Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
doi:10.1016/j.jmb.2003.09.015 J. Mol. Biol. (2003) 333, 907–916 Phase Diagram of Nucleosome Core Particles S. Mangenot1, A. Leforestier1, D. Durand2 and F. Livolant1* 1 Laboratoire de Physique des Solides, CNRS UMR 8502, Bât 510, Université Paris-Sud 91405 Orsay Cedex, France 2 LURE, Université Paris-Sud Bat. 209D, BP34, 91898 Orsay Cedex, France We present a phase diagram of the nucleosome core particle (NCP) as a function of the monovalent salt concentration and applied osmotic pressure. Above a critical pressure, NCPs stack on top of each other to form columns that further organize into multiple columnar phases. An isotropic (and in some cases a nematic) phase of columns is observed in the moderate pressure range. Under higher pressure conditions, a lamello-columnar phase and an inverse hexagonal phase form under low salt conditions, whereas a 2D hexagonal phase or a 3D orthorhombic phase is found at higher salt concentration. For intermediate salt concentrations, microphase separation occurs. The richness of the phase diagram originates from the heterogeneous distribution of charges at the surface of the NCP, which makes the particles extremely sensitive to small ionic variations of their environment, with consequences on their interactions and supramolecular organization. We discuss how the polymorphism of NCP supramolecular organization may be involved in chromatin changes in the cellular context. q 2003 Elsevier Ltd. All rights reserved. *Corresponding author Keywords: nucleosome; phase diagram; ordered phases; supramolecular organization; chromatin Introduction The eukaryotic genome is packaged into chromatin. Recent advances have revealed that chromatin structure is highly dynamic and subject to reversible changes in higher-order folding and nucleosome positioning. The structural changes are largely mediated by enzymatic covalent modifications of DNA and of the flexible N-terminal amino acid residues of the core histones and by non-covalent alterations of nucleosome architecture driven by ATP-dependent chromatin remodeling enzymes.1 – 3 Nevertheless, we seriously lack structural data about these multiple local changes of chromatin organization that occur locally (at the scale of a gene or a group of genes) inside the living cell. To overcome the difficulty of analyzing the structural details of chromatin organization and their changes in situ, simplified experimental models can be used to explore the multiple interactions and possible supramolecular organizations that chromatin elementary units, the nucleosome core particles (NCPs), can form over a Abbreviations used: NCP, nucleosome core particle; EM, electron microscopy; PEG, polyethylene glycol. E-mail address of the corresponding author: [email protected] large range of experimental conditions. 3D crystals obtained with NCP reconstituted from recombinant DNA and histones were used to determine the atomic structure of the NCP.4 – 7 Information on the interactions between the particles has also been collected by the analysis of the contacts between NCPs inside these crystals. It was shown that these interactions depend highly on slight changes in the charges carried by the histone tails.8 Although interactions between NCPs inside the crystals may differ from the interactions that come into play inside the nucleus, they give us information about possible relative positioning of nucleosomes inside chromatin. However, the dramatic limitation of these crystallographic studies comes from the limited sets of experimental conditions that can be explored. To bypass this limitation, some years ago we began a systematic survey of the phases formed by the NCP in changing conditions of ionic strength and osmotic pressure. Multiple phases have been observed that cover a large range of monovalent salt and NCP concentrations. Most of them have been characterized precisely by combining optical and electron microscopy observations9 – 11 and X-ray diffraction experiments.12 Our goal here is to focus our interest on the richness of the phase diagram under conditions of concentration and 0022-2836/$ - see front matter q 2003 Elsevier Ltd. All rights reserved. 908 salt that may be interesting from a biological point of view. We set out to determine the broad conditions under which each phase forms, and where possible to follow interconversions between them. We show how slight changes in the ionic conditions may have tremendous effects on the interactions between particles and produce large changes in their supramolecular organization. Results Dense phases of NCP formed under low salt conditions Lamello-columnar phase of NCP The lamello-columnar phase is found for monovalent salt concentrations ranging from Cs ¼ 3.5 mM to Cs ¼ 25 mM and for pressures ranging from about 3 atm to 25 atm. Within this range of experimental conditions, multiple textures are observed in optical microscopy (Figure 1d –f). The evolution of these textures was followed by increasing the applied osmotic pressures at a salt concentration Cs ¼ 15 mM. Isolated tubes form first, under pressures ranging from 3 atm to 5 atm (Figure 1d). They progressively shorten and new connected tubes form at their extremities (5 – 10 atm) leading to the formation of dense spherulites (15 – 25 atm) (Figure 1e and f). The walls of the tubes are formed by coiling of a series of stacked layers, as seen in freeze-fracture EM when the fracture plane is normal to the axis of the tube (Figure 1c). These various textures are also observed independently of the presence of a dialysis membrane separating the NCP from the PEG. Whatever the observed textures, the phase is lamello-columnar. A first description of this phase was given using cryo-sections of vitrified material observed in cryoEM.10 It was further confirmed by X-ray diffraction analysis.12 The structure of this phase is sketched in Figure 1, in a perspective view (Figure 1a0 ) and in a section plane, normal to the plane of the layers (Figure 1a00 ). Bilayers of NCP columns (L) alternate with layers of solvent (stars in Figure 1b). NCPs are stacked on top of each other in the columns and oriented with their dyad axis more or less normal to the plane of the bilayer. The front sides of the NCP (F, with the DNA ends) are facing the solvent, while the NCP back sides (B) are oriented inwards the bilayer. As seen on the cryo-section of Figure 1b, the circular (or slightly elliptical) shape of the NCP is observed when the column is perfectly seen in top view. The fact that NCP cannot be recognized everywhere along a given bilayer reveals slight deviations of the orientation of the columns. The orientation of the columns may also be different in adjacent bilayers. We have discussed how these bilayers are most probably stabilized by attractive interactions mediated by the amino terminal tail of histone H2B.10 We suspect that the formation of these tubes and of the more complex textures presented Phase Diagram of Nucleosome Core Particles here originate from the chiral and electrostatic interactions that come into play between NCP columns in the bilayers and between bilayers themselves (unpublished results). The lamellar phase was studied by X-ray for monovalent salt concentrations Cs ranging from 15 mM to 25 mM and under pressures ranging from 5 atm to 25 atm. The scattering profiles (Figure 4, below) provide three types of information. (i) In the small q-range three diffraction peaks are characteristic of the lamellar organization with an inter-lamellar distance decreasing from 376 Å to 358 Å when the osmotic pressure is increased. (ii) Other diffraction peaks observed at higher q-values sign the existence of a bi-dimensional monoclinic ordering within each layer of a lamella. (iii) Three broad scattering maxima superimposed to the narrow diffraction peaks indicate that the NCPs are more disordered in some parts of the sample. From X-ray data the NCP concentration can be estimated to 280 – 320 mg/ml in the salt and PEG concentration ranges mentioned above. Inverse columnar hexagonal phase For higher pressures (above 25 atm), NCPs selforganize in a different way. Freeze-fracture electron microscopy does not reveal any lamellar structure any more. Instead, hexagonal patterns can be seen over large domains (Figure 2b0 ). In other regions of the same replicas, parallel columns of stacked NCP can be seen (Figure 2a0 ). Aligned side-by-side, these columns also form bilayers (Figure 2a0 , in the circle) but instead of extending laterally over long distances as they did in the lamellar phase, these bilayers form a honeycomb hexagonal network of parameter a (sketched in Figure 2a and b). The solvent is located in parallel channels separated from each other by the bilayers. The volume of these channels varies, as also the hexagonal parameter a to follow any change in the NCP concentration. As a matter of fact, the diameter of the NCP (11 nm), which is large compared to the dimensions of the hexagonal network, imposes discrete steps of variation of a. Multiple values of a have been measured: a < 38 nm, a < 45 nm and a < 54 nm (^ 2 nm). Two networks may coexist under equilibrium conditions. From these different values of a, and from the corresponding NCP organization (not shown), the NCP concentration was calculated to vary from about 320 mg/ml to 420 mg/ml in this phase. X-ray diffraction analyses will be necessary to get more information on the fine details of the structure. We named this phase “inverse hexagonal phase” by analogy with amphiphilic systems in which polar molecules also form bilayer structures. The textures of this phase, observed in polarizing microscopy, are consistent with the structure described above. Rigid rods, either isolated or organized into 3D spiky spheres (Figure 2c), or more classical hexagonal textures (not shown) can Phase Diagram of Nucleosome Core Particles 909 Figure 1. Lamello-columnar phase of NCP formed under low salt conditions (Cs , 25 mM). a– a00 , Sketch of the lamello-columnar structure. Nucleosomes are first drawn to identify their front (F) and back (B) sides, that correspond, respectively, to the entry and exit sites of their pseudo-dyad axis d. Nucleosomes are then drawn as small cylinders (in perspective view) or circles (in top view) with a black dot that identifies their front side (F) with the two free ends of DNA. Nucleosomes are stacked on top of each other to form columns and these columns are aligned in parallel to form bilayers (L). The front sides of the NCP, with the two free ends of the DNA strand, face the solvent layer separating the bilayers, and the back sides (B) are oriented inwards to the bilayers. Bilayers themselves form a lamellar structure with a period dL. b, Cryo EM section normal to the plane of the bilayers (as drawn in a00 ). Bilayers (L) are separated by solvent layers (p ). NCPs can be seen in top views in regions where the axis of the columns is perfectly normal to the section plane. Their shape is slightly elliptical, revealing that they are slightly tilted in the columns. c, A stacked series of superimposed bilayers, with a periodicity dL, is coiled to form a hollow cylinder, seen almost along its axis on a EM freeze-fracture replica. d – f, Textures of the lamello-columnar phase formed under increasing osmotic pressure conditions (PEG 19 – 35%): isolated tubes in side view (d), tubes budding new tubes at their extremities (d and e), and spherulites (f). d, e and f, Interferential Nomarski contrast. be seen. Between crossed polars, the brightness of each rod is intense and its extinction occurs at once for given orientations of the rod, which reveals a unidirectional orientation of the NCP columns in these monodomains. Dense phases of NCP formed under high salt conditions For salt concentrations ranging from 50 mM to 160 mM, and under applied pressures varying from 4.7 atm to 13 atm, a 3D crystal forms as indicated by X-ray diffraction patterns (Figure 3).12 Columns of stacked NCP are aligned and closely packed to form a columnar quasi-hexagonal structure, as sketched in Figure 3a – a00 . In a first approximation the crystalline cell is orthorhombic, like that observed in crystals obtained with reconstituted NCP in the presence of divalent ions.4,13,14 Some additional Bragg peaks could result from a superstructure along the column axis or from a slight distortion of the orthorhombic unit cell. Increasing the osmotic pressure from 4.7 atm to 13 atm produces similar reductions of all three 910 Phase Diagram of Nucleosome Core Particles Figure 3. Dense phases of NCP prepared under high salt conditions (Cs . 50 mM NaCl). a, Columns align in parallel in a close-packing arrangement, either hexagonal (a00 ) or orthorhombic (quasi-hexagonal) (a0 ). b and c textures observed in polarizing microscopy: typical hexagonal germs with their six branches (noted 1 –6) (b) or hexagonal textures (c). Figure 2. Inverse hexagonal phase, found under high pressure, and under low salt conditions (Cs , 25 mM). a and b, Schematic views of the hexagonal structure of parameter a with the corresponding patterns observed in freeze-fracture EM in planes either parallel (a0 ) or normal (b0 ) to the orientation of the columns. a0 , One can recognize the stacking of NCP in the columns (arrow), and the association of two series of columns to form one bilayer (in the circle). The structure of each bilayer is similar to that described in Figure 1a and b, but it does not extend laterally over large distances. The bilayers separate the solvent channels. In b0 the solvent channels (red dots) form an hexagonal network, that extends over large distances. The bilayer structure of the walls that separate these channels (drawn in b) is not resolved here. This inverse hexagonal phase may appear in the form of rigid birefringent rods in polarizing microscopy (c). orthorhombic parameters: a, b and c decrease from 115.4 Å, 203.6 Å and 119.0 Å to 110.1 Å, 194.0 Å and 112.7 Å, respectively. A 2D hexagonal mesophase may also form under identical pressure and salt conditions: we found that the initial NCP concentration before addition of the PEG solution determines the formation of the 2D or 3D columnar phase (see Mangenot et al.12 for details). The lateral packing of the NCP columns is then perfectly hexagonal (Figure 3a and a00 ), and there is no correlation between the longitudinal order of NCP along the column and the lateral organization of columns. Whatever the 2D or 3D ordering of NCP in these phases, the final NCP concentrations are quite the same, close to 500 mg/ml, under a pressure of 4.7 atm. This concentration may rise up to about 610 mg/ml under higher pressures (23.5 atm) at the expense of a good ordering. The important point we raise here is that different final states can be obtained depending on the story of the sample preparation, while all final concentrations are identical. We reported previously the formation of a columnar phase under high salt conditions, based upon optical and electron microscopy observations.9,11 Spectacular macroscopic hexagonal domains form, that further divide into six branches (Figure 3b), as a consequence of the chiral structure of the NCP.11 We suspect that these germs are 3D quasi-hexagonal crystals because their textures and defects are not characteristic of 2D mesophases. They also lack any fluidity and never fuse together. More rarely, more classical hexagonal textures are also observed (Figure 3c). We assume that these latter textures correspond to the 2D mesophase. Micro-focus diffraction experiments on selected regions of both samples would be necessary to assign a structure to each texture unambiguously. 911 Phase Diagram of Nucleosome Core Particles Phases formed at intermediate salt concentrations The organization of NCP is more complex in the intermediate range of salt concentration From X-ray diffraction (25 , Cs , 50 mM). experiments (Figure 4), we observed that a dense phase of columns forms progressively without any long-range ordering between columns and with NCPs irregularly piled along a column. Further on, de-mixing progressively takes place and several months later the lamello-columnar phase and the 2D hexagonal (or 3D quasi-hexagonal) phases, described above under low salt and high salt conditions, respectively, are found to coexist.12 Samples prepared under similar conditions and observed in the polarizing microscope reveal unexpected textures due to microphase separation effects. After a complex process of organization (unpublished results), a superimposed series of layers forms parallel to the observation plane. They are made of platelets with a puzzle-piece shape (Figure 5b and b0 ). It is remarkable that the limits of the domains do not superimpose but are quite systematically out of phase in the successive layers. At a higher magnification, thin tubes (the same as described in the low salt lamello-columnar phase) can be seen on both surfaces of the layers. Figure 5. Microphase separation effect observed in the intermediate salt range (Cs ¼ 37.5 mM) in flat capillaries (0.2– 0.4 mm thick) and observed with interferential Nomarski contrast as sketched in a. b, b0 , The same region of the preparation observed at two different foci, showing puzzle-pieces of the hexagonal phase separated from one another by a layer of solvent. An enlarged view of the surface of these plates reveals a network of tubes of the lamello-columnar phase (c). Their density may be rather low, or quite high as shown in Figure 5c. The tubes do correspond unambiguously to the lamello-columnar structure and we suppose from the X-ray data12 that the layers correspond to the (quasi) hexagonal phase, although they are not characteristic of hexagonal textures. This organization of these two phases into micro domains is sketched in Figure 5a. Isotropic to ordered phases transition Figure 4. Profiles of the integrated intensities IðqÞ for different salt concentrations: 25 mM (lamello-columnar phase), 160 mM (3D orthorhombic crystal), 50 mM (2D hexagonal phase) and 37 mM (biphasic sample before and after phase separation). In a previous work10 we described, by cryoEM of thin films, how NCPs stack to form isolated columns which themselves form an isotropic solution, under low salt conditions (, 25 mM) at a concentration just below the transition to the lamello-columnar phase. NCP solutions prepared for Cs ¼ 15 mM and for NCP concentrations ranging from 5 mg/ml to 250 mg/ml (in the absence of PEG) give IðqÞ profiles slightly different for q , 0:05 Å21 that reflect the evolution of the structure factor with NCP concentration. For CNCP ¼ 250 mg/ml the curve is also slightly different around q ¼ 0.11 Å21 (Figure 6), which 912 Phase Diagram of Nucleosome Core Particles Phase diagram Figure 6. Scattered intensity IðqÞ of NCP solutions of concentration ranging from 5 mg/ml to 250 mg/ml, for two salt concentrations Cs ¼ 15 mM and Cs ¼ 160 mM. The upper curve was equilibrated against a 12% PEG solution at Cs ¼ 160 mM. suggest that a fraction of NCP self-assembled into short columns. At CNCP < 270 mg/ml, the lamellocolumnar phase is already formed. Therefore, isolated columns may exist under low salt conditions over a very narrow concentration range (between 250 mg/ml and 270 mg/ml). Under high salt conditions (Cs ¼ 160 mM), the formation of the columns was not seen for CNCP # 250 mg/ml (Figure 6). They are clearly visible in solutions to which a pressure of 1.6 atm (12% (w/v) PEG) was applied. The X-ray profile exhibits three broad maxima that are characteristic of a disordered phase of columns, with some disorder along the columns. The NCP concentration evaluated from the position of the first maximum is higher than 450 mg/ml (Figure 6). In the absence of samples prepared at intermediate NCP concentrations, we cannot predict the range of concentrations where the columns form. From the spectra analysis, we cannot predict whether the columns form an isotropic or a nematic phase. Optical microscopy observations show that both can exist. The isotropic phase turns birefringent under flow, which is further evidence of the formation of the columns, and the nematic phase is observed for slightly higher NCP concentrations, in a narrow domain of the phase diagram. The combination of experimental data coming from X-ray diffraction, optical and electron microscopy lead us to present the tentative NCP phase diagram given in Figure 7. More than 30 different experimental conditions (PEG and salt) have been analyzed. A larger range of salt conditions has been explored more qualitatively in the absence of applied pressure. The same structures were found using both methods. A remarkable self-assembling property of NCP is to form isolated columns, by stacking on top of each other, in the absence of any divalent cation and of DNA linking the NCPs together. These columns form before any lateral organization is set between NCPs and whatever the ionic conditions are. They are found in a narrow concentration range (250 , CNCP , 270 mg/ml) at low salt. We suspect this range to be larger under high salt conditions, although NCP concentrations were not determined precisely. These columns further align to form a columnar nematic phase. Under an osmotic pressure fixed to 4.7 atm, a series of ordered 2D or 3D-ordered phases are found: a lamello-columnar phase below 25 mM monovalent salt, a 3D orthorhombic crystal or a 2D columnar hexagonal phase above 50 mM. The lamello-columnar phase and the high salt phase (2D or 3D) coexists in the intermediate salt range, and a microphase separation occurs. An inverse hexagonal phase is found in the low salt range for pressures above 23 atm. We observed the formation of 3D crystals using NCP carrying DNA fragments with some polydispersity and longer than 146 bp. The average distance between columns is just slightly larger than in 3D crystals diffracting at high resolution.4,5,13 Nevertheless, the 3D packing is lost when DNA fragments become too long (170 bp) or too polydisperse. On the contrary, the lamellocolumnar and the inverse hexagonal phases afford significant variations of the DNA length carried by the NCP (at least 146– 170 bp). This is not surprising, since the free ends of the DNA may extend on both external faces of the bilayers without any geometrical constraint. We observe that the boundaries between these different phases, determined by ionic and pressures conditions, move with the DNA fragment length. This is due to the increase in the negative phosphate charges that changes the net charge of the NCP. For the same reasons, displacements of the phase boundaries are also observed when the histone tails of the NCP are partly removed. As a consequence, in samples containing NCP with a well-defined DNA length, half intact and half deleted of their H3 and H2B tails, a segregation occurs under low salt conditions: intact NCP form microdomains of the lamello-columnar phase while the other NCPs pack into 3D hexagonal crystals. 913 Phase Diagram of Nucleosome Core Particles Figure 7. Tentative phase diagram of isolated NCP as a function of monovalent salt (3 , Cs , 160 mM) and applied osmotic pressure. Discussion Phase diagrams and kinetic effects Despite our systematic survey of a large range of experimental conditions, we cannot exclude that other phases may exist. Theoretical phase diagrams, with predictions of different structures, could have been of some help to look for other possible phases but, unfortunately, only simple objects have been considered so far: spheres, rods, platelets, spherocylinders or even cut-spheres but with diameter to thickness ratios that do not correspond to NCPs.15 – 17 Moreover, discrepancies are often observed between predictions and observations because of the intricacies of the kinetics of the phase transitions.18 On top of that, structural details and heterogeneous charge distributions are not considered. Such details, that are crucial in the case of NCP, result in specific orientational ordering in the ordered phase and introduce more complex mechanisms in the kinetics of phase formation.19 As expected, owing to the complexity of the NCP, numerous difficulties have been encountered in the elaboration of the phase diagram. First, equilibration times may be extremely long (several months). Using PEG as a stressing polymer, the solvent was progressively extracted from the NCP solution and this process requires times to reach the equilibrium conditions because our samples are macroscopic (a few mm3). The other method that we have been using, i.e. the evaporation of the solvent, may lead much faster to the formation of the dense phases, with the limitation that the concentration of salt in the samples is not controlled any more. Whatever the method, a slow process of organization of the sample is absolutely necessary to obtain the organization of NCP in domains large enough to be observed in optical microscopy (a few mm). Anyway, we cannot certify that samples would not remain trapped in metastable states for years. For example, in the biphasic intermediate salt range, we cannot ascertain that the sample would not evolve further and that another phase could coexist with the lamellar and the hexagonal phases. Second, the formation of the dense phases was shown to depend on the initial states in the phase diagram. The initial concentration of NCP before addition of the PEG determined the formation of either a 2D or a 3D (quasi) hexagonal crystal. Third, we suspect that the stress produced by the addition of PEG (all at once) to the NCP solution may be of some importance in the kinetics of phase transitions. Indeed, it has been demonstrated how dramatically a stress can jam a system and restrict crystallization for years.20 Other methods of preparation of the samples should be tried to overcome such difficulties. Fourth, the formation of oligomers of NCP piled to form columns also interfere with the other parameters. Finally, microphase separation phenomena were also observed in the intermediate salt range. These are reminiscent of the complex structures described in solutions of virus rods in the presence of PEG.21,22 The routes by which the equilibrium phases are formed are thus complex and a lot more remains to be understood in the organization and kinetics of phase transitions of these NCP solutions,23 but this complex behavior may precisely offer extremely rich possibilities of adaptation of chromatin organization in the context of the living cell, as discussed below. Biological relevance To relate our present studies in vitro and the properties of chromatin in the cell nucleus, a lot remains to be done. Of course, the extremely long organization times reported here are not relevant from a biological point of view, for several reasons. First, the organization of NCP could not extend over long distances, but would be restricted to microdomains inside the nucleus. Second, the formation of the higher concentrated states in chromatin never starts from an isotropic solution but probably emerges from an already highly organized state in the interphase (that we do not know) through processes that involve not only physicochemical processes but also the activity of numerous enzymes and cofactors. These probably help chromatin organize in a sophisticated way, following an unknown path in the extremely complex phase diagram of chromatin. We are aware that our experimental system is extremely simplified and we do not claim that we reproduce all details of chromatin of the living cell. A lot more remains to be done to understand the phase transition processes. We may think, for example of a “memory” of the ordered dense 914 states that could be kept up to a certain point in the de-condensation process and help in a further reorganization of the dense states. Nonetheless, our hypothesis is that the different organizations observed here may exist as well locally in the chromatin of the living cell. Indeed, as detailed below, the explored range of salt and NCP concentrations conditions are biologically relevant, and some of these phases could be kept even in the presence of linker DNA and additional proteins. In the absence of reliable data on the local ionic conditions inside the cell nucleus, we explored a large range of monovalent salt concentrations (3.5 – 160 mM), that covers the values given for eukaryotic nuclei.24 – 26 The pressures that we apply range from 1.5 atm to 25 atm. This osmotic pressure is used here as a tool to reach the NCP concentrations that exist inside the cell nucleus (ranging from 100 mg/ml to 500 mg/ml NCP27,28). The cell is well known to be crowded29 but, to our knowledge, there is no experimental data available on the pressure applied by the molecular species forming the eukaryotic chromatin environment. Nevertheless, for comparison, a pressure of 1 atm maintains the bacterial nucleoı̈d at its initial volume after its extraction out of the bacteria30 and DNA is maintained inside the bacteriophage capsid under a pressure of about 30 – 50 atm.31,32 We also obtained the formation of condensed phases of NCPs by adding divalent or multivalent cations (Ca2þ, Mg2þ, spermidine3þ or spermine4þ for example) to dilute NCP solutions, in the absence of any applied pressure.33,34 In the context of the cell, all these cations and also charged proteins, may interact with nucleosomes in combination with the macromolecular crowding effects. The path followed by the DNA molecule inside the chromosome is not known, and despite the actual consensus about it, the 30 nm chromatin fiber observed in vitro may not be the only possible organization of the chromatin fiber in vivo. Other possible models may be considered and liquid crystalline states are good candidates. Numerous models may be proposed. As shown above, slight changes in the ionic condition are enough to induce phase transitions. For example, the NCP concentration drops from 500 mg/ml to 300 mg/ ml under a 5 atm pressure, when the monovalent salt concentration, Cs, changes from 50 mM to 25 mM salt. The accessibility of NCP changes accordingly. Similar phase transitions are likely to occur in the living cell, and may be useful to control the accessibility of enzymes to the DNA molecule. Together with the diversity of organization of the isolated NCP, we observe that the nature of the phase may depend on the initial state and on the path followed in the phase diagram. The kinetic problems reported here may be even more complex in chromatin, where NCPs are linked together and interact with other proteins and multiple kinds of ions. We guess that this complexity may offer extremely rich possibilities of adaptation of chromatin organization to Phase Diagram of Nucleosome Core Particles multiple local constraints in the context of the living cell. Local phase transitions may be triggered by local changes in ionic conditions or by changes in the distribution of charges on the histone tails, under the action of specific enzymes. Such transitions may be coupled to macromolecular transitions (helix-coil transition) as suggested by Samulski35 and/or be involved in synchronous expression of genetic information (euchromatin to heterochromatin). Materials and Methods NCP Most of the data presented here were obtained with two different batches of NCPs prepared from calf thymus chromatin. DNA fragments associated to the histone core were 155(^7) bp and 165(^ 10) bp, respectively. A few preparations were also made with NCP with exactly 146 bp associated DNA. The integrity of the histones was checked carefully for each batch. Stock solutions were extensively dialyzed against TE buffer (10 mM Tris –EDTA (pH 7.6)) at a concentration of 1 – 3 mg/ml and concentrated by ultrafiltration up to 150– 200 mg/ml in the same buffer. These stock solutions were stored at 0 8C. Sample preparation Aliquots of the stock NCP solutions were diluted and dialyzed against TE buffer eventually supplemented with NaCl. The concentration of monovalent ions in the dialysis buffer (Cs) (including Trisþ and Naþ) was adjusted to concentrations ranging from 3.5 mM to 160 mM. To prepare NCP solutions of different concentrations, while keeping constant the Cs values of the equilibration buffer, samples were prepared under controlled osmotic pressure. Different methods were used: for moderate pressures (0.1– 4 atm), (i) samples were placed in dialysis bags and immersed in a large volume of a neutral polymer solution (polyethylene glycol, Mr 20,000 from Sigma), prepared in the same buffer, with the same Cs concentration; or (ii) samples were equilibrated by ultrafiltration through a nitrocellulose membrane under nitrogen pressure. To reach higher pressures (4.7– 23 atm), the NCP solution was first concentrated by one of the two methods described above up to a concentration of about C ¼ 225– 260 mg/ml. Equilibrated samples for X-ray and microscopy experiments are prepared as detailed below. The osmotic pressures are related to the PEG concentration (calculated with the help of the empirical expression in Ref. 36: 12% PEG ¼ 1.6 atm; 16% PEG ¼ 3.1 atm; 19% PEG ¼ 4.7 atm; 23% PEG ¼ 7.6 atm; 25.5% PEG ¼ 10 atm; 28% PEG ¼ 13 atm; 35% PEG ¼ 23 atm. For microscopy and X-ray experiments, the following protocols were used. Microscopy A 20 ml sample of the NCP solution was introduced into flat capillaries (Vitro Dynamics). The PEG solution (with the same salt concentration, Cs) was then introduced in the capillary, pushing the NCP solution and creating a PEG gradient from one extremity of the 915 Phase Diagram of Nucleosome Core Particles capillary to the other. The two extremities of the capillary were sealed with wax and the specimen left to stabilize. We did not observe any change in the specimen after a few weeks. To explore rapidly the phase diagram, another method was used in some cases: drops of NCP solutions of defined salt and NCP concentration were deposited between slide and coverslip and left to concentrate progressively by slow dehydration. Salt and NCP concentrations corresponding to the observed textures were calculated a posteriori from the volume variation between the initial and final steps. Observations were made with a Nikon polarizing microscope equipped for interferential Nomarski contrast. For electron microscopy, drops of the phases prepared in flat capillaries or in dialysis bags were deposited onto gold discs and stabilized in a humid atmosphere. Drops of the same samples deposited onto glass slides were used as controls to check the presence of the expected textures in the polarizing microscope. Samples were frozen by projection onto a copper surface cooled down to 10 K with liquid helium (Cryovacublock Reichert). Freeze fracture was realized in a Balzers BAF 400T apparatus and replicas observed in a Philips CM12 TEM at 80 kV. Cryo-sections (40– 80 nm thick) were realized at 113 K under nitrogen atmosphere in a cryo-ultramicrotome (Leica) and observed at 100 K in a Philips CM12 cryo-TEM at 80 kV. The vitreous state of water was checked by electron diffraction, and images recorded in low-dose mode at a direct magnification of 45,000 £ and 600– 700 nm defocus. X-ray diffraction About 15– 20 ml of nucleosome solution and 200 ml of polymer solution were successively added into quartz capillaries ,1.5 mm in diameter and left to equilibrate for more than four weeks at room temperature. In some cases, in particular for the dense phases obtained at high salt, the sample-containing capillaries were placed in a magnetic field higher than 7 T immediately after addition of the PEG solution to the NCP solution and kept during the whole formation of the dense phase. It produced an alignment of the NCP columns that was kept after removal of the magnetic field and helped in the understanding of diffraction patterns. X-ray experiments were carried out on the D24 instrument installed on the storage ring LURE-DCI (Orsay, France) and on beam line ID2 at the European Synchrotron Radiation Facility (ESRF, Grenoble, France). On D24 (ID2) the wavelength was 1.488 Å (0.989 Å) and the sample-todetector distance was 2500 mm (4730 mm), respectively. These setups gave access to scattering vectors q (where q ¼ 4p sin u=l; 2u is the scattering angle) ranging from 0.01 Å21 to 0.25 Å21 (0.01 – 0.35 Å21) and to an instrumental resolution Dq (full width at half maximum FWHM value) of 0.00145 Å21 (0.0008 Å21). Acknowledgements This work has been supported by The National Center for Scientific Research (CNRS), by the French Research Department and the Research Association Research against Cancer (ARC) who provided one of us (S.M.) with a three year and a three month fellowship, respectively. References 1. Kornberg, R. D. & Lorch, Y. (2002). Chromatin and transcription: where do we go from here. Curr. Opin. Genet. Dev. 12, 249– 251. 2. Wu, J. & Grunstein, M. (2000). 25 years after the nucleosome model: chromatin modifications. Trends Biochem. Sci. 25, 619– 623. 3. Becker, P. B. (2002). Nucleosome sliding: facts and fiction. EMBO J. 21, 4749– 4753. 4. Luger, K., Mader, A. W., Richmond, R. K., Sargent, D. F. & Richmond, T. J. (1997). Crystal structure of the nucleosome core particle at 2.8 Å resolution. Nature, 389, 251– 260. 5. Harp, J. M., Hanson, B. L., Timm, D. E. & Bunick, G. J. (2000). Asymmetries in the nucleosome core particle at 2.5 Å resolution. Acta Crystallog. sect. D Biol. Crystallog. 56, 1513– 1534. 6. White, C. L., Suto, R. K. & Luger, K. (2001). Structure of the yeast nucleosome core particle reveals fundamental changes in internucleosome interactions. EMBO J. 20, 5207–5218. 7. Davey, C. A., Sargent, D. F., Luger, K., Maeder, A. W. & Richmond, T. J. (2002). Solvent mediated interactions in the structure of the nucleosome core particle at 1.9 Å resolution. J. Mol. Biol. 319, 1097 –1113. 8. White, C. L., Suto, R. K. & Luger, K. (2001). Structure of the yeast nucleosome core particle reveals fundamental changes in internucleosome interactions. EMBO J. 20, 5207–5218. 9. Leforestier, A. & Livolant, F. (1997). Liquid crystalline ordering of nucleosome core particles under macromolecular crowding conditions: evidence for a discotic columnar hexagonal phase. Biophys. J. 73, 1771 –1776. 10. Leforestier, A., Dubochet, J. & Livolant, F. (2001). Bilayers of nucleosome core particles. Biophys. J. 81, 2414 –2421. 11. Livolant, F. & Leforestier, A. (2000). Chiral discotic columnar germs of nucleosome core particles. Biophys. J. 78, 2716– 2729. 12. Mangenot, S., Leforestier, A., Durand, D. & Livolant, F. (2003). X-ray diffraction characterization of the dense phases formed by nucleosome core particles. Biophys. J. 84, 2570– 2584. 13. Davey, C. A., Sargent, D. F., Luger, K., Maeder, A. W. & Richmond, T. J. (2002). Solvent mediated interactions in the structure of the nucleosome core particle at 1.9 Å resolution. J. Mol. Biol. 319, 1097 –1113. 14. Harp, J. M., Hanson, B. L., Timm, D. E. & Bunick, G. J. (2000). Asymmetries in the nucleosome core particle at 2.5 Å resolution. Acta Crystallog. sect. D Biol. Crystallog. 56, 1513– 1534. 15. Frenkel, D. (1991). Statistical mechanics of liquid crystals. In Proceedings of the Les Houches Summerschool session on Liquids, Freezing and the Glass transition (Hansen, J., Levesque, D. & Zin-Justin, J., eds), pp. 689– 762, North-Holland, Amsterdam. 16. Veerman, J. & Frenkel, D. (1991). Relative stability of columnar and crystalline phases in a system of parallel hard spherocylinders. Phys. Rev. A, 43, 4334 –4343. 17. Blaak, R., Frenkel, D. & Mulder, B. (1999). Do 916 18. 19. 20. 21. 22. 23. 24. 25. 26. Phase Diagram of Nucleosome Core Particles cylinders exhibit a cubatic phase? J. Chem. Phys. 110, 11652– 11659. Anderson, V. J. & Lekkerkerker, H. N. (2002). Insights into phase transition kinetics from colloid science. Nature, 416, 811 – 815. Lomakin, A., Asherie, N. & Benedek, G. B. (1999). Aeolotopic interactions of globular proteins. Proc. Natl Acad. Sci. USA, 96, 9465– 9468. Kegel, W. (2000). Crystallization in glassy suspensions of colloidal hard spheres. Langmuir, 16, 939– 941. Adams, M., Dogic, Z., Keller, S. L. & Fraden, S. (1998). Entropically driven microphase transitions in mixtures of colloidal rods and spheres. Nature, 393, 349– 352. Adams, M. & Fraden, S. (1998). Phase behavior of mixtures of rods (tobacco mosaic virus) and spheres (polyethylene oxide, bovine serum albumin). Biophys. J. 74, 669– 677. Evans, R. M., Poon, W. C. & Renth, F. (2001). Classification of ordering kinetics in three-phase systems. Phys. Rev. E. Stat. Nonlin. Soft Matter Phys. 64, 031403. Richey, B., Cayley, D. S., Mossing, M. C., Kolka, C., Anderson, C. F., Farrar, T. C. & Record, M. T., Jr (1987). Variability of the intracellular ionic environment of Escherichia coli. Differences between in vitro and in vivo effects of ion concentrations on protein – DNA interactions and gene expression. J. Biol. Chem. 262, 7157–7164. Cameron, I. L. (1985). Electron probe x-ray microanalysis studies on the ionic environment of nuclei and the maintenance of chromatin structure. Prog. Clin. Biol. Res. 196, 223– 239. Bloomfield, V. (2000). Polymer and polyelectrolyte behaviour of DNA. In Nucleic Acids: Structures, Properties and Functions (Bloomfield, V., Crothers, D. M. & Tinoco, I. J., eds), University Science Books, Mill Valley, CA. 27. Bohrmann, B., Haider, M. & Kellenberger, E. (1993). Concentration evaluation of chromatin in unstained resin-embedded sections by means of low-dose ratio-contrast imaging in STEM. Ultramicroscopy, 49, 235– 251. 28. Daban, J. R. (2000). Physical constraints in the condensation of eukaryotic chromosomes. Local concentration of DNA versus linear packing ratio in higher order chromatin structures. Biochemistry, 39, 3861– 3866. 29. Minton, A. P. (2001). The influence of macromolecular crowding and macromolecular confinement on biochemical reactions in physiological media. J. Biol. Chem. 276, 10577– 10580. 30. Cunha, S., Woldringh, C. L. & Odijk, T. (2001). Polymer-mediated compaction and internal dynamics of isolated Escherichia coli nucleoids. J. Struct. Biol. 136, 53 – 66. 31. Tzlil, S., Kindt, J. T., Gelbart, W. M. & Ben-Shaul, A. (2003). Forces and pressures in DNA packaging and release from viral capsids. Biophys. J. 84, 1616– 1627. 32. Kindt, J., Tzlil, S., Ben-Shaul, A. & Gelbart, W. M. (2001). DNA packaging and ejection forces in bacteriophage. Proc. Natl Acad. Sci. USA, 98, 13671– 13674. 33. de Frutos, M., Raspaud, E., Leforestier, A. & Livolant, F. (2001). Aggregation of nucleosomes by divalent cations. Biophys. J. 81, 1127– 1132. 34. Leforestier, A., Fudaley, S. & Livolant, F. (1999). Spermidine-induced aggregation of nucleosome core particles: evidence for multiple liquid crystalline phases. J. Mol. Biol. 290, 481– 494. 35. Samulski, E. T. (1985). Macromolecular structure and liquid crystallinity. Faraday Discuss. Chem. Soc., 7– 20. 36. Parsegian, V. A., Rand, R. P., Fuller, N. L. & Rau, D. C. (1986). Osmotic stress for the direct measurement of intermolecular forces. Methods Enzymol. 127, 400– 416. Edited by J. O. Thomas (Received 28 May 2003; received in revised form 3 September 2003; accepted 6 September 2003)