Nội dung

MINIREVIEW Chromatin under mechanical stress: from single 30 nm fibers to single nucleosomes Jan Bednar1,2,3 and Stefan Dimitrov4 1 2 3 4 CNRS, Laboratoire de Spectrometrie Physique, St Martin d’Heres, France Charles University in Prague, First Faculty of Medicine, Institute of Cellular Biology and Pathology, Prague, Czech Republic Department of Cell Biology, Institute of Physiology, Academy of Science, Prague, Czech Republic Institut Albert Bonniot, Grenoble, France Keywords chromatin, micro-manipulation, nucleosome, optical tweezers Correspondence J. Bednar, CNRS, Laboratoire de Spectrometrie Physique, UMR 5588, BP87, 140 Av. de la Physique, 38402 St Martin d’Heres Cedex, France Fax: +33 476 51 45 44 Tel: +33 476 51 47 61 E-mail: jbedn@lf1.cuni.cz (Received 22 November 2010, revised 7 April 2011, accepted 28 April 2011) doi:10.1111/j.1742-4658.2011.08153.x About a decade ago, the elastic properties of a single chromatin fiber and, subsequently, those of a single nucleosome started to be explored using optical and magnetic tweezers. These techniques have allowed direct measurements of several essential physical parameters of individual nucleosomes and nucleosomal arrays, including the forces responsible for the maintenance of the structure of both the chromatin fiber and the individual nucleosomes, as well as the mechanism of their unwinding under mechanical stress. Experiments on the assembly of individual chromatin fibers have illustrated the complexity of the process and the key role of certain specific components. Nevertheless a substantial disparity exists in the data reported from various experiments. Chromatin, unlike naked DNA, is a system which is extremely sensitive to environmental conditions, and studies carried out under even slightly different conditions are difficult to compare directly. In this review we summarize the available data and their impact on our knowledge of both nucleosomal structure and the dynamics of nucleosome and chromatin fiber assembly and organization. Introduction Since the pioneering use of micromechanical and single molecule manipulation approaches to probe biological systems back in the late 1980s and 1990s (e.g. [1–5]), their use has continuously expanded. In this review we will focus mainly on the approaches using optical and magnetic tweezers for studying the structure and conformational transitions of chromatin. The basic repeating unit of chromatin, the nucleosome, represents the first level of the chromatin organization [6]. The major part of the nucleosome (termed the chromatosome [7]) is composed of an octamer of core histones (two each of H2A, H2B, H3 and H4), a linker histone and  166 bp ( 56 nm) of DNA [6]. The histone octamer alone associates with 146 bp of DNA ( 50 nm) wrapped round in 1.65 left-handed superhelical turns (Fig. 1) to form the nucleosome core particle (NCP), the structure of which has been solved to 1.9 Å resolution by X-ray crystallography [8]. The neighboring chromatosomes are connected by linker DNA. The linear array of nucleosomes folds into 30 nm fiber, the second level of chromatin organization. The linker histones and the core histone NH2 tails and their post-translational modifications are essential for both the folding process and the maintenance of the chromatin fiber [9–11] as well as for the maintenance Abbreviations ACF, ATP-dependent chromatin assembly and remodeling factor; HMG, high-mobility group; NAP-1, nucleosome assembly protein 1; NCP, nucleosome core particle. FEBS Journal 278 (2011) 2231–2243 Journal compilation ª 2011 FEBS. No claim to original French government works 2231 Chromatin under mechanical stress J. Bednar and S. Dimitrov Fig. 1. Chromatin organization and scheme of chromatin array stretching under different force regimes. A nucleosomal core particle, formed by 147 bp of DNA and a histone octamer, is complemented with linker histone (H1) and an additional 20 DNA bp to form the chromatosome. Linker DNA completes and links consecutive nucleosomes which fold into the 30-nm chromatin fiber. During stretching the nucleosomal array is first stretched to its contour length. Additional stretching leads to the rupture of inter-nucleosomal interactions and the array is stretched to the beads-on-a-string configuration. Further force increase results in progressive eviction of histone octamers. The force values are approximate (see text) (adapted from [69,80]). of mitotic chromosomes [12,13]. The globular domain of the linker histone is internally located in the 30 -nm chromatin fiber [14], although how it interacts with both the NCP and the linker DNA remains a subject of debate [15,16]. The conformation of the 30 nm chromatin fiber is sensitive to ionic conditions [9]. The fiber adopts a relaxed zigzag structure at low ionic strength and undergoes compaction with increasing salt concentration, reaching a very compact form under physiological conditions. The linker DNA arrangement in the most compact form of the chromatin fiber continues to be a controversial issue [15–19]. Micromechanical approaches were used to study three different aspects of chromatin organization: the mechanical properties of (a) mitotic chromosomes and (b) an individual nucleosome or a single 30 nm chromatin fiber, and (c) the rheology of chromatin in vivo (e.g. [20–22]). Mitotic chromosomes have been the subject of several ‘mechanical’ studies [23–29] and some of the stretching experiments were performed long before the invention of optical tweezers [30–33]. These studies have recently been thoroughly reviewed [34] and this review will thus concentrate on reviewing single molecule studies of individual nucleosomes, nucleosomal arrays and 30 nm chromatin fibers. Chromatin samples ‘eligible’ for single molecule experiments All micromechanical experiments applied to a nucleosome or chromatin fiber require an adaptation of the substrate in order to make it suitable for attachment to a ‘micro-handle’. In the case of experiments with optical tweezers, micro-beads of dielectric material (silica, polystyrene) are the most frequently used type of ‘handle’. A typical configuration of the optical tweezers stretching experiment is depicted in Fig. 2. The 2232 Fig. 2. Optical tweezers experimental setup. The laser beam (LB) is conducted via a dichroic mirror (DM) to the back aperture of the objective lens (OL) which focuses the beam and creates the optical trap (OT) at the focal point. The filament (F) (DNA, chromatin fiber etc.) is tethered between a trapped bead (TB) and a bead (FB) held by suction onto a micropipette (MP). The micropipette is coupled to a high precision micro-positioning system (typically a piezoelectric XY plate). The image of the bead is projected onto a position sensor (PS). As the fiber is stretched beyond its curvilinear length, the bead in the trap will start to displace from the center of the trap and the force which is trying to bring the bead back is linearly proportional to this displacement. Thus, the change of the fiber length as a function of the force can be measured, resulting in a so-called force–extension curve. chromatin substrate is tethered between the two beads by means of a very tight interaction, typically using biotin ⁄ streptavidin or digoxigenin ⁄ anti-digoxigenin coupling between the fiber ends and beads. Four distinct types of chromatin substrates have been used for stretching experiments: (a) native chromatin, isolated from nuclei after microccocal nuclease digestion, (b) chromatin reconstituted in vitro by salt FEBS Journal 278 (2011) 2231–2243 Journal compilation ª 2011 FEBS. No claim to original French government works J. Bednar and S. Dimitrov Chromatin under mechanical stress dialysis, (c) nucleosome assembly protein 1 (NAP-1) and ATP-dependent chromatin assembly and remodeling factor (ACF) assembled chromatin and (d) chromatin assembled in nuclear extracts. These distinct substrates have different properties and advantages ⁄ disadvantages for the experiments. Native chromatin fibers, isolated after light micrococcal nuclease digestion from nuclei, exhibit heterogeneous lengths (different numbers of nucleosomes per individual fiber) and both their protein composition and state of histone modifications are poorly defined. The nucleosomal arrays reconstituted by salt dialysis on tandem repeats of positioning DNA sequences (601 [35] or 5S [36]) have defined length, and the nature and modification state of histones can be controlled, but the proper association of linker histones in vitro is difficult. Therefore, this material is mostly studied in their absence. The use of NAP-1 ⁄ ACF systems has allowed the reconstitution of chromatin using any DNA sequence, but it does not solve the issue of linker histone assembly. The preparation of chromatin fragments in nuclear extracts does not require a DNA substrate bearing positioning sequences and the number of nucleosomes will depend only on the length of DNA used. The linker histone will be present, although its type will vary depending on the type of extract used. Unfortunately, in addition to the chromatin assembly proteins, the extracts contain a large number of other proteins which can eventually form distinct DNA–protein complexes. These could affect the physical properties of the assembled chromatin fiber and consequently the interpretation of the measured elastic parameters. the fiber and its stretch modulus at low salt conditions were determined to be 30 nm and 5 pN, respectively. The chromatin fiber showed similar elastic behavior when the experiments were performed in 40 mm NaCl. However, the forces necessary to achieve the same extension of the fiber were significantly higher. This was attributed to the more compact initial conformation of the fiber. Although the compaction level of native chromatin fibers (containing the linker histone) is significantly higher in 150 mm NaCl than in 40 mm [9], quite surprisingly the experiments in 150 mm did not show significant differences in the fiber elastic characteristics compared with those at 40 mm. As mentioned earlier, about 166 bp of DNA is associated with the chromatosome and 145 bp with the NCP. The nucleosome structure can thus be considered as a ‘DNA length’ buffer. When the stretching forces applied on the chain of nucleosomes exceeds the mechanical resistance of the DNA ⁄ histone contacts, a mechanical unwrapping of DNA from the histone octamer will occur. If this release is discontinuous (i.e. a certain characteristic length of DNA is released in an all-or-none event) this will lead to a drop of instant stretching force and a sawtooth profile of the force ⁄ extension curve will appear. This is referred to as a ‘disruption’ event. As the nucleosome will not reassemble, the length of the fiber will remain increased and in the next stretch ⁄ relaxation cycle a different extension curve will be observed. In this study [37], the sawtooth pattern could not be directly observed as the stretching was effected in discrete steps of about 50 nm, a value similar to the total length of nucleosomal DNA. Mechanical properties of native 30 nm fiber Mechanical properties of chromatin reconstituted in egg extract The first ever single molecule micromanipulation experiment on chromatin focused mostly on the elastic behavior of the 30-nm chromatin fiber as a function of its environmental conditions [37]. In the low ionic strength (5 mm NaCl) and low force regime (< 10 pN), the measured stretching curve exhibited a rather extended plateau, which was interpreted as fiber accordion-like extension and disruption of inter-nucleosomal interactions. The energy of these interactions was estimated to be around 3.4 kBT per nucleosome. The authors observed the onset of a hysteresis in repeated stretch ⁄ relaxation cycle curves at a force of about 20 pN. Its origin was attributed to eviction of some histone octamers from the fiber by mechanical stress. These experiments allowed the determination of several physical parameters. The persistence length of Another work used chromatin fibers reconstituted in Xenopus laevis egg extract [38]. Stretching these fibers at a continuous speed of 1 lmÆs)1 revealed a sawtooth profile, which started to appear at forces above 20 pN and continued until about 40 pN. The analysis revealed three distinct characteristic DNA release lengths: 65, 130 and 195 nm. The 65 nm was attributed to single nucleosome disruption and the two others were attributed to the simultaneous dissociation of two and three nucleosomes, respectively. The direct attribution of the observed released lengths to a single nucleosomal DNA unwrapping, however, appeared not to be straightforward. Upon eviction of the histone octamer and histone H1 (i.e. the disruption of the chromatosome) 166 bp of DNA is expected to be released, i.e. 56.4 nm and not 65 nm. FEBS Journal 278 (2011) 2231–2243 Journal compilation ª 2011 FEBS. No claim to original French government works 2233 Chromatin under mechanical stress J. Bednar and S. Dimitrov To explain this, it was suggested [38] that other nonhistone proteins, namely high-mobility group (HMG) family members, abundant in the X. laevis egg extract, were associated with the nucleosome. This would result in reinforcing the nucleosome mechanical resistance and in locking of additional DNA into the complex. These suggestions were not experimentally addressed, however. These experiments allowed also calculation of the assembly rate of the nucleosomes, which was found to be about three nucleosomes per second. For the length of k DNA (48 kbp) and the nucleosomal repeat length (200 bp), the assembly of chromatin would thus be complete in about 80 s under these experimental conditions. This is far shorter than the chromatin assembly time (typically a few hours) in bulk in vitro reconstitution in egg extracts [39]. When a force countering the DNA shortening (due to nucleosome formation) was applied, the rate of DNA shortening gradually decreased and was finally halted at forces above 10 pN. A similar fast rate of nucleosome assembly was also observed in experiments where DNA was stretched by hydrodynamic shear forces and incubated in nuclear extracts [40]. This apparent contradiction between the rates of assembly of single molecules and bulk chromatin could be explained by the very high histone : DNA ratio in the single molecule experiment compared with those in the bulk experiments. When competitive DNA (in amounts needed to reach the DNA : histone ratio typical for bulk experiments) is added to such a system, the nucleosome assembly rate dramatically decreases (Claudet, Bednar and Dimitrov, unpublished results). Unwrapping individual nucleosomes in reconstituted nucleosomal arrays A detailed study of the mechanical behavior of in vitro (by salt dialysis) reconstituted nucleosomal arrays (on DNA templates containing 17 tandem repeats of the 5S positioning sequence from sea urchin) was accomplished by Brower-Toland and colleagues [41]. In their experiments, the arrays did not contain linker histones and the stretching was performed in 100 mm NaCl, 1.5 mm MgCl2. The stretching profiles were recorded with either constant stretching speeds or at constant force. The force–extension curves showed characteristic sawtooth patterns at forces starting at about 20 pN with  17 nominal peaks and a regular length of substrate elongation steps of about 27 nm. Very similar values were observed in the constant force regime. Interestingly, the authors also observed a continuous non-DNA stretching profile under a low force regime 2234 (< 15 pN). This part of the curve was interpreted as a continuous unwinding of nucleosomal DNA from the histone octamer, mainly from contacts with histones H2A ⁄ H2B where the DNA–histone interactions are supposed to be weak [42]. The total amount of DNA released was calculated to be 158 bp per nucleosome, a value slightly higher than the 147 bp expected. It was concluded that 76 bp of DNA per nucleosome is unwound continuously in the low force regime, and 82 bp dissociates under stresses higher than 20 pN in an all-or-none fashion. The calculated energy (using dynamic force spectroscopy theory [43]) necessary to dissociate the DNA from the histone octamer was 21– 22 kcalÆmol)1. In the multiple stretching cycle experiments, the reappearance of peaks was observed when the time gap between successive cycles was sufficiently long (at least 10 s) and the stretching force in the preceding cycle did not exceed 50 pN. It was suggested that forces below this value did not cause a complete eviction of all histone octamers. Some of the octamers may have remained attached to the DNA (probably at the dyad region, where the DNA–histone interactions are the strongest), and upon DNA relaxation nucleosome reassembly could occur. This phenomenon was not observed in the case of stretching experiments using chromatin reconstituted in X. laevis egg extracts [38]. The part of the stretching profile interpreted as ‘continuous release of the outer DNA turn’ [41] is very similar to the initial plateau in the experiments with native chromatin under similar ionic conditions [37], interpreted as chromatin fiber accordion-like extension and reflecting the disruption of the nucleosome–nucleosome interactions. Unfortunately, in [41] there is no comparison with stretching profiles in low salt conditions, which would help to clarify the contribution of inter-nucleosomal interactions or elastic contributions of chromatin fiber compaction. Note that the direct comparison of results reported in these two studies [37,41] is rather difficult as the fibers used in [37] were about 15-fold longer and contained linker histones. In addition, the experiments were carried out under different ionic conditions. The experiments were further refined with arrays reconstituted with tail-less histones or histones with modified NH2-termini [44]. The removal of N-termini of all core histones had a strong impact on both the length of the outer DNA turn and the peak force, which dropped by nearly 40 bp (from 65 bp in intact to 28 bp in tail-less octamer nucleosomes) and to 3 pN, respectively. Also the released DNA lengths in the case of the nucleosome with intact tails were revised FEBS Journal 278 (2011) 2231–2243 Journal compilation ª 2011 FEBS. No claim to original French government works J. Bednar and S. Dimitrov and found to be 65 bp (instead of 76 bp in [41]) for continuous release of the outer DNA turn and 72 bp (instead of 82 bp) for disruption of the inner turn. In all studied cases the removal or modification of histone tails influenced the stretching profile and the effect concerned mainly the outer DNA turn while the inner turn was only minimally affected. A similar phenomenon was also observed for nucleosomal arrays reconstituted with the H2A.Bbd histone variant octamer. In this case a 2 pN drop of threshold disruption forces (from 19 pN for conventional nucleosomes to 17 pN for H2A.Bbd nucleosomes) was measured [45]. The last result is in agreement with the data obtained from other methods showing a weaker association of the variant H2A.Bbd octamer with DNA [45,46]. Obviously, similar experiments performed on nucleosomal arrays prepared by salt dialysis and by assembly in egg extracts (see above) gave quite divergent results. While the threshold force values were very similar (about 20 pN), the lengths of DNA released upon mechanical disruption of nucleosomes were quite different. The values of 65 nm or 130 nm measured in the case of egg extract assembled fibers [38] were never observed for chromatin reconstituted by salt dialysis. Gemmen et al. [47] performed analogous experiments on nucleosomal arrays prepared in vitro by using the histone chaperone NAP-1 and ACF which forms nucleosomal arrays on random DNA sequences with nucleosomal repeat of about 168 bp [48,49]. Although the features of the measured stretching profiles were generally comparable with the results of [41] (including the DNA re-wrapping in repeated stretching cycles), some important differences were observed. The disruption length varied from 55 bp to 95 bp and the threshold forces ranged from 5 to 65 pN. In addition, the authors found a clear dependence of the average threshold force on ionic conditions, ranging from 24 pN in 100 mm NaCl to 31 pN in 5 mm NaCl. The wide range of measured threshold forces was attributed to the variation of histone octamer affinities to the given underlying DNA sequence. We have studied the elastic properties of both native chromatin samples (isolated from chicken erythrocytes and containing linker histones) and nucleosomal arrays reconstituted by salt dialysis [50]. We found the same values of basic characteristics of the majority of disruption events, i.e. the peak force and the released DNA length, as reported in [42] (20 pN and 25 nm). However, a minor population of events exhibited a disruption length centered at 50 nm, thus corresponding very closely to the 147 bp of DNA released upon disruption. How can we explain this finding? It was previously reported that the integrity of the nucleosomal Chromatin under mechanical stress structure depends on two factors: (a) the ionic conditions and (b) the concentration of the chromatin itself [51]. At very low concentrations of chromatin, the structure is destabilized and a progressive dissociation of the linker histone and H2A–H2B dimers from the nucleosome is observed [50,51]. When the stretching experiments with native chromatin fibers were repeated under conditions favoring histone octamer stability (presence of exogenous chromatin or low ionic strength) a significant increase in the number of 50-nm events was observed. This was interpreted as an effect of histone octamer stabilization and the release of all the DNA associated with a histone octamer in an allor-none event [50]. A similar effect was observed with arrays containing 12 nucleosomes reconstituted on 5s positioning sequences. Further analysis showed that indeed under conditions typical for single molecule experiments (where the chromatin concentration is usually extremely low) H2A–H2B dimers as well as linker histones readily dissociated from the nucleosomes even at moderate ionic concentrations [50]. The remaining (H3–H4)2 tetramers associate with only one superhelical turn of DNA and consequently, upon stretching, the release of only 25 nm in a single disruption event will be observed. Why then were the peaks with 25-nm release length not observed in experiments with egg extract reconstituted chromatin? One of the possible explanations is the association of non-histone proteins (e.g. HMG proteins) with chromatin, leading to an additional stabilization, mainly of the outer turn. The study of Pope and colleagues [52] showed that the situation might be even more complex. In their work they focused mainly on the elastic response of chromatin fibers assembled in X. laevis egg extract to different loading rates (i.e. the force increase per time unit). They detected three typical disruption lengths: 30 nm, 59 nm and 117 nm. These data differed from the results of Bennink et al. [38], where the 30-nm disruption length was not detected, and revealed two distinct energy barriers having values of 25 and 28 kBT (14.5 kcalÆmol)1 and 16 kcalÆmol)1). With high loading rates the value of the first barrier dropped to 20 kBT (12 kcalÆmol)1). The individual lengths were attributed to a disruption of the entire nucleosome in one event (60 nm), simultaneous release of the DNA from two nucleosomes (117 nm) or the partial unraveling of one DNA turn (30 nm). In addition to the explanation of Brower-Toland and Wang [53], Pope et al. [52] also considered the possibility of disruption of an incomplete nucleosome – missing either one or both H2A–H2B dimers. The individual energy barriers were attributed to nucleosomes with and without linker histone B4 FEBS Journal 278 (2011) 2231–2243 Journal compilation ª 2011 FEBS. No claim to original French government works 2235 Chromatin under mechanical stress J. Bednar and S. Dimitrov (the embryonic linker histone variant present in the egg extract). Based on the analysis, the linker histone contribution to nucleosomal ‘stability’ was estimated to be rather low, about 3 kBT, which would reflect the fact that no significant difference in threshold force was observed between nucleosomal arrays with and without linker histones [50]. In repeated stretching experiments the number of events with high energy barriers (28 kBT) rapidly decreased suggesting the permanent removal of B4 from the nucleosomes during the initial stretching. No correlation between the disruption length and the energy barrier was found. The value of the barrier was significantly lower than that reported in [41] (16 kcalÆmol)1 versus 20–22 kcalÆmol)1) but again the experiments were carried out in different ionic conditions (10 mm Tris ⁄ HCl, pH 7.5, 1 mm EDTA, 150 mm NaCl, 0.05% BSA and 0.01% NaN3) and the chromatin samples were assembled by different techniques. The mechanical properties of nucleosomal arrays reconstituted on African green monkey alpha-satellite DNA were studied by Bussiek et al. [54]. They found disruption of alpha-satellite nucleosomes to occur at a higher force on average – 26.4 pN versus 21.7 pN for random DNA nucleosomes. The authors hypothesized that the increased bending flexibility of alpha-satellite DNA (due to the presence of clustered CA ⁄ TG steps) would result in the formation of more stable nucleosomes as less energy is needed for DNA bending. Zooming in on the stretching of a single nucleosome Analysis of the experiments with nucleosomal arrays is always complicated by the elastic contribution of the inter-nucleosomal interactions at different ionic concentrations. This could be overcome by analyzing the properties of a single mononucleosomal template. Mihardja et al. [55] prepared a mononucleosomal template on a 2582-bp long DNA construct containing a single 601 positioning sequence [35]. Stretching profiles of these particles showed several features not previously observed. Pulling the template at very low loading rates, the first discontinuity in the stretching curve was observed at forces centered at  3 pN and the length of the released DNA was determined to be 21 nm. A second peak occurred at forces around 8–9 pN with a similar length, 22 nm. These events were interpreted as a successive release of the outer and the inner wrap of the nucleosomal DNA. The first unwrapping was reversible, provided the stretching curve did not reach the second discontinuity. Experiments conducted under a constant force regime 2236 ranging between 2 and 3 pN revealed a bistable character of the first event with a dwell time in the unwrapped state depending on the force value, increasing with increased force. From these measurements the free energy of the outer turn unwrapping was calculated to be  6 kcalÆmol)1. The unwrapping of the second, inner turn represented by the second peak at about 8 pN was not reversible. Its analysis with loading rates in the range 2.4–11 pNÆs)1 revealed that the dependence of the probability of unwrapping on the force was not linear. Therefore, the unwrapping of the inner turn cannot be considered as a simple two-state process but will involve some intermediate states as well. The same experiments were also performed at high salt concentrations (200 mm potassium acetate). Under these conditions, the first low force transition was transformed into a nearly continuous plateau rather than a sharp peak and the high force transition was shifted to lower forces. These experiments identified at least two novel features of the nucleosome elasticity behavior. First, the value of the disruption force was lowered to about half of that originally reported (9 pN versus 20 pN) and, second, the experiments clearly showed that unwrapping of the outer turn was not continuous as reported previously [41]. The differences in these experimental data from those obtained with nucleosomal arrays could reflect both the differences in the experimental conditions and the nature of the starting material. The single nucleosome experiments avoid all contributions coming from the fiber-like behavior of the nucleosomal arrays, which is strongly dependent on the ionic conditions. The forces needed to stretch the fibers containing native linker histone without disrupting the nucleosomes (5 pN [37]) are roughly equal to or greater than the threshold forces for unwrapping of the outer turn (3 pN [55]). It is thus likely that at forces up to 5 pN two events are happening simultaneously – an unwrapping of the outer DNA turn and stretching of the folded nucleosomal array. The resulting elastic profile would reflect a superposition of these two events. This would in turn result in a smeared, plateau-like characteristic of the stretching curve at low forces rather than resolved peaks. The different composition of buffers used in the experiments make the comparison even more difficult. As the nucleosome stability strongly depends on ionic conditions, in order to directly compare the data the experiments have to be carried out under exactly the same ionic conditions. Rather high concentrations of Mg2+ (10 mm magnesium acetate) together with  50 mm potassium acetate, however, were used in the single nucleosome experiments [55], while 100 mm FEBS Journal 278 (2011) 2231–2243 Journal compilation ª 2011 FEBS. No claim to original French government works J. Bednar and S. Dimitrov NaCl and 1.5 mm MgCl2 were used for nucleosomal array stretching [41]. An interesting approach to investigate the stability of a single nucleosome was used by Shundrovsky et al. [56]. Instead of pulling tethered nucleosomal templates, they ‘unzipped’ the DNA of a reconstituted template containing a single 601-positioned nucleosome. The nucleosome was flanked by free DNA arms and, upon stretching, the first 220 bp of naked DNA were unzipped before the histone octamer was reached. The unzipping of DNA associated with the histone octamer was affected by histone–DNA contacts within the nucleosome and reflected the strength of the histone– DNA interactions. The unzipping profile of the nucleosome showed three distinct high force regions (contrary to the first two, the third region was not regularly observed). Within these regions, forces up to 45 pN had to be applied in order to overcome the barrier. The first peak was observed at about 50 bp from the dyad upon applying an average force of 31 pN, while the second one was observed in the vicinity of the nucleosome dyad and at 37 pN average force. These peaks were attributed to the disruptions of the strong interactions between H2A–H2B dimers and H3–H4 tetramers, respectively. The attribution of the first peak to the disruption of the H2A ⁄ H2B–DNA interaction was confirmed by stretching a particle reconstituted with the (H3–H4)2 tetramer only. The unzipping profile of this tetrameric particle exhibited only the second, high force peak. According to the authors, the third peak was associated with the instability of the nucleosome when most of the nucleosomal DNA was unzipped. These experiments were further refined [57], allowing analysis of the DNA–histone interactions with near base-pair resolution. The unzipping was carried out under a constant force regime using a 28-pN trapping force. The strength of the interaction was found to be proportional to the time needed for its disruption. This allowed mapping of the interaction strength of the different regions with a resolution of about 1.5 bp. The recorded data again revealed three regions of strong interactions (longer dwell times): one was located close to the dyad, while the other two were symmetrically located at positions ±40 bp from the dyad. All three exhibited a 5-bp periodicity. The data demonstrated that the unzipping of the first 20 bp of nucleosomal DNA had the same characteristics as those of naked DNA, indicating a loose interaction of the histones with DNA at the entry ⁄ exit points of the NCP. Very similar results were obtained in continuous stretching regime measurements with loading rates of 8 pNÆs)1, as well as when random DNA sequences instead of Chromatin under mechanical stress positioning sequences were used for nucleosome reconstitution. Magnetic tweezer experiments Several experiments with magnetic tweezers have also been reported (for the principles of magnetic tweezers see for example [58,59]). Magnetic tweezers can measure forces about 1–2 orders smaller than optical tweezers and, unlike optical tweezers, they can also control the torsion of the fiber. Leuba et al. [60] studied NAP-1 mediated assembly of chromatin fibers on k DNA using magnetic tweezers. They observed an inhibition of the fiber assembly at forces of  10 pN, but they also registered disassembly events (in an otherwise progressive assembly process) at forces of about 5 to 7.5 pN. This suggested that the equilibrium forces were in this range. Experiments using a similar strategy, but in X. laevis egg extracts, were realized by Yan et al. [61]. The experiments were carried out either in ATP-depleted extract or in extract containing a defined concentration of ATP or non-hydrolyzable ATP. They found that in ATP-depleted extract forces of only 4 pN resulted in inhibition of nucleosome assembly. At forces below 3.5 pN, the extract was able to accomplish the assembly although the number of assembled nucleosomes was significantly lower relative to the nucleosomal array reconstituted under optimal conditions (the measured nucleosomal repeat was only 280 bp in contrast to the 180–160 bp repeat reported for fully extractassembled chromatin [39]). The 3.5 pN value was determined as an equilibrium force of ATP-independent nucleosome assembly giving straightforwardly the free energy of DNA–histone octamer association as 27 kcalÆmol)1. Once the assembly was completed, the fibers were stretched with different loading rates. During this process, a step-wise fiber lengthening was observed with a predominating step value of 50 nm, attributed to an unwrapping of one complete nucleosome. The presence of 30- and 100-nm steps was also detected. Interesting changes were induced by addition of ATP to the extract. In this case, the disassembly threshold force decreased to  1 pN. Non-hydrolyzable ATP did not affect the nucleosome assembly ⁄ disassembly equilibrium force determined in ATPdepleted extract. Why did these two very similar experiments give rise to such different results? First, the egg extract contains a poorly defined composition of proteins compared with the purified NAP-1 assembly system. It is quite possible that some ATP-independent protein complexes present in the egg extract can associate with the FEBS Journal 278 (2011) 2231–2243 Journal compilation ª 2011 FEBS. No claim to original French government works 2237 Chromatin under mechanical stress J. Bednar and S. Dimitrov nucleosomes and modify their mechanical stability. The steep drop to  1 pN in the stall force in the presence of ATP, however, is quite surprising. The events observed in the stretching profile under these conditions did not correspond to an assembly of individual nucleosomes, but rather to formation and release of rather long ‘loops’ (200–400 nm). The fact that the energy provided by the added ATP in the system was not even partly used for assisted nucleosome assembly is also surprising. However, the authors have observed nucleosome-like disassembly steps of 50 and 100 nm when the force was increased to over 5 pN. Importantly, no reverse (i.e. assembly) events were detected even at low forces. Kruithof et al. [62] carried out experiments on strongly subsaturated oligonucleosomal arrays (one to four nucleosomes present on 17 tandem repeats of 5s DNA) using magnetic tweezers with sub picoNewton resolution. This experiment is directly comparable with the work of Mihardja et al. [55]. Although both groups used very similar conditions, Kruithof et al. did not observe any DNA unwrapping from the nucleosome below forces of 6 pN, even though they used a positioning sequence with lower affinity for the histone octamer (5s versus 601). The data obtained by force spectroscopy of chromatin are not always easy to interpret unambiguously and to explain in terms of changes in nucleosome and fiber structure and dynamics. While in lower salt concentrations (50 mm) and in the absence of bivalent ions the inter-nucleosomal interactions can be neglected, the situation becomes more complex when the fiber is studied in its compact form, where the presence ⁄ absence of linker histones, the higher concentration of monovalent ions, and the presence of bivalent or polyvalent ions contribute significantly to the fiber properties. Kruithof et al. [63] used improved techniques of linker histone association [64,65] to prepare defined chromatin arrays of 25 nucleosomes with two different nucleosomal repeat lengths (197 and 167 bp) and used magnetic tweezers to study their elastic behavior. The stretching curves of these samples exhibited four major regions. The first was attributed to the extension of the DNA segments flanking the array of 25 nucleosomes which serve as a handle for tethering. The second region (at forces up to 4 pN) represented the extension of the chromatin fiber. The third region (a plateau observed at 4–4.5 pN) was attributed to the disruption of inter-nucleosomal interactions. The last region was interpreted to reflect extension of the beads-on-a-string fiber. The incorporation of the linker histone had only a minor effect on the overall form of the stretching profile. Upon H5 association, the third 2238 region (plateau) was shifted to a higher force value – 7 pN – suggesting that linker histone stabilizes nucleosomal stacking. However, its absence did not compromise chromatin folding when Mg2+ was present (1.5 mm MgCl2). When Mg2+ ions were depleted from the solution, the behavior of the fibers without linker histones changed. A disruption of the inter-nucleosomal interactions at forces of about 3.5 pN and an increasing irreversibility upon repeated stretching cycles (in the presence of 100 mm NaCl) were observed. Reintroduction of Mg2+ resulted in a complete recovery of the original folding pattern, suggesting that, at least under these conditions, the linker histone might not be required for proper chromatin folding. The analysis of fiber stretching profiles, their Hookian behavior, their length and transition to extended beads-on-a-string structures in the third and fourth regions of the stretching curve led the authors to conclude that in its compact form the fiber is organized in a one-start solenoidal topology. The data obtained on fibers with 167 bp nucleosomal repeat were significantly different [63]. Surprisingly, their contour length at 0.5 pN stretching force was longer than for fibers with 197 bp nucleosomal repeat and their measured stiffness was found to be 2.7-fold higher (0.052 versus 0.019 pNÆnm)1). This was interpreted as a consequence of their different topological organization and a two-start helix topology was suggested as best fitting the observed data. However, the story of chromatin fiber folding is apparently more complex. Other studies have demonstrated that for longer nucleosomal repeats both linker histone and Mg2 + ions are required in order to reach maximal packing levels of the chromatin [66]. This is not valid for short nucleosomal repeats where, even in the absence of linker histone, the fiber can maximally pack in a regular manner [66]. It would therefore be interesting to see whether the same elastic behavior (i.e. linker-histone-independent compaction in the presence of bivalent ions) would be observed for longer nucleosomal repeats. The presence of Mg2+ also resulted in a substantial increase of the inter-nucleosomal stacking energy to about 17 kBT, compared with the value of 3.4 kBT observed in [37] for native chromatin fibers in the absence of bivalent ions. This clearly demonstrates an important role for bivalent ions in chromatin fiber stabilization. It should also be mentioned that the presence of bivalent ions not only influences chromatin stability [10,67,68] but may also direct the topology of its folding. Indeed, analysis of the data obtained on native chromatin fibers with rather long nucleosomal repeat and in the absence of bivalent ions [37], using metropolis–Monte Carlo FEBS Journal 278 (2011) 2231–2243 Journal compilation ª 2011 FEBS. No claim to original French government works Native, chicken erythrocytes Chromatin reconstituted in Xenopus egg extract on k DNA Chromatin reconstituted on 17 tandem repeats of 5S DNA, no linker histone Chromatin reconstituted on random DNA using NAP-1 ⁄ ACF system, no linker histone Chromatin reconstituted on 12 tandem repeats of 5S DNA, no linker histone, Native chromatin from chicken erythrocytes Chromatin reconstituted in Xenopus egg extract Single nucleosome reconstituted on 601 sequence Strongly subsaturated nucleosomal arrays reconstituted on 17 tandem repeats of 5S DNA, no linker histone Chromatin reconstituted in Xenopus egg extract either ATP depleted or ATP enriched Chromatin reconstituted on tandem repeat alpha-satellite DNA and random DNA using NAP-1 ⁄ ACF system, no linker histone Cui & Bustamante [37] Bennink et al. [38] Brower-Toland et al. [41] Gemmen et al. [47] Mihardja et al. [55] Kruithof et al. [62] FEBS Journal 278 (2011) 2231–2243 Journal compilation ª 2011 FEBS. No claim to original French government works Buissek et al. [54] Yan et al. [61] Pope et al. [52] Claudet et al. [50] Type of chromatin substrate References 30, 59 and 117 nm  20 pN 22 pN random DNA 26 pN alpha-satellite DNA 3.5 pN ATP) < 2 pN ATP+ > 6 pN (no DNA unwrapping observed below) 3 pN outer turn 8–9 pN inner turn 23 nm 50 nm (ATP)) 21 nm outer turn 22 nm inner turn 25 and 50 nm  20 pN 24 pN in 100 mM M+ 31 pn in 5 mM M+ 76 bp outer turn, continuously unwrapped, 82 bp for inner turn 55–95 bp 65, 130, 195 nm Disruption length < 15 pN for outer turn  20 pN for inner turn > 20 pN > 20 pN Threshold force (pN) Table 1. Comparison of experimental conditions and results of selected experiments. 27 kcalÆmol)1 (ATP)) 6 kcalÆmol)1 for outer turn 14.5 and 16 kcalÆmol)1 22 kcalÆmol)1 Calculated energy of DNA–histone octamer dissociation 10 mM Tris ⁄ HCl pH 7.5, 0.05% BSA, 100 mM NaCl 10 mM Tris ⁄ HCl pH 7.5, 1 mM EDTA, 150 mM NaCl, 0.05% BSA and 0.01% NaN3 10 mM Tris-acetate, 50 mM potassium acetate, 10 mM magnesium acetate, 1 mM dithiothreitol, 0.1 mgÆml)1 BSA 10 mM Hepes, pH 7.6, 100 mM KAc, 2 mM MgAc, 10 mM NaN3, 0.1% (v ⁄ v) Tween-20, 0.2% (w ⁄ v) BSA Egg extract ATP depleted or enriched 10 mM Tris ⁄ HCl pH 7.5, 1 mM EDTA, 50–100 mM NaCl, exogenous chromatin 10 mM Tris, 2 mM EDTA pH 7.5, 5, 40 and 150 mM NaCl, 2 mgÆmL)1 BSA, exogenous chromatin 10 mM Tris ⁄ HCl pH 7.5, 1 mM EDTA, 150 mM NaCl and 0.01% (w ⁄ v) NaN3 10 mM Tris ⁄ HCl pH 8.0, 1 mM Na2EDTA, 100 mM NaCl, 1.5 mM MgCl2, 0.02% (v ⁄ v) Tween-20, 0.01% (w ⁄ v) milk protein 20 mM Tris pH 7.8, 1 mM EDTA and 5–100 mM NaCl Ionic conditions J. Bednar and S. Dimitrov Chromatin under mechanical stress 2239 Chromatin under mechanical stress J. Bednar and S. Dimitrov simulation [69], proposed the zigzag organization of the fiber as the best fitting to measured elastic profiles. Therefore, the organization of the chromatin fiber in its compact state remains an open issue and it is very likely that variable topologies can be adopted depending on the given conditions [18]. Chromatin arrays under twist The group of Viovy used magnetic tweezers to study the behavior of a 36 nucleosome long array reconstituted on the tandem repeat of 5s DNA under torsional stress [70]. The acquired data allowed the determination of several elastic parameters of the fiber, namely the persistence length (28 nm) and the stretch modulus (8 pN), which are quite close to the values obtained for native chromatin fibers (30 nm persistence length and 8 pN stretch modulus) determined in [37]. However, the determined torsional persistence length (5 nm) differed markedly from the value of 35 nm obtained by WLC (worm-like chain) modeling of similar arrays, using canonical nucleosomes [71]. A new model of the fiber was therefore proposed, where the nucleosomes could exist in three different configurations according to the crossing of the entry ⁄ exit DNA segments: negatively crossed, open and positively crossed. Transitions between the different configurations are possible and energies of 0.4 kcalÆmol)1 and 1.2 kcalÆmol)1 from negative to open and positive to open nucleosome states, respectively, fitted the experimental data very well. As linker histone was not present in the system, transitions between individual configurations (crossings) of nucleosomes could be facilitated. Further experiments have revealed that the behavior of the fiber differed significantly during stress relaxation [70]. While in the case of negative twist the process was essentially reversible, in the case of positive twist a very significant hysteresis was observed, as if the stress (and the resulting shortening) was released in time by an internal structural rearrangement of the fiber. When the H2A–H2B histone dimers were selectively removed from the nucleosomes, the hysteresis disappeared. Based on previous detailed studies on nucleosomal polymorphism [72–76], the authors proposed a specific mechanism for this rearrangement, which required a flip of the nucleosomal chirality from left-handed to right-handed. Concluding remarks In this review, we have summarized available data on the mechanical properties of nuclesomes and chroma2240 tin. We did not include single molecule experiments in which functional aspects of nucleosomal interactions with other complexes were examined (e.g. [77,78]) or experiments where single molecule techniques other than micromechanical manipulation were used (e.g. [79]). Still, the situation appears to be rather complex, as documented in Table 1 where data obtained in selected studies are compared. As can be seen, in some cases the data from very similar experiments are quite divergent. This reflects the high sensitivity of the studied chromatin samples to a number of parameters. Obviously, the traction parameters, i.e. the loading rate, turns out to be particularly important. It is therefore not surprising that data from early experiments, using in general quite high loads, are quite similar (e.g. a disruption force around 20 pN), but very different from the latest data (3–11 pN). The ionic conditions and the buffer composition are also very important factors, as they can influence the octamer stability or the DNA–octamer association strength. It is also clear that the choice of DNA substrate has an impact on the results [57]. The question of the effect of the linker histone association still remains an open issue as most of the array stretching experiments were carried out in the absence of linker histone. Although substantial progress has been made in the micromanipulation of chromatin substrates, many additional experiments will certainly be needed in order to evaluate the effects of individual factors that potentially influence the mechanical properties of chromatin substrates. Acknowledgement This work was supported by grants from INSERM and CNRS. S.D. acknowledges ANR-09-BLANNT09-485720 ‘CHROREMBER’. J.B. acknowledges the support of the Ministry of Education, Youth and Sports (MSM0021620806 and LC535) and the Academy of Sciences of the Czech Republic (Grant #AV0Z50110509). References 1 Ashkin A & Dziedzic JM (1987) Optical trapping and manipulation of viruses and bacteria. Science 235, 1517–1520. 2 Ashkin A, Dziedzic JM & Yamane T (1987) Optical trapping and manipulation of single cells using infrared laser beams. Nature 330, 769–771. 3 Svoboda K, Schmidt CF, Schnapp BJ & Block SM (1993) Direct observation of kinesin stepping by optical trapping interferometry. Nature 365, 721–727. FEBS Journal 278 (2011) 2231–2243 Journal compilation ª 2011 FEBS. No claim to original French government works