-nm, 30-nm, and 100 + −nm “chromonema” fiber-folding hierarchy (Belmont and Bruce 1994) as well as a polymer with a random walk path (Mirny 2011). While we did not see any structural evidence of self-similarity of higher order structures in our tomograms of either interphase or mitotic cells, they might be “fractal” in that the path of individual nucleosome strings might be random walks.
O. tauri chromatin has characteristics of a polymer melt
Polymer melt chromatin may facilitate conserved nuclear functions
Compared to 30-nm fiber chromatin (Fig. 5c), polymer melt-like chromatin is more compatible with key cell-biological processes in tiny nuclei. For example, we previously proposed that O. tauri might cluster mitotic chromosomes together at their kinetochores so mitosis could be completed in a single round of anaphase (Gan et al. 2011). Since kinetochores are nucleated by centromeric chromatin and positioned by pericentromeric chromatin (Blower et al. 2002; Marshall et al. 2008; Zinkowski et al. 1991), the chromatin path must make tight turns in order to cluster 20 kinetochores in a <1-μm nucleus. We speculate one possible configuration in which centromeric nucleosomes could cluster in a ring surrounding the spindle microtubules (Fig. 5d). Polymer melt chromatin could be flexible enough to make such turns because a loosely ordered beads-on-a-string nucleosome arrangement is likely to have a persistence length similar to naked dsDNA (<50 nm) (Brinkers et al. 2009). The polymer melt could therefore be instrumental to chromosome and kinetochore organization in some Trypanosome species, which may also have fewer kinetochores than chromosomes (Solari 1995). It is currently unknown if 30-nm fibers also exhibit such flexibility because their reported persistence length ranges from 30 to 220 nm (Bystricky et al. 2004; Cui and Bustamante 2000; Dekker et al. 2002; Kepper et al. 2008; Wedemann and Langowski 2002). While we have not yet visualized kinetochore distribution in O. tauri, kinetochore clustering has indeed been shown in yeasts (Appelgren et al. 2003; Jin et al. 2000). Genome-wide chemical mapping and two-color fluorescence light microscopy studies have argued, however, that budding yeast chromatin organizes as 30-nm fibers (Brogaard et al. 2012; Bystricky et al. 2004). It is therefore important to confirm whether or not 30-nm fibers do in fact exist in budding yeast and determine how interchromosomal interactions are mediated.
Chromatin organization also plays a role in transcriptional initiation by modulating the accessibility and positioning of both cis- and trans-regulatory elements. Transcriptional regulation models in humans must therefore take into account both the gene density and long intergenic sequences. In chromosome 11, for example, there are 10.6 genes/megabase and genes are separated by an average of 86 kb (Taylor et al. 2006). As illustrated in a compelling model of the β-globin locus (in human chromosome 11), an extended 30-nm fiber could act as a mechanical scaffold that loops in order to position an array of RNA polymerase II complexes (a “reading head”) on the coding sequences located several kilobases away (Wong et al. 2009). Such a transcriptional-regulation mechanism does not appear plausible in O. tauri and possibly other eukaryotes due to their much higher gene density (Derelle et al. 2006). Since 80 % of the O. tauri genome codes for genes and these are separated on average by less than 200 bp, the gene density is 100-fold larger than the human average. The typical gene would therefore span 1.3 kb—just 6.5 nucleosomes, with a single additional nucleosome separating adjacent genes. In the context of the 30-nm fiber, the average O. tauri gene would span less than one helical turn! According to the β-globin model, activating just one O. tauri gene could then force the shutdown of tens of other genes, which we find unlikely. Polymer melt chromatin would allow independent regulation of each gene in O. tauri and therefore finer control of transcriptional programs. A recent study furthermore showed that due to its flexible nature, polymer melt chromatin can even facilitate chromatin accessibility in both interphase and mitotic chromosomes (Hihara et al. 2012). Other gene-dense organisms may use polymer melt chromatin for similar purposes.
It remains unknown just how common 30-nm fiber-like chromatin is among eukaryotes. There is now an incipient consensus that somatic mammalian cells (Chinese hamster ovary; HeLa; mouse embryonic fibroblasts, spleen, and liver) do not have 30-nm fibers; instead they pack chromatin either as a polymer melt or as 10-nm fibers (Eltsov et al. 2008; Fussner et al. 2012; McDowall et al. 1986; Nishino et al. 2012). In contrast, specialized transcriptionally silent eukaryotic cells (chicken erythrocyte, starfish, and sea cucumber sperm) have 30-nm fibers as the predominant form of chromatin (Scheffer et al. 2011; Woodcock 1994). Here we have presented evidence that a unicellular picoeukaryote also packs chromatin as a polymer melt. Since the O. tauri cells analyzed here were isolated from both interphase and mitotic cultures, they are most analogous to somatic cells in higher eukaryotes. Cells with substantially differing gene densities and spatial and evolutionary constraints can therefore package chromatin without using 30-nm fibers. Unlike higher eukaryotes, however, O. tauri does not undergo large-scale chromosome condensation during mitosis. In other words, O. tauri chromatin does not reorganize into discrete chromatids separated by large cytoplasmic spaces; for an example of condensed mitotic chromatin, see Fig. 2 in Eltsov et al. (2008). It is possible that this picoplankton does not have the genes needed to condense chromosomes, which may also be needed to achieve 30-nm fibers, a form of local chromatin condensation. To test if the features of O. tauri chromatin are unique to this “untypical” organism or an adaptation, our analyses should be applied to other tiny eukaryotes that also have high gene density and small nuclear size.
Materials and methods
Cell preparation and electron cryotomography
Details of cell culture, synchronization, freezing, cryosectioning, imaging, and tomographic reconstruction are described (Gan et al. 2011). In summary, cell cultures of strain RCC745 were grown in artificial seawater and were naturally synchronized to a 12-h light/12-h dark cycle. To enrich for mitotic cells, cultures were arrested sequentially with hydroxyurea and propyzamide. Cells released from propyzamide were able to complete mitosis, progressing through prometaphase, metaphase, and anaphase. These cells could also be arrested in metaphase by treatment with MG132. The mitotic cells analyzed in this study were isolated after release from propyzamide treatment. Cell cultures were then mixed with 22 % dextran (an extracellular cryoprotectant) and 10-nm colloidal gold (fiducial markers for tomographic image alignment) and then rapidly frozen in an HPM-010 high-pressure freezer (Leica Microsystems). Cryosections were cut using an EM-UC6/FC6 cryoultramicrotome (Leica Microsystems) at −145 °C with a nominal feed of 130–150 nm. Ribbons of cryosections were controlled with a micromanipulator (Leitz model “M,” Leica Microsystems) and secured onto a C-flat CF422C-T grid (Protochips, Inc.). Tomographic imaging was done on a FEI “Polara” electron cryomicroscope, operated at 300 kV and −193 °C. Tilt series images were recorded using UCSF Tomo or Leginon (Suloway et al. 2009; Zheng et al. 2004) at a magnification of 18,000 or 22,500, corresponding respectively to 1.26 or 0.96 nm pixels at the specimen level. Nominal underfocus values ranged from 8 to 10 μm, which places the first CTF zero between 4 and 4.5 nm resolution. The nominal tilt range was ±66°, tilt increment was 1.5° or 2°, and cumulative dose was 100–140 electrons/Å2. Tomograms were reconstructed using IMOD (Kremer et al. 1996; Mastronarde 1997).
Tomographic slices were generated using IMOD (Kremer et al. 1996; Mastronarde 1997) and both Fourier transforms and rotational averaging of the amplitudes was done using ImageJ (Schneider et al. 2012). Since knife marks and crevasses at the cryosection surfaces could interfere with this analysis (Dubochet et al. 2007), we used subvolumes containing only the interior of the cryosection.
Nucleosomes were located automatically using the template-matching function in Jsubtomo (www.opic.ox.ac.uk/jsubtomo) (Huiskonen et al. 2010). An initial search was done using a low correlation cutoff—typically less than 0.05. Hit lists having higher correlation cutoffs were then created manually. For example, this list can be filtered so that the number of hits is less than, equal to, or greater than the number of nucleosomes calculated from other types of data (see “Nucleosome concentration calculations” section). Positions corresponding to each hit were mapped to a 3D volume using the IMOD program “point2model.” The hits were then visualized with the tomographic densities using the IMOD programs “3dmod” and “slicer.”
Atomic models of polynucleosomes were adjusted manually using UCSF Chimera (Pettersen et al. 2004) and tomograms were simulated using Bsoft (Heymann and Belnap 2007) following our previous protocol (Gan et al. 2008).
Micrococcal nuclease digests
O. tauri nucleosome-repeat length was determined using the classic micrococcal nuclease digestion experiment (Hewish and Burgoyne 1973). O. tauri cells were pelleted and then resuspended with digestion buffer (50 mM Tris, pH 8, 5 mM CaCl2, 0.1 % saponin, 25 μg/ml RNase A). Next, micrococcal nuclease (Roche Applied Science) was added and incubated at room temperature for 5 min. The reaction was stopped by the addition of and incubation with 0.2 mg/ml Proteinase K for 2 min, followed by the addition of one volume of 2× Stop Buffer (20 mM Tris, pH 7.5, 20 mM EDTA, and 20 mM EGTA). Digested DNA was purified using the DNeasy blood & tissue kit (Qiagen) according to the manufacturer’s instructions. DNA bands were then resolved in a 1.5 % agarose/0.5× Tris–borate–EDTA gel, electrophoresed at 50 V for 3 h, and visualized with FloroSafe stain (1st BASE). DNA bands were photographed in a G:BOX gel imager (Synoptics, Ltd.). Since limited micrococcal nuclease digestion products migrate as a DNA ladder corresponding to oligonucleosomes, the nucleosome-repeat length was determined by least-squares linear-regression analysis of log(band size) versus nucleosome number. The mono- and dinucleosomes were excluded from the analysis because they undergo more extensive digestion, producing shorter DNA fragments that, when included in the linear fit, result in a longer apparent nucleosome-repeat length.
Nucleosome concentration calculations
To calculate the nucleosome concentration, we used the haploid nuclear genome size (12.56 Mb) known from genome sequencing (Derelle et al. 2006; Grimsley et al. 2010). We then used the O. tauri nucleosome-repeat length (198 ± 6 bp, Fig. S2) to convert the genome size to nucleosomes (63,000). Finally, we used the volume of early interphase nuclei (0.155 ± 0.02 femtoliter), which we previously measured in our tomograms of intact O. tauri cells (Henderson et al. 2007). From these data, we estimate that early interphase cells pack nucleosomes at 680 ± 90 μM average concentration. We could not estimate the nucleosome concentration of mitotic cells because the contrast of these large plunge-frozen cells was too low to permit accurate segmentation of the nuclei. To calculate the local nucleosome concentration from Daban’s DNA concentration estimates, we assumed that the average base pair has a molecular weight MWbp = 610 Da and that a typical nucleosome wraps 200 bp of DNA. The local DNA concentration, ρDNA, was defined by Daban as “the mass of DNA per unit volume of the structure that contains it” (Daban 2003). For interphase chromosomes, ρDNA = 100 g/l (0.1 g/ml). Based on these values, the molar concentration per base pair is Cbp = ρDNA/MWbp or 0.16 M. The local nucleosome concentration is therefore 0.16/200 = 8.2 × 10−4 mol/l or 820 μM.
Figures were created using 3dmod (Figs. 1, 2, 3, 4, and S1), UCSF Chimera (Figs. 5 and S1), Adobe Illustrator (Fig. 5), Microsoft Excel (Figs. 1 and S2), and Syngene (Fig. S2) and arranged using Adobe Photoshop.
We thank Drs. J. Huiskonen for advice on Jsubtomo, M. Swulius for discussions on template matching, H. Wong and J. Mozziconacci for sharing their 30-nm fiber model, and Drs. A. McDowall and D. Rhodes for discussions on chromatin. This work was supported by the Howard Hughes Medical Institute and the Gordon and Betty Moore Center for Integrative Study of Cell Regulation. MSL was supported by NIH grant 2 R37 AI041239-06A1 to P. Björkman. LG was supported by a fellowship from the Damon Runyon Cancer Research Foundation (DRG-1940-07) and startup funds from NUS.
- Alberts B (2008) Molecular biology of the cell. Garland Science, New YorkGoogle Scholar
- Derelle E, Ferraz C, Rombauts S, Rouzé P, Worden AZ, Robbens S, Partensky F, Degroeve S, Echeynié S, Cooke R, Saeys Y, Wuyts J, Jabbari K, Bowler C, Panaud O, Piégu B, Ball SG, Ral J-P, Bouget F-Y, Piganeau G, De Baets B, Picard A, Delseny M, Demaille J, Van de Peer Y, Moreau H (2006) Genome analysis of the smallest free-living eukaryote Ostreococcus tauri unveils many unique features. Proc Natl Acad Sci USA 103:11647–11652PubMedCrossRefGoogle Scholar
- Fussner E, Strauss M, Djuric U, Li R, Ahmed K, Hart M, Ellis J, Bazett-Jones DP (2012) Open and closed domains in the mouse genome are configured as 10-nm chromatin fibres. EMBO reportsGoogle Scholar
- Lieberman-Aiden E, van Berkum NL, Williams L, Imakaev M, Ragoczy T, Telling A, Amit I, Lajoie BR, Sabo PJ, Dorschner MO, Sandstrom R, Bernstein B, Bender MA, Groudine M, Gnirke A, Stamatoyannopoulos J, Mirny LA, Lander ES, Dekker J (2009) Comprehensive mapping of long-range interactions reveals folding principles of the human genome. Science 326:289–293PubMedCrossRefGoogle Scholar
- Lodish HF (2013) Molecular cell biology. Freeman, New YorkGoogle Scholar
- Maeshima K, Hihara S, Eltsov M (2010) Chromatin structure: does the 30-nm fibre exist in vivo? Current Opinion in Cell BiologyGoogle Scholar
- Taylor TD, Noguchi H, Totoki Y, Toyoda A, Kuroki Y, Dewar K, Lloyd C, Itoh T, Takeda T, Kim DW, She X, Barlow KF, Bloom T, Bruford E, Chang JL, Cuomo CA, Eichler E, FitzGerald MG, Jaffe DB, LaButti K, Nicol R, Park HS, Seaman C, Sougnez C, Yang X, Zimmer AR, Zody MC, Birren BW, Nusbaum C, Fujiyama A, Hattori M, Rogers J, Lander ES, Sakaki Y (2006) Human chromosome 11 DNA sequence and analysis including novel gene identification. Nature 440:497–500PubMedCrossRefGoogle Scholar
Open Access This article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited.