Quantifying ATP-Independent Nucleosome Chaperone Activity with Single-Molecule Methods

Abstract

The dynamics of histone-DNA interactions govern chromosome organization and regulates the processes of transcription, replication, and repair. Accurate measurements of the energies and the kinetics of DNA binding to component histones of the nucleosome under a variety of conditions are essential to understand these processes at the molecular level. To accomplish this, we employ three specific single-molecule techniques: force disruption (FD) with optical tweezers, confocal imaging (CI) in a combined fluorescence plus optical trap, and survival probability (SP) measurements of disrupted and reformed nucleosomes. Short arrays of positioned nucleosomes serve as a template for study, facilitating rapid quantification of kinetic parameters. These arrays are then exposed to FACT (FAcilitates Chromatin Transcription), a non-ATP-driven heterodimeric nuclear chaperone known to both disrupt and tether histones during transcription. FACT binding drives off the outer wrap of DNA and destabilizes the histone-DNA interactions of the inner wrap as well. This reorganization is driven by two key domains with distinct function. FD experiments show the SPT16 MD domain stabilizes DNA-histone contacts, while the HMGB box of SSRP1 binds DNA, destabilizing the nucleosome. Surprisingly, CI experiments do not show tethering of disrupted histones, but increased rates of histone release from the DNA. SI experiments resolve this, showing that the two active domains of FACT combine to chaperone nucleosome reassembly after the timely release of force. These combinations of single-molecule approaches show FACT is a true nucleosome catalyst, lowering the barrier to both disruption and reformation.

1 Introduction

The nucleosome coordinates the histone octamer and ~2 wraps of DNA to form the basic subunit of chromatin (Fig. 1a). Histones are the most abundant nuclear protein, and four specific types, H2A, H2B, H3, and H4, combine in a symmetric disc that contacts the wrapped DNA in hundreds of histone-DNA and water-mediated histone-DNA contacts [1]. A total of 147 base pairs wrap into the nucleosome, centered on the (H3-H4)₂ tetramer, and two H2A/H2B dimers (Fig. 1b). A central 80 base pairs constitute the inner wrap, beginning at the dyad center of the tetramer and extending to each dimer, in direct contact with these core or strong sites [2, 3]. The remaining bases bind ~ 1/2 wraps above and below the inner wrap (Fig. 1c). Though this structure is stable, it is also highly dynamic, as local regions of DNA-histone separation, or breathing, occur. Histone-DNA binding must be disrupted to facilitate many cellular processes including replication, transcription, and repair [4, 5].

Fig. 1

Structures examined in this study. (a) An individual nucleosome consists of four pairs of histones: H4 (blue), H3 (cyan), H2B (pink), and H2A (red). These pairs are arranged symmetrically about a dyad axis, joined at the H3-H3 interface. A total of 147 base pairs of DNA (black/silver) wraps ~1.7 times in this structure (pdb: 1kx5) [47]. (b) DNA is centered about the dyad axis, upon the (H3-H4)2 tetramer and extending along two H2A/H2B dimers. Together, these core histone sites coordinate ~80 base pairs of DNA. (c) The outer 1/2 wraps of DNA (~base pairs each) are more loosely held above and below the inner wrap, a variable linker (of 60 base pairs here) connect adjacent nucleosomes. (d) Human FACT is a heterodimer, consisting of the SPT16 and SSRP1 subunits, joined at dimer domains. (e) Binding to the nucleosome occurs with several DNA-protein and histone-protein contacts throughout this X-ray structure (pdb: 6upl) [12]. (f) An isolated HMGB domain (not present in (e)), binds DNA alone, and as a part of SSRP1, with a specific affinity for bent DNA (which has been bent by cross-linking here, pdb: 1ckt) [48]. (g) The single HMG box of yeast NHP6A contains a longer charged tail, enhancing DNA binding (pdb: 1j5n) [49] (Figure adapted from [50])

Among the variety of proteins that disrupt isolated nucleosomes, the non-ATP-driven heterodimer FACT consists of the 140 kDa SPT16 (SuPpressor of Ty elements 16) and 80 kDa SSRP1 (Structure Specific Recognition Protein 1), joined at dimer domains (DD). Several domains within these subunits contact to both disrupt and reassemble the nucleosome [4, 6,7,8,9,10,11]. Recent cryoEM structures show tetramers bound to FACT (missing the HMGB domain), displacing the H2A/H2B dimer [12], while another series of structures show FACT (plus an accessory protein) tethering histones during the passage of human RNA polymerase II (RNAPII) [13]. Domains of human HMGB exist in the nucleus both as part of FACT and alone, while the HMG box found in yeast, NHP6A, is not directly bound to the yeast analogue of FACT, but coordinates with it. Both HMGB and NHP6A are known to bind to and disrupt DNA wrapped around the nucleosome [21]. Single-molecule experiments on chromatin have yielded insights into chromatin structure ranging from measurements of the mechanical elasticity of a chromosome down to the protein-DNA interactions holding DNA to single nucleosomes [17, 22,23,24]. Here, we will assemble a short array of individual nucleosomes that facilitate the rapid collection of data on both energies and kinetics of histone-DNA interactions. Force disruption (FD) experiments characterize the strength of histone-DNA interactions and facilitate the construction of a free energy landscape. Confocal imaging (CI) experiments probe the kinetics of disrupted DNA-histone interactions. Finally, survival probability (SP) experiments reveal the extent of nucleosome reformation as tension is released.

The following experiments reveal that FACT binds to the nucleosome, releasing the outer 1/2 wraps of DNA and destabilizing the DNA-histone interactions of the remaining inner wrap. Within FACT, two domains, the SPT16 MD and SSRP1 HMGB, function effectively to bind the histone and the DNA, substituting direct DNA-histone contacts with contacts mediated by FACT. SSRP1 HGMB plays a significant role in nucleosome destabilization, while yeast NHP6A shows even greater activity. Yet only the full protein is able to simultaneously destabilize the nucleosome and tether the components for reassembly after release of tension. Full FACT acts as a true nuclear chaperone.

2 Materials

Several studies have detailed the assembly of human histone octamers, as well as their reconstitution onto DNA [25,26,27,28]. Furthermore, protocols for purification of the proteins studied here have been described. The goal of this writing is to present a general technique to study the effectiveness of nucleosome chaperones. Some familiarity with octameter assembly, fluorescent dye labeling, and protein purification is assumed. The following protocol will outline the key steps to create the DNA template and then combine octamers to create a 12× nucleosome array for study. A brief outline of the expression and purification of the proteins specifically is used here, though these techniques are useful for any nucleosome binding ligand.

2.1 Instruments, Reagents, and Buffers for Sample Preparation

• 1. E. coli Rosetta(DE3)pLysS and E. coli BL21(DE3)pLysS cells for expression (MilliporeSigma).

• 2. Plasmid pET28 (modified), pTEV, and pCOLADuet cloning vectors.

• 3. Ni-NTA column (Qiagen).

• 4. Bind buffer: 50 mM NaH2PO4-Na2HPO4, pH 7.5, 300 mM NaCl.

• 5. Wash buffer: 50 mM NaH2PO4-Na2HPO4, pH 7.5, 300 mM NaCl, 20 mM imidazole.

• 6. Elution buffer: 50 mM NaH2PO4-Na2HPO4, pH 7.5, 300 mM NaCl, 200 mM imidazole.

• 7. Superdex 200 column (GE Healthcare) for size-exclusion chromatography with Bind buffer as running buffer.

• 8. TEV protease (stock in glycerol) with 50 mM sodium phosphate pH 6.5, 50 mM NaCl, for SSRP1 HMGB storage.

• 9. Protein dialysis and storage buffer: 20 mM HEPES pH 7.5, 100 mM KCl, 1 mM EDTA, 1 mM DTT, and 5% glycerol.

• 10. pU19 plasmid and 601 Widom positioning sequences.

• 11. Bsal I restriction enzyme.

• 12. DNA pol I.

• 13. Digoxygenin-11-dUTP (Roche), biotin-16-dUTP (Sigma), biotin-14-dCTP (Invitrogen) and biotin-14-dATP (Invitrogen).

• 14. Human octamers and human octamers carrying Alexa488 or Atto647N on each H2B (T112C).

• 15. Nucleosome dialysis buffer, high salt: 10 mM Tris–HCl, pH 7.5, 1 mM EDTA, with Na⁺ of 1 M.

• 16. Nucleosome dialysis buffer, intermediate salt: 10 mM Tris–HCl, pH 7.5, 1 mM EDTA, and Na⁺ of 750 mM.

• 17. Low salt nucleosome dialysis buffer and storage solution: 10 mM Tris–HCl, pH 7.5, 1 mM EDTA, with Na⁺ of 2.5 mM.

• 18. Individual solution of 5 M NaCl.

2.2 Instruments and Buffers for Single-Molecule Experiments

• 1. A fiber coupled laser, up to 200 mM at 850 nm (see Note 1), split 50:50 and directed to form a dual beam counterpropagating single trap. A confocally arranged pair of water-dip** 1.0 NA objectives is adjusted to overlap the two traps. Sub-picoNetwon forces on the trapped bead are determined using a lateral effect detector (Melles Griot) to measure trap** beam deflection, which is proportional to the applied force.

• 2. A home-built flow cell and sample chamber. A custom plexiglass spacer is machined and covered on the sides with 22 × 30 mm glass cover slips, fastened with UV-curing epoxy. Inlet and outlet tubing is also glued in place at opposite ends of the chamber.

• 3. Pre-pulled glass micropipettes with an opening of 0.5 micrometers (World Precision Instruments), glued into place through a hole drilled at 45 degrees from vertical (see Note 2).

• 4. The LUMICKS dual beam imaging C-trap is a commercial system capable of simultaneous force disruption (FD) experiments (complementing the instrument described above) and confocal imaging (CI).

• 5. Polystyrene beads, 1.0% w/v, functionalized with a coating of streptavidin, and with a diameter of 1.7, 3.1, or 5 micrometers (see Note 3).

• 6. 2 μm diameter polystyrene beads, 0.5% w/v, with digoxygenin coating.

• 7. Experimental buffer: 10 mM HEPES, pH 7.5 and Na⁺ of 50 or 100 mM.

3 Methods

Many excellent reviews summarize various optical tweezer designs and experiments [18,19,20,21, 29, 30]. Here, tweezers will apply tension to reconstituted DNA arrays described above in force disruption (FD), confocal imaging (CI), and survival probability (SP) assays. These experiments determine the stability of DNA-histone contacts and the length of DNA wrapped around the octamer. Protein binding affects both the wrapped length and stability, allowing the binding affinity to be determined. The kinetics of histone-DNA interactions and the chaperone activity of these proteins after the release of tension is characterized.

3.1 FACT Protein and Isolated Domain Production

• 1. Human FACT SPT16 and SSRP1 as well as isolated SPT16 MD and SSRP1 HMGB are to be expressed in E. coli and purified to homogeneity (see Note 4). Full SPT16 and SPT16 MD are cloned in a modified pET28 vector, while SSRP1 HMG and SPT16 CTD are cloned in a pTEV vector. An N-terminal His₆ tag is included to facilitate purification. SSRP1 is cloned in a pCOLADuet plasmid, without a tag.

• 2. E. coli Rosetta(DE3)pLysS and E. coli BL21(DE3)pLysS cells are used to express all subunits/domains, grown in LB broth at 37°C to an OD_{600 nm} of approximately 0.6, and then induced with 0.5 mM isopropyl-β-D-thiogalactoside at 15°C for 16–20 h.

• 3. Harvest the cells by centrifugation. Discard the supernatant. Resuspend the pellets in 10 mL of Bind buffer. Lyse the cells using an Avestin EmulsiFlex C5 homogenizer. PMSF is added before cell lysis (see Note 5).

• 4. Centrifuge at 48,000 × g for 45 min to separate supernatant from pellet. Apply the supernatant onto a Ni-NTA column pre-equilibrated with Bind buffer at 4 °C. Extensively wash the column with Wash buffer. Elute the protein with Elution buffer.

• 5. Concentrate the protein solution. Cut the His₆-MBP tag of SPT16 (501-1006) using TEV protease at 4 °C overnight.

• 6. Separate the FACT complex from His₆-MBP by size-exclusion chromatography with a Superdex 200 column using Bind buffer as running buffer. Pool all the fractions containing FACT, concentrate to proper volume and store at −80 °C until use.

3.2 NHP6A Protein Expression and Purification

• 1. Recombinant untagged yeast NHP6A protein is expressed in bacteria and purified by HPLC [31].

• 2. Further purification is achieved via size exclusion chromatography in phosphate buffered saline on a Superdex 200 10/30 column. Elution occurs at a flow rate of 0.4 mL/min.

• 3. Fractions are finally pooled and dialyzed against protein dialysis buffer.

• 4. Proteins are stored at −20 °C in dialysis buffer and supplemented with 50% (v/v) glycerol.

3.3 Combining Nucleosome Position Sequences and Handles to Form Templates

• 1. Two DNA constructs (pJ1937 and pJ2774) are created utilizing 12 tandem copies of the Widom 601 nucleosome positioning sequences, inserted into a pUC19-based plasmid.

• 2. Each Widom positioning sequence is comprised of 147 base pairs that contact the histone octamer and 60 base pairs to serve as a linker (see Note 6).

• 3. The first construct, a linear “pJ1937” template, is prepared by BsaI restriction endonuclease cleavage to create flanking non-nucleosomal “handles” of 1340 and 1360 base pairs. BsaI, a type IIS restriction endonuclease, creates distinct four-base overhanging termini allowing DNA polymerase I repair of the overhangs in the presence of single digoxygenin (digoxigenin-11-dUTP) and biotin (biotin-14-dATP) nucleotide analogs on opposing termini, enabling capture for force-extension experiments (Fig. 2a, see Note 7).

• 4. The second construct, the linear “pJ2774” template, employs longer (3400 bp) “handles” flanking the 12 Widom 601 sequences (Fig. 2b). The template is digested with BsaI restriction endonuclease, and overhangs are filled in with the Klenow fragment of DNA polymerase I in the presence of biotinylated nucleotide analogs labeling both DNA termini (biotin-16-dUTP, biotin-14-dATP, biotin-14-dCTP; see Note 8).

• 5. The biotin- and digoxigenin-labeled DNA was purified by spermine DNA precipitation and resuspended in water prior to assembly with the human histone octamers.

Fig. 2

Constructs used to interrogate chaperone activity. (a) 12× 601 positioning sequences (including 60 base pair linkers) are flanked by DNA handles, each capped by distinct labels for attachment to trapped beads. Human octamers are reconstituted onto this array as described. (b) For imaging, each H2B (T112C) incorporates a fluorescent tag (Alexa488 or Atto647N). Labeled nucleosomes are reconstituted onto the same 12× sites but are flanked by longer handles, coupled with smaller beads and a simpler attachment

3.4 Reconstituting Nucleosome Arrays

• 1. Pre-assembled octamers are combined with the DNA constructs at high concentrations (ng/μL, see Note 9), in low salt dialysis buffer, titrated with isolated 5 M Na⁺ solution to a final concentration of 2 M Na⁺.

• 2. Combined samples are placed into a small volume dialysis button (Hampton Research, 50 μL, see Note 10), and the button is sealed with dialysis membrane (see Note 11) and placed in 200 mL of high salt dialysis buffer.

• 3. Dialysis decreases Na⁺ from 2 M to 1 M over 4–6 h.

• 4. Discard the high salt buffer and replace with intermediate buffer, dialyzing for 4–6 h.

• 5. Finally, replace buffer with low salt buffer and dialyze for another 4–6 h.

• 6. Reconstituted arrays remain stored in low salt dialysis buffer for storage and are stable for several weeks at 4 °C (see Note 12).

3.5 Force Disruption (FD) Using Optical Tweezers

• 1. Stock solutions of streptavidin coated and anti-digoxigenated coated beads are diluted 1000:1 and 1000:2, respectively, into the experimental buffer at ambient lab temperature of 22 °C (see Note 13).

• 2. One 3-micron diameter streptavidin coated bead is caught and held in the optical trap, while a 2-micron diameter anti-digoxigenin coated bead is pulled onto the micropipette tip (see Notes 14 and 15).

• 3. Flow reconstituted 12-mer nucleosome arrays diluted 10,000:1 in experimental buffer into the sample cell (see Note 16). Arrays quickly saturate the flow cell chamber. An array is secured between the beads by rubbing the smaller bead on the micropipette tip against the bigger bead on the trap and slightly tugging out and looking for some resistance (see Note 17).

• 4. Once the nucleosome array is tethered, the piezo stage increases the extension stepwise in 4 nm increments, with a force measurement after each step (see Notes 18 and 19). After a short pause at ~40 pN, the array should be gradually released, also in 4 nm steps (see Note 20).

• 5. Data for the nucleosome array stretching shows a smooth increase in extension with force at the beginning (see Notes 21 and 22).

• 6. Nucleosome outer wrap release will be seen as a shoulder in the low force region (~2–5 pN, see Fig. 3a).

• 7. As force increases further (>10 pN), sudden, consecutive increases in extension will be seen. These correspond to nucleosome inner wrap (core) openings. Discrete rips appear, in the which the extension change is due to base pairs unwrapped and released from each nucleosome. All wild-type nucleosomes will be unwrapped once the applied tension reaches ~40 pN.

• 8. Opening events at higher forces are due to 2× or even 3× arrays tethered simultaneously between the same streptavidin-anti-digoxigenin bead complex. It is difficult to resolve the data due to multiple arrays and to physically separate multiple arrays from the same bead. These curves and beads should be discarded, and the fresh beads caught.

Fig. 3

Force disruption (FD) experiments and analysis. (a) A single cycle of extension (solid circles) and release (open circles) for an array reveals DNA handle elasticity punctuated by DNA unwinding from the octamer in two distinct steps. The outer 1/2 wraps are released gradually and indistinctly below 5 pN. Red lines trace polymer elasticity for the wrapped and the unwrapped states, and the pink shaded region the equilibrium energy of release for all 12× nucleosomes. Above 10 pN sudden changes in extension characterize the non-equilibrium release of the inner wrap from the strong sites. Here, the blue lines show the change in length for a single event, and the cyan region characterizes the non-equilibrium work done. Some rewinding is evident upon release, as the histones may remain attached to the DNA at the dyad. (b) The measured force that precipitates the release of the inner wrap, arranged in order observed (A is the number remaining in the array). Events from n = 25 arrays are shown in open blue circles and the average over all arrays is shown in black for each value of A. The solid line is a fit to the model of Eq. (2), determining the natural rate of opening (k_o) in the absence of force. (c) Measured extensions (x_core) released during release of the inner wrap for n = 25 arrays (same as in (b)) are shown in blue and averaged in black. Solid line shows the averaged value over all A, as no change within uncertainty across A was observed. Error bars, where visible, are SEM, and figure adapted from [50]

3.6 Quantifying Force Disruption Data

• 1. With calibration of the force and extension, the full cycle of extension and release is methodically compared to a well-known model of polymer elasticity. The extensible Worm-Like Chain (eWLC) is used to characterize double-stranded DNA over a wide range of solution conditions. An analytical solution is known for the high force limit [19, 32,33,34,35],

  $$ b(F)=B\left[1-\frac{1}{2}{\left(\frac{k_BT}{PF}\right)}^{1/2}+\frac{F}{S}\right]\cdot $$

  (1)

  • The measured extension as a function of the applied tension, b(F), is determined by a set of intrinsic polymeric parameters: the end-to-end contour length (B), the polymeric bending stiffness known as the persistence length (P), and the enthalpic stretch modulus (S, see Note 23). Values of these parameters have been measured in many circumstances (see Note 24), and typical values match those fitted to the constructs measured here: B = 0.340 ± 0.001 nm/bp, P = 42 ± 1 nm, and S = 1000 ± 100 pN.

• 2. Changes in the measured length of the DNA (Fig. 3a) are attributed to DNA unwrap** from the nucleosome. Beginning with the fully unwrapped construct, at maximum extension, a custom algorithm searches for divergences between the eWLC model and the data (see Note 25). Each disruption is attributed to a disruption of the inner wrap, and both the force and extension change (see Note 26) are recorded and plotted against the number remaining (A, see Fig. 3b, c as well as Table 1 for n = 25 arrays).

• 3. Continuing to low force, a small plateau may be observed below 10 pN, and the extension change here is due to collective unwrap** of the outer 1/2 wraps. Table 1 summarizes the sum of the length of all released DNA.

• 4. Unwrap** the core DNA above 10 pN gives distinct events that may be arranged in order of observation (here by the number remaining, A). Despite variations, the averaged force shows a clear increase as the number remaining decreases. This is due to the nonequilibrium rate of release, relative to the loading rate of the instrument, and can be described by a simple model [17]:

  $$ F=\frac{k_BT}{x_{\mathrm{core}}^{\dagger }}\cdot \ln \left[\frac{\mathrm{d}F}{\mathrm{d}t}\cdot \frac{x_{\mathrm{core}}^{\dagger }}{k_BT\cdot {k}_0\cdot A}\right]\cdot $$

  (2)

  • The observed rip** force (F) will increase as the number remaining (A) decreases, depending upon the loading rate (dF/dT), the distance from the wrapped state to the barrier (x†_core), and the natural (zero force) rate of DNA-histone opening fluctuations (k_o). Fits to the data averaged at each A (Fig. 3b) determine this rate and the distance to the barrier.

• 5. The measured extension change, once corrected for force dependent polymer elasticity (Eq. 1), does not vary within uncertainty across the number released (Fig. 3c) and may be simply averaged across all A.

Table 1 Nucleosome stability affected by FACT domains

3.7 Evaluating the Free Energy and Transition Barrier

• 1. The free energies of the inner and core releases are found in distinct ways. The release of the outer 1/2 wraps, though indistinguishable, is in equilibrium, and the free energy is simply determined by the area between the measured change in the extension (Fig. 3a shows ΔG_outer and see Note 27). The measured value of ΔG_outer = 14 ± 2 k_BT compares well with previous measurements (see Table 2) [3].

• 2. The inner wrap release is a nonequilibrium event, and the area between the polymer models is now formally the work done by the instrument (W_A) at each release,

  $$ {W}_A=\Delta {G}_A-\Delta {G}_{A-1}+{W}_{\mathrm{stiffness}}\cdot $$

  (3)

• This work is the difference between energy required to extend the array with A nucleosomes remaining less the array with A − 1 nucleosomes (shaded in blue in Fig. 3a, see Note 28).

• 3. The free energy of the inner wrap release is deduced from the work, using the method of Jarzynski (see Note 29) [36],

  $$ \Delta {G}_{\mathrm{core}}=-{k}_BT\cdot \ln \left[\sum {e}^{-W/{k}_BT}\right]. $$

  (4)

• The measured value of ΔG_core = 62 ± 2 k_BT is substantially stronger than the energy holding the outer 1/2 wraps.

• 4. Distributions of each force release (at each A, shown in Fig. 3b) can determine transition state parameters, using a simplified form of the solution of the transition state theory of Dudko [37, 38],

  $$ {G}_{\mathrm{core}}^{\dagger }=\frac{F_{\mathrm{max}}\cdot {x}_{\mathrm{core}}^{\dagger }}{k_BT}\left[\frac{x_{\mathrm{core}}^{\dagger }}{x_{\mathrm{core}}^{\dagger }-{P}_{\mathrm{max}}\cdot {k}_BT\cdot e}\left(1-\nu \right)\right]\cdot $$

  (5)

• The transition energy barrier (G†_core) is calculated from the distribution peak probability (P_max, see Note 30) at the peak probability force (F_max, see Note 31). The value of the distance to the transition state (x^†_core) is chosen using the values determined from fits to Eq. 2 (see Notes 32 and 33).

Table 2 Energy landscape of the nucleosome as shifted by FACT

3.8 Characterizing the Effects of a Chaperone Protein

• 1. Nucleosomes are incubated in specific protein concentration and tethered in the flow cell. A full FD cycle is sought and analyzed as for isolated arrays shown above (see Note 34).

• 2. Force extension curves (Fig. 4a for examples) are scrutinized for changes in length due to unwinding of the outer 1/2 wraps and the inner wrap, as described above.

• 3. The measured force of the inner wrap release is the most sensitive method for judging ligand activity, as compared to the measured length, or other parameters that result from additional analysis steps. Forces are averaged across the entire array (Fig. 4b) for a single cycle of extension and release, and fresh arrays are averaged together (n is typically three or greater). Proteins that destabilize core DNA-histone interactions include human HMGB isolated from SSRP1 and yeast NHP6A, and lower forces are required to fully disrupt the core wrap. In contrast, STP16 MD stabilizes histone-DNA interactions as this protein must bind both histones and DNA and must be removed to facilitate further unwrap**.

• 4. The concentration-dependent changes to the measured disruption force is fit to a simple model that combines the Hill equation with a linear model of changing force with protein binding [17, 35],

  $$ F\left(\Theta \right)={F}_{\mathrm{nucl}}-\left({F}_{\mathrm{nucl}}-{F}_{\mathrm{protein}}\right)\cdot \Theta \cdot $$

  (6)

• Here, F(Θ) is the measured average force of inner wrap disruption for a given fractional occupancy Θ (Θ, which is related to the ligand concentration through the Hill Eq. (see Note 35). F_nucl is the average force in the absence of any ligand, and F_protein is the value determined at saturation. Nonlinear fits determine the equilibrium binding dissociation constant (K_D), summarized in Table 1 (see Note 36). The equilibrium dissociation constant for full FACT (K_D = 26 ± 3 nM) is less than the individual active domains and compares well with other techniques [39].

• 5. Both the released length of the outer 1/2 wraps and the core wrap are measured to find the total length (Fig. 4c and see Note 37). Saturating concentrations of protein clearly displace the outer 1/2 wraps and modestly release of the core wrap is seen, though the histone-DNA interactions of the core sites (the tetramer and the dimer) appear largely intact.

• 6. To separately distinguish protein affinity for the nucleosome and for free DNA, the titration may be repeated on the DNA construct in the absence of any nucleosome. As there is no nucleosome disruption to characterize, measured force extension curves are fit to the eWLC model of Eq. (1). Protein binding will alter the flexibility of the DNA, and this is quantified through fitted changes to the measured persistence length, P(Θ) (see Note 38) [17, 35],

  $$ \frac{1}{P\left(\Theta \right)}=\frac{1-\Theta}{P_{\mathrm{DNA}}}+\frac{\Theta}{P_{\mathrm{protein}}}. $$

  (7)

• As above, Θ is the fractional occupancy of DNA binding sites. In the absence of any protein, DNA has a persistence length of P_DNA (see Note 39), and when saturated with protein, the value will be P_protein. Measured and fitted changes in the persistence length vary with different protein concentrations (Fig. 4d). Nonlinear fits find the equilibrium binding dissociation constant (K_D), describing the affinity to DNA, summarized in Table 2 (see Note 40).

• 7. For all proteins studied here, the values of K_D reveal a stronger affinity for the complete nucleosome than for bare DNA (Fig. 4e and Table 1). The reason varies by protein. In the case of both NHP6A and SSRP1 HMGB, this is due to the well-known preference of HMG for bent DNA, which binds the entry/exit points of the nucleosome, while binding more weakly to bare DNA. By contrast, SPT16 MD simultaneously binds both DNA and the histones of the nucleosome. The activity of full FACT is driven by a combination of these effects, plus the coordinated DNA/histone binding of other subunits (notably SPT16 CTD) [39].

Fig. 4

Characterizing the activity of nucleosome chaperones in Force Disruption (FD) experiments. (a) Typical FD curves for nucleosomes (black), and nucleosomes exposed to saturating conditions of full FACT (purple), as well the isolated domains SSRP1 HMGB (blue) and SPT16 MD (red). A color-coded key locates these domains in the full protein. (b) Averaged release force for the inner wrap versus concentration of added FACT (purple), SPT16 MD (red), SPT16 CTD (pink), SSRP1 HMGB (blue), and NHP6A (green). The activity of the protein is reliably measured and then fit to a simple model to determine the binding affinity (K_D) of the domains to the nucleosome. Note that SPT16 CTD shows little effect. (c) Total released base pairs from both the outer 1/2 wraps (x_outer) and inner wrap (x_core) for the full protein and the active domains. Dotted lines here are guides to the eye. (d) Fitted persistence length (P, to Eq. 1) versus concentration for active proteins, characterizing binding to DNA without any nucleosomes. Lines are fits to simple binding isotherm to determine the binding affinity of these domains to DNA. (e) Summary of fitted values of (K_D) showing all domains have a weaker affinity for DNA, compared to the nucleosome. Values are also summarized in Table 1. All errors are SEM for n ≥ 3 arrays and figure adapted from [17, 35, 50]

3.9 Quantifying Changes to the Energy Landscape

• 1. Fitting the release forces to averaged arrays in the presence of protein shows increases in the kinetics of histone-DNA interactions (Fig. 5a, b) for full FACT and SSRP1 HMGB.

• 2. In the presence of saturating concentrations of protein, the measured energy of nucleosome release increases for SPT16 MD, while increasing for all other proteins.

• 3. The transition state barrier is found as above. Values are summarized in Table 2 and graphically in Fig. 5c and compare to the values obtained for the nucleosome alone (see note).

• 4. These landscape parameters confirm the stabilizing role of SPT16 MD, while SSRP1 HMGB both weakens DNA-protein interactions across the core sites, increasing the kinetics of histone-DNA fluctuations.

Fig. 5

Kinetics and energy landscapes of unwrap**. (a) Fitted values of the opening rate (k_o) as a function of protein concentration for nucleosomes in the presence of saturating FACT (purple), SPT16 MD (red) and SSRP1 HMGB (blue). Lines are a guide to the eye. (b) The values of the opening rate at saturation. Only the HMGB domain leads to more rapid DNA-histone fluctuations. (c) Simple energy landscape is shown for nucleosomes (gray), where the free energy (ΔG) and length change of unwrap** (Δx) is determined for both the outer 1/2 wraps (“outer”) and the inner wrap (“core”). Though the full landscape is not determined, the last, rate-limiting barrier to core unwrap** is found (both x^†_core and G^†_core). While FACT, SSRP1 HMGB, and SPT16 MD displace the outer 1/2 wraps, only SPT16 MD apparently stabilizes the core as both FACT and SSRP1 HMGB destabilize core DNA-histone interactions. Values are summarized in Table 2 (Figure adapted from [50])

3.10 Confocal Imaging (CI) and Kymographs in Combined Instrument

• 1. Experiments begin with construct assembly in a laminar flow cell, commercially produced for the LUMICKS confocal C-Trap (Fig. 6a). Streptavidin-coated beads are caught at the focus of a split 1064 laser (see Note 41). Nucleosome arrays are tethered, translated into a buffer (protein and array free) channel. A quick confocal image verifies a clean catch of a single array (Fig. 6a, inset).

• 2. Confocal images are first collected at a fixed force of ~1 pN (Fig. 6b, see Note 42) and then at 40 pN (Fig. 6c), after progressive disruption of the full array (with all released). Though the images are diffraction limited, the extension change due to wrap** may be observed (Fig. 6b, c).

• 3. Once a fresh array is tethered and translated into a free channel (see Note 43) or a channel containing saturating values of protein (see Note 44), a kymograph measures array intensity versus time (see Notes 45 and 46). Kymographs are initiated at 1 pN, and the force is increased again to 40 pN to complete core disruption (the resulting change in extension is visible in Fig. 6d, e, see Note 47).

• 4. Kymograph data is first processed in custom LAKEVIEW software (see Note 48), to produce an image of photon counts. From there, the image is quantified in FIJI, to plot the photon count versus time (Fig. 6f).

• 5. This plot is fitted using the single exponential algorithm in FIJI (results in Fig. 6f), which gives a half-life of observed fluorescence. Results for nucleosomes and those exposed to FACT and SSRP1 HMGB and SPT16 MD are compared (Fig. 6g).

• 6. Upon disruption, histones are not immediately lost to solution, a result seen previously for nucleosomes labeled on H3 and more generally across all of the octamer (see Note 50) [40]. Interestingly, both SSRP1 HMGB and SPT16 MD induce rapid loss of disrupted histones, with full FACT driving the fastest loss (see Note 51).

Fig. 6

Observing the loss of disrupted histones in Confocal Imaging (CI). (a) The progressive assembly of experimental constructs in a laminar flow cell, where protein is exposed to arrays before or after tethering (shown here). Insets show confocal images for Alexa488 or Atto647N, with corresponding diffraction limited spots at the location of the array. (b) CI of array (red), with bead autofluorescence (green) at low force. Individual nucleosomes are not resolved. (c) As the array is extended, DNA release from the nucleosome expands the fluorescence signal. (d) Kymograph of array fluorescence versus time (the experiment begins at the top of the image and proceeds downward) shows that force disruption does not lead to the immediate dissociation of the histones from the DNA. (e) The presence of saturating conditions of FACT leads to faster loss of disrupted histones from DNA. (f) Measure fluorescence signal decreases with time after disruption (F = 40 pN) for nucleosomes (black) and nucleosomes with FACT (purple), SPT16 MD (red) and SSRP1 HMGB (blue). (e) Fitted values of the fluorescence half-life show that both the subunits and the full protein increase the rate of histone loss from the DNA after disruption. Errors are SEM for n ≥ 5 and figure adapted from [50]

3.11 Survival Probability (SP) Across Cycles of Disruption and Release

• 1. In the FD experiments above, only the first extension release cycle is collected/analyzed. Yet subsequent cycles can also be collected and analyzed as above (Fig. 7a, b) (see Note 52).

• 2. The survival probability (Fig. 7c) is scaled to the number of inner wrap releases observed compared to the first cycle (see Note 53), and the rupture forces (Fig. 7d, see Note 54) and extensions (Fig. 7e, see Note 55) are analyzed as above.

• 3. Only full FACT, combining SPT16 MD and SSRP1 HMGB, restores the nucleosome after release, over as many as 30 cycles (see Note 56). Note that here, failure to observe any rewrap** indicates not only dimer loss but likely tetramer loss as well, though it should be noted that the tetramer may remain bound to the DNA but not able to wrap it (due to displacement of the of the DNA from the central dyad).

• 4. Decreases in the observed rupture forces and the opening length indicate some progressive weakening/loss of the histone/DNA interactions in the core holding the inner wrap.

Fig. 7

Measuring rewrap** in Survival Probability (SP) experiments. (a) After rapid release of force, a new cycle of disruption shows some fraction of the original 12× nucleosomes has rewrapped. Three consecutive cycles are shown for an array of nucleosomes, where only some are observed to reform. (b) In saturating concentrations of FACT, not only is destabilization evident relative to (a), but complete reformation of the inner wrap is observed. (c) The survival probability, the fraction of each array observed in subsequent disruption experiments, is shown for nucleosomes (black), nucleosomes in the presence of saturating FACT (purple), SPT16 MD (red) and SSRP1 HMGB (blue). Only the full protein acts as a chaperone. (d) The measured disruption force of the inner wrap (F) decreases with the number of disruption cycles, indicating progressive DNA-histone disruption. (e) Measured change in extension during disruption of the inner wrap (x_core) varies only slightly with increasing cycle, indicating most of the inner wrap is able to reform (Figure adapted from [50])

3.12 Comparisons

The results of combined FD, CI, and SP experiments show that FACT both destabilizes the nucleosome and chaperones its reassembly after disruption (Fig. 8). Furthermore, destabilization is driven chiefly by two domains, SSRP1 HMGB and STP MD, though each acts distinctly. SPT16 MD stabilizes overall nucleosome interactions, via binding to both DNA and the octamer. SSRP1 HGMGB destabilizes the nucleosome, through binding to the bent DNA of the entry/exit points. Interestingly, yeast NHP6A is shown to have an even stronger effect compared to the human analogue. Though the binding of each subunit disrupts the nucleosome and causes the release of the outer 1/2 wraps of DNA, individually neither can reunite the disrupted DNA with the histones. Only full FACT can restore the nucleosome after disruption, increasing the kinetics of DNA-histone interactions and acting as a true chaperone. The combinations of FD, CI, and SP experiments here can characterize the activity of a wide range of nuclear chaperones and determine their role in transcription, replication, and repair.

Fig. 8

Full FACT catalyzes reorganization of the nucleosome. Simplified diagram illustrating the unwrap** of DNA in FD experiments and the dislocation and release of histones from the DNA seen in both CI and SP experiments. Adding either SPT16 MD (red) or SSRP1 HMGB (blue) disrupts DNA-histone interactions and drives the loss of the outer 1/2 wraps. SSRP1 HMGB achieves this through direct DNA binding while SPT16 MD binds both DNA and histones. While full FACT also disrupts histone-DNA interactions and drives off the outer 1/2 wraps, the release of the disruption force facilitates the restoration of the inner wrap, as the two key subunits coordinate to tether the histones

4 Notes

• 1. The wavelength of this laser is specified to be in the near IR (850 nm), to avoid water absorption lines at longer wavelengths, which would generate local heating. Systems that use 1064 nm lasers for trap** must be carefully adjusted to minimize this effect.

• 2. A 45-degree angle minimizes protein/DNA aggregation that accumulates at the bead-tip interface.

• 3. Smaller beads provide better resolution in FD and CI experiments, though larger beads facilitate catching and allow higher trap** forces and require smaller corrections to the stiffness.

• 4. Full FACT lacks the SPT16 NTD and SSRP1 CTD, as these constructs are difficult to reliably purify without significant aggregation.

• 5. For the full protein, cell suspensions containing the two subdomains are combined at a 1:2 volume ratio (SPT16:SSRP1) just before lysis and purification. This ratio ensures proper stoichiometry of the SPT16-SSRP1 complex after purification using the N-terminal His₆-MBP tag on SPT16.

• 6. Linker length determines the degree of interaction between adjacent nucleosomes. The 60 base pair linkers shown here lead to ~20 nm separation between adjacent nucleosomes, even at low external forces of ~1 pN. Thus, adjacent nucleosomes are expected to have minimal direct interaction.

• 7. Distinct labels are required on each terminus for constructs whose length is very small relative to the size of the beads used in the force disruption experiments. Generally, the construct length should be comparable to the diameter of the smallest bead, and here, the reconstituted array construct is ~1 micron at 1 pN stretching force, while the smallest bead is 2.11 microns in diameter.

• 8. Longer “handles” facilitate clearer confocal images, which may be otherwise complicated by bead autofluorescence. Longer “handles” also allow the same label to be used on each terminus. Here, the reconstituted array construct is ~3 microns long at 1 pN stretching force, while the beads are 1.76 microns in diameter.

• 9. A mass ratio 1.2 octamers to DNA sites (excluding handles) results in arrays with nearly 12× octamers on the positioning sites while they are minimized on the handles.

• 10. While smaller volumes require less sample, sealing the dialysis membrane over a smaller volume is difficult to do without also entrap** an air bubble that will inhibit dialysis.

• 11. Dialysis membranes were chosen with a cutoff of 6–8 kDa and secured in place with an O-ring.

• 12. Sample quality was checked in AFM images, where ~12× octamers could be counted.

• 13. The experimental buffer used here includes 100 mM Na⁺, 10 mM HEPES at 7.5 pH at room temperature. Slightly lower salt buffer (50 mM Na⁺) is also used as the lower salt stabilizes the nucleosomes arrays. However, at very low salt buffer, the biotin–streptavidin link is more difficult to establish.

• 14. For the experiments shown throughout this work, both commercial anti-digoxigeninated (A-D) beads from the LUMICKS company and in house beads are used. Custom A-D beads can be made by coating anti-digoxygenin to protein G labeled 2 micron diameter beads. We have found that the commercial beads are more uniform in size and have better tethers to the nucleosome arrays.

• 15. Varying sizes of polystyrene beads are useful, though limited by the laser trap diameter. Larger bead sizes are easier to catch and offer more surface area for tethers but are more likely to simultaneously catch multiple arrays. It is also preferable that differently labeled beads be of easily distinguishable sizes. Furthermore, within each stock solution, the beads should not vary strongly (>10%) in diameter, to prevent error in either instrument calibration or stiffness as described.

• 16. The reconstituted nucleosome array should be freshly diluted 1:10,000 in experimental buffer on every data collection state. Diluted nucleosomes last 5–6 h, after which they steadily degrade.

• 17. It is helpful to flow in buffer in between cycles of catching fresh beads as the large concentration of nucleosome array constructs quickly saturate the flow cell. The constructs coat the perimeter of the micropipette tip and the sides of the flow cell.

• 18. Detectors, running at 20 kHz, typically each measure 100 data points of the trap** laser deflection (linearly proportional to the trap** force). Data shown here is averaged at this rate.

• 19. The rate of the force increase is the loading rate. Here, that rate is ~10 pN/s, in the middle of the useful range of this instrument and typical for instruments of this type.

• 20. The time for a full cycle is 5–10 s, depending on the number of data points required to reach the fully disrupted state from the starting extension.

• 21. The instrument is calibrated via overstretched long DNA (phage-λ, 48,500 base pairs), which provides a clear force plateau for calibration of 60–65 pN (including known variations due to experimental conditions).

• 22. Measuring the trap stiffness corrects for the movement of the trapped bead with increasing force. To find this value, first a bead should be pulled on the micropipette tip. Then, the piezo moves the bead across the length of the empty optical trap. The stiffness correction is defined by the displacement of the bead as the trap** laser is deflected over the relevant trap** range. For the 3 micron diameter beads (in the center range of those used here), this stiffness was measured to be ~110 pN/μm.

• 23. The enthalpic stiffness is a phenomenological addition, not found in the original solution.

• 24. Variations due to construct length and solution conditions are well known [41, 42].

• 25. The force extension data follows the eWLC model until a “rip” is observed, which presents as both a sudden increase in the extension coupled with a drop in the measured force (due to the finite instrument stiffness and averaging rate). Thus, simultaneous force and length increases (relative to the unwrapped state) are sought to discriminate unwrap** events from noise. Typical thresholds were 0.3 pN and 7 nm, and note the length threshold is significantly lower than the typical lengths measured (between 20 and 25 nm). Finally, as mentioned above, the instrument is averaging the data during collection.

• 26. The change in extension is converted to the number of base pairs, using Eq. (1).

• 27. In practice, the extension change between the force thresholds of 5 and 2 pN is found, corrected for the expected change due to Eq. (1) and converted to bases.

• 28. An estimate of the small correction due to the finite instrument stiffness (W_stiffness) and representing the area during the drop in force as the extension increases is also added.

• 29. The more robust methods of Crooks and Bennett are difficult to implement as closing events were difficult to reliably observe [43, 44]. Where data could be taken, preliminary calculations produced values that appeared to agree with the results shown here.

• 30. P_max is formally a probability density, with units of (pN⁻¹).

• 31. F_max and P_max are both a function of the loading rate.

• 32. Values are determined at each value of A and averaged. The values should be similar across all A, though there are often variations due to experimental uncertainty, especially in P_max.

• 33. The shape factor, ν, is formally 1/2 or 2/3. The value of 1/2 is used here.

• 34. Here, arrays are selected where the number of inner wrap releases (N) was nine or greater, though N > 10 was typical. For every concentration, at least two arrays where N = 12 was sought to avoid biasing the measured force toward higher values.

• 35. The simple Hill Eq. relates the occupancy (Θ) at each binding site to the applied ligand concentration (c) through the equilibrium dissociation constant (K_D),

  $$ \Theta =\frac{1}{1+{\left(\frac{K_{\mathrm{D}}}{c}\right)}^H}. $$

  (8)

• When K_D is equal to c, the occupancy is 1/2. As the sites are assumed to be non-interacting, H is set to unity, which roughly agrees with previous affinity measurements [39]. However, in the case of NHP6A binding, a value of H = 2 fit the data best implying a small degree of cooperativity.

• 36. While in principle all three parameters could be independently fit to the data, in practice, F_nucl is well characterized and thus is fixed.

• 37. Errors in the total length will be largely determined by the error in the measuring the outer 1/2 release.

• 38. Equation (1) is fitted to the extension data using a nonlinear χ² Levenberg–Marquardt algorithm, and details of this analysis have been shown previously.

• 39. The value of P_DNA was measured here, and the value of 42 ± 2 pN agrees with previous work that measured variations for shorter lengths of DNA.

• 40. Now P_DNA is fixed and K_D and P_protein are determined by the fits. The Hill parameter (H) was fixed at one, though in the case of NHP6A, H = 1.5 gave the best fits and indicate weak cooperativity.

• 41. Trap** laser power was reduced to ~200 mW, to minimize heating due to absorption by water at 1064 nm.

• 42. Confocal laser power must be systematically reduced to minimize photobleaching while maximizing signal intensity.

• 43. The free channel may be passivated with a sequential application of BSA and Pluronic. An alternative and simpler approach prerinses each channel with 0.2% Tween20.

• 44. To minimize protein attachment (and dye contamination) of the flow cell chamber, arrays may be caught and then translated into a protein containing chamber previously passivated. Alternatively, arrays may be incubated with the protein, as this binding is stable. Protein exposed arrays are then translated into a buffer only chamber to minimize background.

• 45. The range of the kymograph is shown in the red box on the image in the inset to Fig. 6a.

• 46. The green blocks correspond to autofluorescence of the beads, also visible in the confocal images.

• 47. Kymographs are also collected at 1 pN, to check the rate of photobleaching (k_bleach). The measured rate is used to correct all subsequent measured release rates (k_release), according to [45],

  $$ I={I}_{\mathrm{o}}\cdot {e}^{-{t}_{\mathrm{release}}\cdot {k}_{\mathrm{release}}}\cdot {e}^{-{t}_{\mathrm{bleach}}\cdot {k}_{\mathrm{bleach}}}\cdot $$

  (9)

• 48. A more robust technique combines an oxygen scavenger (GODCAT, glucose oxidase and catalase) with a triplet suppressor (Trolox) [46].

• 49. As the relative intensity may be saturated or washed-out during image processing, image data is processed on the photon counts for this comparison.

• 50. In this study, H3 is labeled with an anti-H3 antibody attached to a dye, and NHS ester labeling is used to label histones generally.

• 51. Critically, the H2A are dye labeled here, and the loss in signal here indicates only dimer loss. The tetramers may be lost at the same time or may also be lost shortly after (see results below).

• 52. The measured half-lives from the previous sections indicate that to forestall histone loss, the arrays should remain unwrapped for less than 1–2 s.

• 53. Only arrays were N > 10 are extended/released over these experiments, to minimize any potential disruption from the tethering process.

• 54. As the cycle number increases and the number of nucleosomes left on each array decreases, Eq. (2) predicts that the average disruption force should increase. Yet this is not observed, indicating some progressive destabilization relative to native nucleosomes.

• 55. The observed unwrap** length appears nearly constant with increasing cycle number, in contrast to the observed force. While DNA appears to wrap up to the dimers, the rewrapped state is not fully stable compared to the non-disrupted nucleosome.

• 56. Restoration was limited to 30 cycles as the observed release force gradually decreased with each cycle, eventually crossing below the low force limit of the threshold chosen here.