Proteins often exist as ensembles of interconverting states in solution which are often difficult to quantify. In the present manuscript we show that the combination of MS under nondenaturing conditions and AUC-SV (analytical ultracentrifugation sedimentation velocity) unambiguously clarifies a distribution of states and hydrodynamic shapes of assembled oligomers for the NAP-1 (nucleosome assembly protein 1). MS established the number of associated units, which was utilized as input for the numerical analysis of AUC-SV profiles. The AUC-SV analysis revealed that less than 1% of NAP-1 monomer exists at the micromolar concentration range and that the basic assembly unit consists of dimers of yeast or human NAP-1. These dimers interact non-covalently to form even-numbered higher-assembly states, such as tetramers, hexamers, octamers and decamers. MS and AUC-SV consistently showed that the formation of the higher oligomers was suppressed with increasing ionic strength, implicating electrostatic interactions in the formation of higher oligomers. The hydrodynamic shapes of the NAP-1 tetramer estimated from AUC-SV agreed with the previously proposed assembly models built using the known three-dimensional structure of yeast NAP-1. Those of the hexamer and octamer could be represented by new models shown in the present study. Additionally, MS was used to measure the stoichiometry of the interaction between the human NAP-1 dimer and the histone H2A–H2B dimer or H3–H4 tetramer. The present study illustrates a rigorous procedure for the analysis of protein assembly and protein–protein interactions in solution.
- analytical ultracentrifugation
- nucleosome assembly protein 1 (NAP-1)
- protein assembly
- sedimentation velocity
Protein assembly is important in Nature for the exertion of function. Some proteins self-assemble into higher oligomers and are thought to adapt their oligomeric state according to the cellular environment. For example α-crystallin, a small heat-shock protein, plays a key role in maintaining lens transparency by chaperoning structurally compromised proteins. MS under non-denaturing conditions revealed the formation of α-crystallin higher oligomers with assemblies containing between 20 and 30 subunits . Although MS analysis gives us the molecular mass of such oligomers with extremely high resolution, buffer conditions that can be used in MS measurements are limited. On the other hand, AUC (analytical ultracentrifugation) provides the opportunity to investigate interaction states in various solution conditions. In addition, recent developments in the computational analysis of AUC-SV (AUC sedimentation velocity) data have offered the possibility of quantitative analysis for protein interactions. For example, self-assembly states of tubulin were analysed by direct curve fitting of the sedimentation boundary [2,3]. C(s) analysis, implemented by SEDFIT , is a very useful and frequently exploited software for investigating the distribution pattern of existing oligomers in solution. Although the sedimentation coefficient (s-value) distribution pattern for each oligomer is precisely evaluated by c(s) analysis , c(s) analysis isn't suited to the evaluation of molecular mass, because the analysis assumes that each oligomer has the same frictional ratio (f/f0), whereas the shapes of individual oligomers are not identical and thus have different frictional ratios. Estimation of molecular mass requires correct f/f0 values for each oligomer with different s-values. Alternatively, SEDPHAT developed by Schuck et al.  provides the possibility of determining the combined s-value and f/f0 when useful prior knowledge, such as the assembly unit, is available. Although 2DSA (two-dimensional spectral analysis) in ULTRASCAN  also provides discrete s-values and f/f0 for each oligomer. Where heterologous oligomers exist, however, accurate determination of the molecular mass for individual oligomers is still difficult mainly due to the uncertain f/f0 estimation.
In the present study, we developed a new method for establishing the assembly states of protein oligomers using the combination of MS under non-denaturing conditions and AUC-SV. The assembly state of NAP-1 (nucleosome assembly protein-1) was investigated.
The nucleosome is the fundamental unit of assembly of chromatin fibres in higher eukaryotes, and consists of DNA plus four distinct histones, H2A, H2B, H3 and H4 [7–9]. Nucleosome assembly is not required for translation, but it is required for chromatin replication . During these processes, histones are delivered to naked DNA by proteins known as histone chaperones [11–13]. It has been proposed that assembly of the nucleosome in vivo takes place in two steps . Initially, CAF-1 (chromatin assembly factor 1) mediates H3–H4 deposition on to naked DNA [15,16] and subsequently NAP-1 delivers the H2A–H2B dimer. Additional nucleosome/histone chaperones have been described since 2003 [17–20] and NAP-1 is regarded as a multi-functional protein in interphase since its function in nuclear import and also RNA processing has been reported previously [21,22].
Nucleosome formation from DNA and histones can be achieved in vitro in the presence of NAP-1 . The current picture that is emerging is of NAP-1 facilitating exchange of the nucleosome histones in various processes that require deformation of the rigid nucleosome assembly so that the genomic information of DNA becomes accessible.
Biochemical or biophysical characterization of NAP-1 has been carried out previously using yNAP-1 (yeast NAP-1) [19,24,25] or dNAP-1 (Drosophila NAP-1) . These previous studies showed that dNAP-1 and yNAP-1 mainly form dimers [24,26]. In addition, the formation of yNAP-1 higher oligomers at physiological ionic strength has been suggested . A subsequent study concluded that there is an equilibrium between dimers, octamers and hexadecamers of yNAP-1 on the basis of AUC . In vitro studies suggest the formation of yNAP-1 complexes with both H2A–H2B dimers and (H3–H4)2 tetramers [18,24,25,28–31]. In these previous studies, proteins from heterologous sources were employed such that histones from Xenopus laevis and yNAP-1 were used for the analysis. It should be noted that the similarity between histones from yeast and Xenopus is relatively high (H2A, 84%; H2B, 77%; H3, 87%; H4, 90%); however, the similarity of NAP-1 in yeast and Xenopus is low (29%). Thus the studies outlined above have yet to reveal a general understanding of NAP-1.
We have now investigated extensively the biophysical properties, self-assembly and histone-binding states of hNAP-1 (human NAP-1) and yNAP-1 in solution using the combination of MS and AUC-SV. Both hNAP-1 and yNAP-1 constitute a stable dimer in solution as the assembly unit. Unexpectedly, the dimers form only even-numbered higher oligomers, with a small proportion of monomer (less than 1%) in the concentration range investigated in contrast with previous reports for yNAP-1. On the basis of these results and SOMO (SOlution MOdeller) simulation, assembly models for higher oligomers are proposed.
In addition, the stoichiometries of the NAP-1–histone interactions were determined unambiguously by ESI-MS (electrospray ionization MS) and used to interpret the differences between hNAP-1 and yNAP-1 in terms of histone recognition.
MATERIALS AND METHODS
Preparation of recombinant hNAP-1 and yNAP-1
Full length hNAP-1 was cloned from the Human Leukocyte Large-Insert cDNA Library, HL5509u (Clontech) using the primers CACCGATGATGATGATAAGATGGCAGACATTGA-CAACAAAGAACAGTCTGAAC and TCACTGCTGCTTGCA-CTCTGC. The core region of hNAP-1, amino acids 37–347, was amplified using the primers CACCGACGAC-GACGACAAACAGCTAACTGTTCAGATGATGCAAAATCC and TCAAATAGCTTCTCCAGTAAAATATAACACTGATC. Vector construction was performed using Gateway technology (Invitrogen) with the pENTR Directional TOPO Cloning Kit according to the manufacturer's instruction. Vectors pDEST14 and pDEST15 were used for protein expression. For each preparation, the expression vector was introduced into Escherichia coli Rosetta (DE3) cells and expression was induced with 1 mM IPTG (isopropyl β-D-thiogalactopyranoside) at a D600 of 0.6–1.0 in LB (Luria–Bertani) broth. The bacteria were incubated for a further 3 h at 37°C. Cells were collected by centrifugation for 20 min at 13000 g and frozen with liquid nitrogen. All hNAP-1 purification steps were carried out at 4°C. The cell pellet was thawed by re-suspending in PBS buffer (140 mM NaCl, 25 mM Na2HPO4, 2.7 mM KCl, 1 mM EDTA and 1 mM 2-mercaptoethanol, pH 7.3). The suspension was sonicated on ice in the presence of 1 mg/ml of lysozyme for five 60 s bursts at full output in a cup horn sonicator (TOMY), and then centrifuged for 2 h at 20000 g. The supernatant was dialysed against GST (glutathione transferase)-binding buffer (10 mM Na2HPO4, 1.8 mM KH2PO4, 140 mM NaCl, 2.7 mM KCl, 1mM EDTA and 1 mM 2-mercaptoethanol, pH 7.3) followed by application on to a GST Trap column (GE Healthcare). After washing with GST-binding buffer, GST–NAP-1 was eluted using GST elution buffer (50 mM Tris/HCl, 10 mM glutathione, 1 mM EDTA and 1 mM 2-mercaptoethanol, pH 7.3). Fractions containing GST–NAP-1 were collected and dialysed against digestion buffer (50 mM Tris/HCl, 1 mM CaCl2 and 1 mM 2-mercaptoethanol, pH 8.0). GST was cleaved using EKMax Enterokinase at 10 units per mg of protein for 16 h. The cleaved sample solution was purified on a Hi-Trap Q column (GE Healthcare) equilibrated with the digestion buffer. The fraction containing hNAP-1 was further purified on a Superdex 200 pg column (GE Healthcare). The biochemical characterization of hNAP-1, such as supercoiling activity assay and binding analysis by pull-down method, is a subject for future study.
yNAP-1 was a gift from Professor Zlatanova (Department of Molecular Biology, University of Wyoming, Laramie, WY, U.S.A.). Concentrations of samples were determined spectrophotometrically using extinction coefficients of 29870 cm−1·M−1 for hNAP-1, and 36040 cm−1·M−1 for yNAP-1.
Preparation of recombinant human histones
Expression and purification of histones was carried out according to methods published previously [24,32,33]. Each expression vector was introduced into E. coli Rosetta (DE3) cells and expression was induced with 1 mM IPTG at a D600 of 0.6–1.0 in LB broth. The bacteria were incubated for 5 h at 37°C. All purification procedures were carried out at 4°C unless otherwise noted. Cells were collected by centrifugation for 20 min at 13000 g and frozen with liquid nitrogen. The cell pellet was thawed by re-suspension in wash buffer (300 mM NaCl and 50 mM Na2HPO4, pH 7.8). The suspension was sonicated on ice for five 60 s bursts at full output in a cup horn sonicator. Inclusion bodies were collected by centrifugation (30 min, 30000 g, 4°C), and the supernatant was discarded. The pellet was re-suspended in an ice-cold desalting buffer (50 mM Na2HPO4, pH 7.8), and the supernatant was separated by centrifugation for 10 min at 20000 g. The pellet containing inclusion bodies was re-suspended in unfolding buffer (8 M urea and 50 mM Na2HPO4, pH 7.8) with the help of a short sonication pulse at room temperature (25°C). Insoluble material was removed by centrifugation for 1 h at 20000 g and the supernatant was applied to a SP-Sepharose column (GE Healthcare) equilibrated with the unfolding buffer. Proteins were eluted with salt gradient buffer (1 M NaCl, 8 M urea and 50 mM Na2HPO4, pH 7.8). Histone proteins were separated on a Protein C4 reverse phase chromatography column (Grace Vydac) using HPLC. Elution was performed with a linear gradient of 0–100% acetonitrile over 10 min in 0.1% trifluoroacetic acid. Chromatography was monitored by absorbance at 215 nm. The peak fractions containing histone proteins were frozen and lyophilized. The purity of the prepared proteins was confirmed to be greater than 95% by SDS/PAGE on 18% gels with Coomassie Brilliant Blue staining.
Preparation of recombinant histone dimers, tetramers and octamers
For octamer production, each lyophilized histone was diluted to a concentration of approximately 2 mg/ml in unfolding buffer (7 M guanidinium HCl, 20 mM Tris/HCl, pH 7.5, and 10 mM dithiothreitol). The concentration of the unfolded histone protein was determined by measuring the A276 of the undiluted solution against unfolding buffer. The four core histone proteins were mixed at equimolar ratios and adjusted to a total final protein concentration of 1 mg/ml by addition of unfolding buffer. The mixture was dialysed against at least three changes of 2 litres of refolding buffer (2 M NaCl, 10 mM Tris/HCl, pH 7.5, 1 mM Na-EDTA and 5 mM 2-mercaptoethanol). The second and third dialysis steps were performed overnight. The histone octamer was maintained at 4°C throughout. Precipitated proteins were removed by centrifugation for 30 min at 20000 g. Samples were concentrated to a final volume of 1 ml before purification on a HiLoad 26/60 Superdex 200 pg gel filtration column (GE Healthcare). Confirmation of the purity and stoichiometry of the histone octamer was performed using SDS/PAGE on 18% gels and the concentration was determined using an extinction coefficient of 29 870 cm−1·M−1 at 276 nm. The histone octamer was concentrated to 3–15 mg/ml, adjusted to 50% (v/v) glycerol, and stored at −20°C. Preparations of H2A–H2B dimer and (H3–H4)2 tetramer were performed according to a method published previously .
Prior to the nucleosome assembly reaction, hNAP-1 and histone octamer were dialysed against supercoiling buffer (10 mM Tris/HCl, pH 8.0, 1 mM EDTA, 150 mM NaCl and 5 mM MgCl2). The nucleosome reassembly reaction was performed according to a method published previously . After the reaction, SDS and Proteinase K (Invitrogen) were added to a final concentration of 0.2% and 100 μg/ml respectively, and incubated for a further 30 min. DNA was purified by phenol/chloroform extraction and separated electrophoretically on a 1% agarose gel, followed by staining with SYBR Green I (Invitrogen). The band intensities were analysed using ImageQuant 5.0 (GE Healthcare).
GST pull-down assay
Glutathione Sepharose 4B (20 μl; GE Healthcare) was mixed with 100 pmol of GST-fused hNAP-1 in 200 μl of PBS followed by incubation with gentle agitation for 3 h at 4°C. The solution was centrifuged at 500 g for 10 min and the precipitant was washed twice with 100 μl of PBST (PBS containing 0.1% Tween 20). Various concentrations of histone octamer in 200 μl of PBST was added to the precipitant followed by incubation with gentle agitation for 16 h at 4°C. The solution was centrifuged at 500 g for 10 min and the precipitant was washed twice with 100 μl of PBST. The precipitant was suspended in 20 μl of 2×SDS sample buffer, boiled for 10 min, and used for SDS/PAGE analysis. Proteins were detected after Coomassie Brilliant Blue staining.
MS under non-denaturing conditions
Solutions of hNAP-1 (100 μM) and yNAP-1 (100 μM) were buffer-exchanged into different concentrations of ammonium acetate, from 150 mM to 1 M, pH 7.5, using a Bio-Spin 6 column before analysis. A solution of hNAP-1 and histone octamer was prepared in buffer (10 mM Tris/HCl, pH 7.5, 1 mM Na-EDTA and 2-mercaptoethanol) and buffer-exchanged with 150 mM ammonium acetate, pH 7.5, using a Bio-Spin 6 column before analysis. The ionic strength I of ammonium acetate is approximately equal to that of sodium chloride (150 mM ammonium acetate I=0.147, 150 mM NaCl I=0.150). All samples were analysed by nanoflow electrospray ionization using capillaries prepared as described previously [34,35]. Capillaries were loaded with 2 μl aliquots of solution, and spectra were recorded on a modified Micromass QToF2 or Synapt HDMS mass spectrometer (Waters) [34,35]. These modified mass spectrometers consist of a nanoflow electrospray interface and a low frequency extended mass range quadrupole, followed by a hexapole collision cell and a ToF mass analyser. Conditions were carefully optimized to allow the ionization and detection of NAP-1 complexes without disrupting the non-covalent interactions that maintain the higher order structures. Typically, the capillary and cone voltages were maintained at 1800 and 120 V respectively, and pressure conditions were maintained at 1×10−2 mbar in the hexapole ion guide. All mass spectra were calibrated against cesium iodide and analysed using MassLynx software (Waters). Molecular mass of each oligomer was calculated from different charge states and m/z.
SV experiments were performed with an XL-A or XL-I analytical ultracentrifuge (Beckman-Coulter) at 20°C using a double-sector charcoal-filled Epon centerpiece (1.2 cm) with sapphire windows. Measurements were carried out in buffer (50 mM Tris/HCl, pH 8.0, 1 mM EDTA and 1 mM 2-mercaptoethanol) with differing sodium chloride concentration between 150 mM and 750 mM. In SV experiments, samples with concentrations of 35 μM for hNAP-1, and 50 μM and 5 μM for yNAP-1 with volumes of 400 μl were centrifuged at 40000 rev./min. Radial scans with 0.001 cm steps were carried out every 5 min in continuous scan mode. The partial specific volumes at 20°C were determined from the amino acid composition to be 0.7244 ml·g−1 for hNAP-1 and 0.7235 ml·g−1 for yNAP-1 using the program ULTRASCAN II 9.6 . Hybrid local continuous distribution and global discrete species analysis were carried out with SEDFIT  and SEDPHAT . The initial s-value for the SEDPHAT analysis was derived from c(s) analysis on SEDFIT. 2DSA GA-MC (2DSA genetic algorithm-Monte Carlo) was carried out using ULTRASCAN II 9.6 .
Biochemical activities of hNAP-1
Supercoiling activity is important for the assembly of histones and DNA into the nucleosome structure. Figure 1 shows the supercoiling activities of hNAP-1. Addition of hNAP-1 results in an increase in supercoiled DNA. These activities are similar to those found for yNAP-1 in analogous experiments . A 3:1 excess of hNAP-1 to histone octamers converted 45% of the DNA in the relaxed form into supercoiled DNA (Figure 1A). The interactions of hNAP-1 with core histones were investigated using a GST pull-down assay. Under the experimental conditions used (150 mM NaCl), the histone octamer is dissociated into two equivalents of the H2A–H2B dimer and one equivalent of the (H3–H4)2 tetramer . As shown in Figure 1(B), hNAP-1 recognizes both the H2A–H2B dimer and the (H3–H4)2 tetramer. When equimolar hNAP-1 is added to histone octamers, hNAP-1 binds the H2A–H2B dimer and the (H3–H4)2 tetramer with equal propensity. These binding properties of hNAP-1 are different from those of yNAP-1, which preferentially binds the (H3–H4)2 tetramer in vitro .
MS of hNAP-1 and yNAP-1 under non-denaturing conditions
Self-assembly of hNAP-1 and yNAP-1 was investigated by two different approaches: MS under non-denaturing conditions; and AUC-SV measurements. Initially, the homogeneity of hNAP-1 and yNAP-1 were assessed by MS under non-denaturing conditions. This provides precise mass information about non-covalently bound complexes co-existing in solution. The mass spectra of hNAP-1 at physiological ionic strength showed two major series of resolved peaks with different charge states in addition to broad low-intensity unresolved peaks at higher m/z (Figure 2A, upper panel). In the case of yNAP-1, four major series of resolved peaks were clearly detected (Figure 2B). It should be noted that the results of both yNAP-1 and hNAP-1 show the existence of monomers with charge states that could be assigned to folded states.
The observed charge states of higher oligomers are assigned to dimer, tetramer, hexamer, octamer and decamer. The molecular masses determined are summarized in Table 1. Considering that no odd-numbered oligomers of hNAP-1 and yNAP-1 are observed except small peaks assigned to monomers, the results indicate that hNAP-1 and yNAP-1 exist as stable dimers and a portion of dimers are further assembled into higher oligomers. The exact masses of unfolded hNAP-1 were obtained by increasing the collision energy. Two series of charge states were assigned to unfolded hNAP-1 molecules with different masses; full-length (45885 Da) and truncated hNAP-1 (45323 Da). Because of the heterogeneity in the primary structure of hNAP-1, the hNAP-1 oligomers have several molecular masses and the observed peaks for hNAP-1 oligomers are broad. At higher ionic strength (750 mM ammonium acetate, I=0.748, Figure 2A, lower trace), the populations of higher oligomers of hNAP-1 were markedly reduced, the dimer being the dominant species.
Similarly, as for yNAP-1, the dimer was predominant at 750 mM ammonium acetate (results not shown). Together, these results indicate that the primary assembly unit of both hNAP-1 and yNAP-1 is a dimer and that higher oligomers are formed at physiological ionic strength. Disruption of higher oligomers at high ionic strength implies that associations of dimers are stabilized by electrostatic interactions.
SV measurements of hNAP-1 and yNAP-1
The assembly states of hNAP-1 and yNAP-1 were examined in solution under physiological ionic strengths by AUC-SV. The results of c(s) analyses for both hNAP-1 and yNAP-1 showed eight different peaks between 0 S and 25 S (Figure 3A). At higher ionic strength (750 mM sodium chrolide), populations of higher oligomers of hNAP-1 and yNAP-1 are not observed and only 5 S species remained. The molecular masses of these 5 S species correspond to ~50 kDa and are assigned to dimeric forms. Small peaks ~3 S observed under physiological ionic strength are assigned to monomeric forms. The s-values of hNAP-1 and yNAP-1 oligomers were found to be concentration-independent (see Supplementary Figure S1 at http://www.BiochemJ.org/bj/436/bj4360101add.htm), indicating the observed sedimentating boundary is not a reaction boundary that occurs in a narrow range of dissociation rate constant between 10−3 and 10−4 s−1. In contrast, the proportions of dimer and oligomers changed depending on the total protein concentration with the proportion of dimer apparently increasing as the concentration decreased.
In such a system, accurate s-values and the concentration of each component can be obtained from c(s) analysis . However, estimation of molecular mass for each oligomer from c(s) analysis is difficult because the analysis doesn't return f/f0 values for individual oligomers, although these are essential for the estimation of molecular masses. We therefore performed ‘hybrid local continuous distribution and global discrete species’ analysis using the program SEDPHAT [5,6] on the basis of the findings from non-denaturing MS measurements where the dimers were found to self-associate to form even-numbered higher oligomers, and a low population of monomers was observed. In this analysis, monomer and even-numbered oligomers, including dimer, tetramer, hexamer, octamer, decamer, dodecamer, tetradecamer and hexadecamer, were assumed, and the dimer molecular mass, s-values of each oligomer and their populations were set as variable parameters. We carried out the analysis with several models composed of 1-2-4-6-8-mers, 1-2-4-6-8-10-mers, 1-2-4-6-8-10-12-mers, 1-2-4-6-8-10-12-14-mers and 1-2-4-6-8-10-12-14-16-mers. As the number of oligomeric species included in the analysis increased, the RMSD (root mean square deviation) value was improved (Figure 3B). The RMSD value plateaued with well-distributed residuals when the 1-2-4-6-8-10-12-mer model was employed, indicating that this model employed in the numerical analysis of AUC-SV data is appropriate and that at least these oligomers are present in solution. Analysis using a previously reported model (2-8-16-mer)  resulted in a significantly higher RMSD value (0.0123) than that with the 1-2-4-6-8-mer model. Similarly, for yNAP-1, the 2-8-16-mer model gave a large RMSD value (0.013), whereas models including 1-2-4-6-8-10-12-mer or higher oligomers had significantly smaller RMSD values. Consistent with the MS results, the AUC-SV results reported here indicates that both yNAP-1 and hNAP-1 have the same assembly states, which are different from those previously reported . All the parameters from AUC-SV and non-denaturing MS are summarized in Table 1. The proportions of each oligomer varied according to the total protein concentration (Figure 3C). As the total protein concentration decreases, the proportion of higher oligomer decreases, whereas the proportion of dimer increases. The proportion of yNAP monomer estimated from AUC is less than 1% at 50 μM and increases slightly to 1.3% at lower concentration (5 μM).
The f/f0 values show that both hNAP-1 dimer and yNAP-1 dimer have 1:5 axial ratios, indicating a rather non-spherical elongated shape. In forming higher oligomers up to octamers, NAP-1 takes a more elongated shape, but beyond octamer, the shape gradually changes from elongated to more spherical (Table 1 and see the Discussion section). In order to assess these results, we analysed the AUC-SV data further using the 2DSA GA-MC method that searches the optimum combinations of s-value and f/f0 with individual populations, without prior information, using high performance supercomputing (see Supplementary Figures S2 and S3 and Supplementary Table S1 at http://www.BiochemJ.org/bj/436/bj4360101add.htm). The three sets of s-values and frictional ratio values estimated from the ULTRASCAN analyses using three independent AUC-SV data were highly reproducible and each s-value of species obtained was closely similar to that estimated by the SEDPHAT analysis. Therefore, the assignment of each species in the ULTRASCAN results could be carried out according to the SEDPHAT result. The ULTRASCAN analysis gave similar f/f0 values for monomer and dimer as the values in the SEDPHAT results. Nevertheless, the combination of s-values and frictional ratio values for each species estimated from the ULTRASCAN analysis gave much smaller molecular masses for most of the higher oligomers (Supplementary Table S1), because the ULTRASCAN analysis provided smaller f/f0 values for oligomers, compared with the f/f0 values for the same oligomers in the SEDPHAT results.
Binding stoichiometry between hNAP-1 and histones investigated by MS
The binding stoichiometry between NAP-1 and histones was investigated by MS under non-denaturing conditions. Prior to the interaction studies, the assembly states of histones in 150 mM ammonium acetate were investigated using MS. The histone octamer prepared in 2 M sodium chloride was buffer-exchanged into 150 mM ammonium acetate buffer immediately prior to ESI-MS analysis. At this ionic strength, the mass spectrum showed four highly charged series. These peaks were assigned to each of the four histone proteins, suggesting the dissociation of octamers into their unfolded component proteins. CD measurements  of the octamers confirmed that the characteristic α-helix peak at 222 nm, corresponding to the histone fold, was only observed at high ionic strength (1 M ammonium acetate). The unfolding curve (Figure 4A) indicated that the unfolding midpoint was at 650 mM, and that at least 700 mM ammonium acetate is necessary to maintain the histone fold. Mass spectra recorded in 750 mM ammonium acetate indicated that the octamer dissociates into two stable components in this buffer. MS/MS (tandem MS) confirmed that the two components were the H2A–H2B dimer and the (H3–H4)2 tetramer (Figure 4B, lower panel and Table 2). Interaction studies between histones and hNAP-1 were therefore carried out in 750 mM ammonium acetate.
The binding stoichiometry of the hNAP-1 dimer to both histone components was investigated using MS (Figure 5 and Table 2). At equimolar ratios of the H2A–H2B dimer to the hNAP-1 dimer, (hNAP-1)2 dimer and (hNAP-1)2(H2A–H2B) heterotetramer was observed. Increasing the molar ratio of the H2A–H2B dimer to hNAP-1 dimer removed free (hNAP-1)2 and peaks corresponding to interactions with two H2A–H2B proteins [(hNAP-1)2(H2A–H2B)2] were predominantly observed. Interactions between the hNAP-1 dimer and the (H3–H4)2 tetramer were investigated in a similar manner. Increasing the amount of (H3–H4)2 tetramer up to a 3:1 ratio of (H3–H4)2 tetramer to hNAP-1 dimer, led to formation of the complex (hNAP-1)2(H3–H4)2, in addition to hNAP-1 dimer and free (H3–H4)2 tetramer (Figure 5 and Table 2). This is a different stoichiometry from that observed with the same molar ratio of hNAP-1 dimer to H2A–H2B dimer.
Assembly states of NAP-1
The results shown in the present study, obtained using AUC-SV and MS under non-denaturing conditions, establish the assembly states of hNAP-1 and yNAP-1. Previous sedimentation studies of yNAP-1 showed the possibility of further oligomerization of dimers at physiological ionic strength . This assembly reaction was thought to occur via a 2–8-16 mechanism . In contrast, our results from the present study indicate that the dimers of hNAP-1 are sequentially assembled into a wide variety of even-numbered higher oligomers, such as tetramer, hexamer, octamer etc. We therefore re-investigated the oligomerization state of yNAP-1 using MS under non-denaturing conditions and AUC-SV. These results clarify that the oligomerization behaviour of NAP-1 is common to human and yeast.
The SEDPHAT results indicate that the proportion of monomer is minor in the current concentration range (μM), although the yNAP-1 dimer is able to dissociate into monomer at more dilute solution conditions. This is quite different from the oligomerization behaviour of the yNAP-1 dimer in this concentration range; the proportion of higher oligomer clearly shows the dependence on total protein concentration. In addition, as ionic strength increases, higher oligomers of yNAP-1 were depleted and the yNAP-1 dimer became the predominant species. The crystal structure of yNAP-1 has revealed that the structural basis for dimer formation is hydrophobic interactions via long α-helices in each molecule . Our results from the present study and the crystal structure suggest that dimerization occurs through rigid inter-molecular hydrophobic interactions between yNAP-1 monomers, whereas higher even-numbered oligomerization occurs through electrostatic interactions. The behavior of hNAP-1 in solution, clarified by AUC-SV and MS, resembles that of yNAP-1. Therefore, the mechanism underling oligomerization is most likely similar for yNAP-1 and hNAP-1.
The results from the present study imply that dimers of both hNAP-1 and yNAP-1 assemble sequentially at physiological ionic strength in solution. These findings differ from previous results on yNAP-1  in which it was reported that yNAP-1 can be described by a dimer-octamer-hexadecamer model based on the AUC-SV and AUC-SE (AUC sedimentation equilibrium) analyses . We show the distribution of only even-numbered higher oligomers for hNAP-1 and yNAP-1 (dimer, tetramer, hexamer and higher oligomers) and reveal the shapes of each oligomer in solution (Table 1). Determination of accurate molecular masses by AUC-SV for co-existing protein complexes having different molecular shapes is difficult, because heterologous f/f0 is not reliably estimated by AUC-SV (Supplementary Table S1). In contrast, determination of the distribution pattern in solution using MS is relatively straightforward, but gas-phase conditions may change the ratio of oligomeric species. The latter does, however, provide accurate molecular masses. The combination of AUC-SV and MS is, therefore, a powerful method for the analysis of complicated protein association systems in solution. 2DSA GA-MC seeks, without prior information, the combination of s-value, f/f0 and concentration that gives the smallest RMSD calculated from the difference between observed sedimentating boundary and numerically generated one employing the combination, under a certain statistical constraints. Therefore, data suitable for estimating accurate s-value and f/f0 might improve the reliability of the 2DSA GA-MC. It is generally considered that the sedimentating boundary at lower rotor speed provides f/f0 with high accuracy whereas that at a higher rotor speed leads to the precise s-value . Therefore, uncertainties in f/f0 as seen in 2DSA GA-MC of yNAP-1 could be improved by global multi-speed AUC-SV analysis using SV data at different rotor speeds as previously reported [6,40].
In order to assess the hydrodynamic parameters obtained experimentally, we calculated f/f0 on the basis of the three-dimensional structures employing the SOMO beads-modelling method . In addition to the crystal structure of yNAP-1 dimer (PDB code 2AYU), we built two assembly models for yNAP-1 oligomers. One is the linear arrangement model (Figure 6D), the other is the circular arrangement model (Figures 6E and 6F). In both models, yNAP-1 dimers are assembled via the same interaction site, β-hairpin, which was suggested as essential for higher oligomerization . The structures of the tetramer are same in both models. Each f/f0 value from AUC-SV analysis (Table 1) and from SOMO simulation were plotted (Figure 6A). The simulated f/f0 value of the dimer (1.38) and tetramer (1.63) agrees well with that from the AUC-SV analysis of the yNAP-1 dimer (1.45) and tetramer (1.68) (Figure 6A). The calculated f/f0 values of higher oligomers compared with tetramers showed distinct trends between the two models; in the linear model, f/f0 values are increasing monotonously as the dimers are assembled. In contrast, the circular models take similar values to each other as the oligomerization state increases, rounded to their circular shapes. The f/f0 values in the circular models have similar trends to those of SEDPHAT analysis of our experimental results obtained using AUC-SV. From these experimental results and theoretical considerations, we suggest that NAP-1 dimers are assembled into higher oligomers through a circular arrangement.
Binding specificity and stoichiometry of hNAP-1 with histones
Previous in vivo studies indicated that yNAP-1 binds to the H2A–H2B dimer, and not the (H3–H4)2 tetramer . Interaction studies of yNAP-1 in vitro also support the binding of yNAP-1 to the H2A–H2B dimer . However, further studies proposed preferential binding of yNAP-1 to the (H3–H4)2 tetramer rather than to the H2A–H2B dimer . It has also been shown that NAP-1 recognizes both the H2A–H2B dimer and the (H3–H4)2 tetramer . The results of pull-down analysis and MS under non-denaturing conditions shown in the present study further support these previous findings. Considering the interaction of hNAP-1 with histones, only one (H3–H4)2 tetramer is bound to one hNAP-1 dimer, whereas two different complexes composed of H2A–H2B dimers and hNAP-1 dimer were observed, with one hNAP-1 dimer recognizing up to two H2A–H2B dimers. Such interactions of hNAP-1 with H2A–H2B dimers may prevent them from self-aggregating before nucleosome formation. We speculate that this could be an alternative functional role of hNAP-1 in vivo.
Masanori Noda and Susumu Uchiyama performed all the experiments and wrote the manuscript. Kiichi Fukui and Carol Robinson designed this research and wrote the manuscript. Adam R. McKay performed MS measurements. Shigeki Misawa, Akihiro Yoshida, Hideto Shimahara and Hiroto Takinowaki performed the gene cloning, protein expression and purification. Yuji Kobayashi performed AUC analysis. Tadayasu Ohkubo and Shota Nakamura performed the structural modelling. Akihiro Morimoto and Sachihiro Matsunaga performed interaction analysis between NAP-1 and core histones.
This work was supported by Grants-in-Aid for Scientific Research from the Ministry of Education, Science, Culture, Sports, Science and Technology of Japan [grant number 18687005 (to K.F. and S.U.)].
We thank Dr Borries Demeler for his useful discussion and analytical help for using ULTRASCAN software, and Dr Peter Schuck for his useful discussion and analytical help using SEDFIT and SEDPHAT software.
Abbreviations: 2DSA, two-dimensional spectral analysis; 2DSA GA-MC, 2DSA genetic algorithm-Monte Carlo; AUC, analytical ultracentrifugation; dNAP-1, Drosophila nucleosome assembly protein 1; ESI-MS, electrospray ionization MS; GST, glutathione transferase; hNAP-1, human nucleosome assembly protein 1; IPTG, isopropyl β-D-thiogalactopyranoside; LB, Luria–Bertani broth; MS/MS, tandem MS; NAP-1, nucleosome assembly protein 1; PBST, PBS containing 0.1% Tween 20; RMSD, root mean square deviation; SOMO, SOlution MOdeller; SV, sedimentation velocity; yNAP-1, yeast nucleosome assembly protein 1
- © The Authors Journal compilation © 2011 Biochemical Society