|Home | About | Journals | Submit | Contact Us | Français|
The histone chaperone Nucleosome Assembly Protein 1 (NAP1) is implicated in histone shuttling, nucleosome assembly and disassembly. Under physiological conditions, NAP1 dimers exist in a mixture of various high-molecular weight oligomers whose size may be regulated by the cell-cycle dependent concentration of NAP1. The functional and structural significance of the observed oligomers are unknown. We have resolved the molecular mechanism by which yeast NAP1 dimers oligomerize by applying x-ray crystallographic, hydrodynamic, and functional approaches. We find that an extended β hairpin that protrudes from the compact core of the yNAP1 dimer forms a stable β sheet with β hairpins of neighboring yNAP1 dimers. Disruption of the β hairpin (whose sequence is conserved among NAP1 proteins in various species) by the replacement of one or more amino acids with proline results in the complete loss of yNAP1 dimer oligomerization. The in vitro functions of yNAP1 remain unaffected by the mutations. We have thus identified a conserved structural feature of NAP1 whose function, in addition to presenting the nuclear localization sequence, appears to be the formation of higher-order oligomers.
The nucleosome core particle (NCP) is the universally repeating unit in chromatin, consisting of an approximately equal mass of protein and DNA in a complex macromolecular assembly of 210 kDa1. Nucleosome assembly is a stepwise process that starts with the association of histones H3 and H4 with the DNA, followed by the incorporation of H2A-H2B dimers to form the nucleosome2. This process is orchestrated by a set of diverse chromatin assembly factors and chaperones 3 4 5.
Nucleosome assembly protein 1 (NAP-1, or NAP1) is a histone chaperone that binds H2A-H2B 6 and linker histones 7 (reviewed in 8). Chromatin assembly activity has been demonstrated in vitro 9 6 10. Since H2A-H2B dimer depositon is likely reversible in vivo, NAP1 has been implicated in promoting chromatin fluidity through the removal of H2A-H2B dimers 11 12 13 14 15. We have shown previously that yNAP1 promotes nucleosome sliding through histone H2A-H2B dimer exchange, which under certain conditions results in histone variant incorporation into existing nucleosomes 11. NAP1 is a member of a growing family of related proteins whose members are implicated in transcription regulation, cell cycle regulation, apoptosis, histone modification, chromatin assembly, and histone shuttling 8.
We have previously determined the crystal structure of yNAP1, and have demonstrated that the protein is an obligate dimer with an extensive dimerization interface 16. Studies of NAP1 or NAP1/histone complexes revealed that NAP1 exists in multiple states of oligomers of varying size 9 6 17 18 19. Using sedimentation velocity experiments, oligomerization was shown to depend on the protein concentration and on the ionic strength of the buffer 17. Similar experiments suggested that yeast NAP1 exists in a dimer-octamer equilibrium 20. Based on sedimentation data, a hexadecameric ring-shaped assembly was proposed for the largest observed yNAP1 complex, similar to the assemblies formed by the unrelated chaperone nucleoplasmin 21. Since formation of higher order assemblies is concentration dependent, and since the nuclear concentration of NAP1 fluctuates in a cell-cycle dependent manner, it has been suggested that the oligomerization of yNAP1 dimers could exist at different times of the cell cycle 22 20.
Due to the strong propensity of yNAP1 dimers to form oligomers under physiological conditions, it is unknown whether dimeric and oligomeric forms of yNAP1 exhibit different functions. Furthermore, the molecular details of the interactions responsible for the oligomerization of yNAP1 dimers are unknown. Here we investigate yNAP1 oligomerization using x-ray crystallography, gel filtration, analytical ultracengrifugation as well as functional assays. We have identified an extended β-hairpin which encompasses the nuclear localization sequence as the site of intermolecular interactions through the formation of an extended β sheet, and have evidence that yNAP1 assemblies become more elongated with increasing size. We have disrupted the β-hairpin using site-directed mutagenesis, and find that this region is indeed responsible for the formation of oligomers, while it has no apparent effect on the various functions attributed to yNAP-1 in vitro.
The central domain of yeast NAP1 (residues 74 to 365; yNAP1(74–365)) retains native-like activity in histone binding and nucleosome assembly 9 23. In contrast to wild type yNAP1 which crystallized at pH 4.85, crystals of NAP1(74–365) were obtained at neutral pH in a tetragonal space group (P42212). The refined structure of this truncated form of yNAP1 is for the most part identical to the previously published structure of full length yNAP1 (Figure 1A). The two structures superimpose with an rmsd of ~ 1Å. Like full length yNAP1, yNAP1(74–365) forms a stable homodimer. However, while the two monomers are related by crystallographic symmetry in full length yNAP1, the asymmetric unit for yNAP1(74–365) encompasses the two monomers of the physiological dimer. The two independently built and refined monomers are identical and superimpose with a least square difference of 0.79 Å.
The only striking difference between the two structures is found in β5 and β6 (Fig. 1A). This region was structurally ill-defined and partially disordered in the previously published structure 16. In yNAP1(74–365) it forms a β- hairpin that extends over 11 amino acids for each β strand and is stabilized by nine hydrogen bonds (Figure 1B). Amino acids 295– 298 form a typical β - turn. The entire hairpin is exceedingly well ordered, as signified by the highly defined electron density (Fig. 1B) and B-factors that are approximately 20 Å2 lower than the average for the entire molecule. The β hairpin protrudes by a distance of 3.7 nm from either side of the compact core of the NAP1 dimer.
The high definition of this region in the present structure can be attributed to its strong contributions to crystal contacts (Figure 2A). A total of eight β-strands are arranged to form a consecutive antiparallel β-sheet that links four yNAP1 dimers. The β-hairpin interacts with the β-hairpin of a neighboring yNAP1 dimer through an extensive hydrogen bonding network over its entire length (Figure 2B). One additional yNAP1 dimer docks on either side of this dimer of dimers. This interaction is slightly weaker due to a larger twist of the β-sheet imposed by lattice constraints. The distance of the adjacent β-strand backbones as well as the distances between facing oxygen and hydrogen atoms are within the canonical range.
It is well established that interactions between exposed edges of β-sheets is an important mode of protein-protein interaction with potentially pathological consequences (for example, 24). To test whether the β-hairpin of yNAP1 is responsible for oligomerization, single point mutations were designed to disrupt the β hairpin (Figure S1A). Arg-290, Arg-301, Thr-302 and Lys-305 were mutated to proline, an amino acid that has a low propensity to form β-sheets and that is generally occluded from naturally occurring β-strands 25 26. A mutant containing all four point mutations (R290P, R301P, T302P, and K305P) was also prepared. All mutations were generated in the full length protein, with the exception of K305P, which was also mutated in yNAP1(74–365). All mutants were expressed in soluble form, and were purified under identical conditions compared to their wild type counterparts (Figure S1B).
The oligomerization state of wild type and mutant NAP1 proteins was investigated by size exclusion chromatography. Since we have previously demonstrated yNAP1 oligomerization below 500 mM NaCl 17, and since the original crystals of yNAP1 were grown at a pH of 4.8 16, we examined the elution profile of wild type yNAP1 (at a concentration of ~50 μM) under three sets of conditions that varied in ionic strength and pH (Figure 3). At physiological pH and low salt, wild type NAP1(1–417) eluted as a broad peak at an apparent size that is indicative of a heterogenous mixture of various oligomeric assemblies (Fig. 3A), as reported 17. At high salt (Fig. 3B), or under acidic buffer conditions (Fig. 3C), the protein eluted from the column at an apparent molecular weight of ~ 95 kDa, consistent with the dimeric form of a 417 amino acid protein. This finding suggests that oligomerization of NAP1 is sensitive to low pH, explaining our ability to crystallize a protein that under physiological conditions is present in various states of oligomerization.
In striking contrast, all mutants eluted as well-behaved apparent dimers under physiological conditions (Fig. 3D), indicating that disruption of the β-sheet by the substitution of a single residue with proline completely abolishes the propensity of yNAP1 to form oligomers. The minor differences in elution volumes between the different mutants may be explained by the degree of disruption of the β-strands. Since they extend prominently from the compact core of the protein (Fig. 1), the presence of residual β strands in some mutants, but not in others, could have a distinct effect on the hydrodynamic radius of the yNAP1 dimer.
The self-association states of wild type and mutant yNAP1 were further analyzed by analytical ultracentrifugation. Sedimentation velocity experiments were performed at ~3 μM protein concentration under all solution conditions. This concentration is ~ 12-fold lower than that used for gel filtration. Boundaries were analyzed by the G(s) method to obtain an integral distribution of sedimentation coefficients that are corrected for diffusion, and are accurate for even highly heterogeneous samples 27. A vertical G(s) plot is indicative of a homogeneous sample while a heterogeneous sample yields a plot with one or more regions of positive slope. For a protein that is >95% pure (as are the samples under investigation here), a significant slope in the G(s) plot is diagnostic for the presence of protein oligomers, and the shape of the plot provides insight into the oligomerization process.
Analysis of full-length wild type NAP1 under moderate salt concentration (100mM NaCl) shows the characteristic G(s) plot of a self-associating system, with solutes exhibiting sedimentation coefficients from ~5–8S (Fig. 4A). This result is in good agreement with a previous sedimentation velocity analysis of NAP1 20, and allows observation of the stable ~4.5S dimers and the larger oligomers. The 4.5S NAP1 homodimer (98,458 Daltons) exhibits a frictional ratio (f/f0) of 1.8, consistent with the asymmetric nature of the dimer 16. In contrast, the G(s) distributions of all mutants tested [NAP1(R290P), NAP1(K305P), and NAP1(R290, R301P, T302P, K305P)], sedimented as a single, homogeneous ~4.5 S species under all conditions tested (Fig. 4A). These distributions are nearly identical to the distribution of wild type NAP1 under high salt conditions (500mM NaCl; Figure 4B, and 17). This suggests that disruption of β5 and β6 does not affect the overall structure of the yNAP1 dimer. The G(s) distributions of NAP1 mutants were unchanged by high salt.
To investigate the possibility that amino acids 1–70 and 370–417 (which are disordered in both structures) contribute to oligomerization, we analyzed the sedimentation behavior of yNAP1(74–365) and found no differences compared to wild type under any condition tested (Fig. 4C, compare with 4A). Thus, neither the N- or C-terminus contributes to yNAP1 oligomerization under these conditions. As expected, the sedimentation coefficient distribution for the proline mutant (K305P) in the context of yNAP1(74–365) shows a nearly vertical plot with a single, ~4.5 S component. However, f/fo is changed from 1.8 for wild type to 1.46 in the mutant (not shown), consistent with loss of frictional drag due to the loss of N- and C-termini. Together, these results show that disruption of the formation of higher order oligomers requires mutation of only one of the residues that contribute to the formation of β5 and β6 of NAP1. Further, it demonstrates that neither the N or C termini of NAP1 contribute significantly to the formation of the higher-order oligomers.
Most in vitro assays of yNAP1 activity were done under conditions where yNAP1 dimers form large macromolecular assemblages, and thus it is unknown whether the various activities of yNAP1 require oligomerization beyond the dimeric state. We tested the ability of all yNAP1 mutants to form histone H2A-H2B/NAP1 complexes, to assemble mono-nucleosomes in vitro, and to dissociate H2A-H2B dimers from assembled nucleosomes. Histone binding was analyzed by GST pulldown assays under increasingly stringent conditions (Fig. 5A). No differences were observed between wild type and mutant proteins, in that all complexes appeared to be stable up to the same ionic strength. GST alone did not interact with histones except for at the lowest ionic strengths (100 and 150 mM; Fig. 5A, lowest panel). To assess the nucleosome assembly activity of NAP1 mutants, histone octamer, DNA, and wild type or mutant NAP1 were mixed, and the assembled nucleosomes were resolved by native PAGE followed by SYBR-Gold staining. The NAP1 mutants showed only minor (if any) defects in this activity (Fig. 5B). The higher migrating bands are routinely observed with salt-dependent as well as chaperone-mediated assembly, and probably represent aggregated/misfolded nucleosomes. Finally, the ability of wild type and mutant yNAP1 to remove H2A-H2B dimers from the nucleosome was analyzed by monitoring the appearance of a (H3-H4)2 – DNA band (in addition to a nucleosomal band) in a native gel (“S3” in Fig. 5C, lanes 2–4), with a concomitant appearance of a yNAP1 – H2A-H2B dimer band (“GST-S2” in Fig. 5C, lanes 10–12), as described earlier 11. These latter experiments were also performed with GST-yNAP1 to achieve better separation between the nucleosome and the yNAP1 – H2A-H2B dimer complex. Together, these data indicate that the β hairpin is not involved in the major yNAP1 in vitro functions identified previously 16.
We have shown that the large macromolecular assemblies previously observed for NAP1 under physiological conditions 28 17 20 29 are the result of intermolecular interactions involving an extended β-hairpin that encompasses the nuclear localization sequence. The structural integrity of the β-hairpin is not required for histone binding, nucleosome assembly, and nucleosome disassembly in vitro, even though all 24 amino acids are conserved among different species. We have thus identified a conserved structural feature of NAP1 whose function, in addition to presenting the nuclear localization sequence, appears to be the formation of higher-order oligomers, with as yet undetermined implications for its in vivo function.
The intermolecular antiparallel β-sheet interactions observed in the crystal lattice resemble those seen in amyloid fiber structures 30 31. An extensive network of hydrogen bonds maintains the intermolecular contacts between individual NAP1 dimers. About 510 Å2 of surface area is buried between two β-hairpins of adjacent yNAP1 dimers in the crystal lattice, an area that is much smaller than the area buried by the interaction of two α2 helices to form the yNAP1 dimer 16. This is consistent with the finding that the interface formed between yNAP1 dimers is destabilized at moderate levels of chaotropic agents without a measurable loss of secondary structure, while complete unfolding is required to disrupt the yNAP1 dimer interface 17. The disordered N and C terminal regions, comprising about 30% of the total mass of yNAP1, have no effect on the dimer-oligomer equilibrium, consistent with a recent report for Xenopus laevis NAP1 29.
In yeast, NAP1 is a mostly cytoplasmatic protein 32. It participates in the active transport of histones from the cytoplasm to the nucleus by forming a ternary complex with H2A-H2B dimers and a member of the karyopherin family of proteins (Kap114). Kap114 recognizes the nuclear localization sequence (NLS) of histones and yNAP1, presumably in a non-canonical manner 32. The NLS has been located to the equivalent of β5 of yNAP1 β hairpin in a variety of species 33 34 32 35. Although the interaction interface between yNAP1 and histones remains to be characterized, we have evidence that the β hairpin does not contribute to the histone interaction (A.J. Andrews and K. Luger, unpublished results). This is consistent with the earlier result that yNAP1 persists as a ternary complex with histones and the NLS-interacting protein Kap114 in the nucleus 34.
We hypothesize that the nuclear localization of yNAP1 may be regulated by altering the accessibility of the nuclear export sequence (as suggested earlier, 16) and NLS. Our structural studies suggest that the oligomeric version of yNAP1 is incapable of interacting with karyopherins, whereas the NLS is highly accessible in the dimeric form of yNAP1. The nuclear localization of NAP1 is regulated through cell-cycle specific phosphorylation of residues located in the immediate vicinity of the NES and NLS in Drosophila and presumably also in other species 36. Although the sites of phosphorylation have not yet been identified in yeast, it is possible that phosphorylation in and around the β hairpin has the potential to disrupt oligomerization, thus regulating interaction with karyopherins and nuclear transport.
Wild type and mutant yeast NAP-1 were expressed and purified as described before 23. For the histone exchange assay, NCPs were reconstituted with CPM labeled DNA or H2B T112C. Histone binding, nucleosome assembly, and histone exchange assays were performed as described 23 11. GST-NAP1 pulldown assays were performed as follows: 0.25 nmoles of GST-tagged wild type or mutant NAP1 was immobilezed on 50μl of Glutathione Sepharose 4B resin. The resin was then mixed with 100 pmoles of refolded H2A-H2B dimer and incubated for 3 hours. Unbounded H2A-H2B was removed by washing three times with HEPES buffer (20mM HEPES, 0.5 mM EDTA, 10% glycerol, 0.05% Nonidet p-40, 5μM ZnSO4, 2.5 mM MgCl2, and 100mM to 450 mM KCl). GST alone was used as a control to test the non-specific binding of H2A-H2B. The results were analyzed by 15% SDS PAGE and stained with coomassie brilliant blue.
Nucleosome assembly was assayed by incubating 2.5 μM of NAP1 (wild type or mutant, respectively) with recombinant Xenopus laevis histone octamer (0.1–0.4 μM) and 0.1 μM of a 146 bp DNA fragment derived from the 5S rRNA gene. The reaction (in 10 mM HEPES at pH 7.5, 100 mM NaCl, 5 mM MgCl2, 0.5 mM EGTA, and 0.2 mM PMSF) was allowed to proceed for 1 hour at 4°C, and the resulting nucleosomes were analyzed by native PAGE and SYBR-Gold staining.
H2A-H2B dimer dissociation from the nucleosome was analyzed by electrophoretic mobility shift assay (EMSA). CPM labeled DNA or CPM-labeled H2A-H2B dimer containing nucleosomes were prepared as described before 11. Fluorescently labeled nucleosomes on the gels were visualized without staining on a transilluminator at 365nm.
Recombinant yNAP-1(74–365) was crystallized at 4°C by sitting drop vapor diffusion. yNAP1(74–365) at 15 mg/ml protein, in 20 mM Tris-HCl pH 7.5, 1 mM dithiothreitol, 1 mM EDTA, 100 mM NaCl was combined in a 1:1 ratio with reservoir solution containing 40% MPD, 0.1M Sodium citrate and 0.2M Ammonium acetate at pH 7.5. Crystals (average size 0.06 × 0.03 × 0.03 mm) were obtained after about 20 days at 4 C. Diffraction data were collected at beamline 5.0.2 at the Advanced Light Source. Data were processed with Denzo, and reduced with SCALEPACK 37. The structure was solved by a molecular replacement 38, using PDB code 2ayu as an initial search model. The model was built with O 39, and refined using CNS 38. All of the images were prepared using PYMOL 40. Structure-superpositions were carried out using LSQMAN 41. The buried surface area for the complex was calculated using AREAIMOL.
Gel filtration chromatography was done on an AKTA FPLC using a Superdex S-200 16/60 (120 mL) column (Amersham Biosciences) as described 17. Following equilibration in either 100 or 500 mM NaCl chromatography buffer, 1 mL of 5mg/ml samples were applied and run at 1 mL/min and 4 C. The elution solution used in the experiments contained 100 or 500 mM NaCl in 20 mM Tris-HCl, pH 7.5, 1 mM dithiothreitol, 1 mM EDTA. For the low pH experiment, 100 mM NaCl in 20 mM Bicine, pH 5.6, 1 mM dithiothreitol, 1 mM EDTA was used. Peak fractions were checked by SDS-PAGE (data not shown).
Sedimentation velocity experiments were performed in a Beckman XL-series analytical ultracentrifuge using the absorbance optical system. 400 μl of sample and 420 μl of reference (buffer) were loaded into two-sector, charcoal-filled Epon centerpieces (12mm pathlength) assembled with quartz windows and centrifuged in an AN60Ti 4-hole rotor at 23°C. All samples were prepared to absorbance values at 230 nm of 0.5, which results in an absorbance signal of 0.6 in the AUC. NAP1 concentrations were determined by absorbance spectrometry using a calculated molar extinction coefficient of 36,100 M-1 cm-1 at 276 nm 42. Velocity data were edited and analyzed using the boundary analysis method of Demeler and van Holde 43 as implemented in UltraScan version 7.1 for MS Windows. This analysis yields the diffusion-corrected, integral distribution of S (G(s)). Sedimentation coefficients (s) are reported in units of Svedbergs (S), where 1 S = 1 × 10−13 s, and are corrected to that of water at 20°C (s20,w). The partial specific volume (v-bar) of NAP1 (0.7174 cm3/g at 20°C) was calculated from the primary amino acid sequence, and solvent densities (ρ) were calculated within UltraScan. For these experiments, we assumed no significant change in the ε276 due to the change in oligomeric state of the proline substitutions mutants. Frictional ratios (f/f0) were calculated using the observed, experimental sedimentation coefficients and the known molecular weight of NAP1 within the simulations menu in UltraScan. f/f0 values account for the geometric symmetry/asymetry of the molecule as well as the frictional drag imparted by unstructured or loosely structured domains in the molecule.
(A) Sequence of the βhairpin. The NLS is shown in bold, residues that were mutated to proline (either individually (R290P or K305P) or cumulatively (R290P, R301P, T302P, and K305P) are indicated by black circles. (B) Purified wild type and mutant proteins were analyzed by 15% SDS PAGE. Lane 1: wild type NAP1(1-417). Lane 2: R290P NAP1(1-417). Lane 3: K305P NAP1(1-417). Lane 4: R290P, R301P, T302P, and K305P NAP1(1-417). Lane 5: wild type NAP1(74-365). Lane 6: K305P NAP1(74-365). Protein size markers (KDa) are 10; 15; 20; 25; 37; 50; 75; and 100 KDa. Gels were stained with coomassie brilliant blue. Since wild type NAP1(1-417) and NAP1(74-365) have extremely low pI 4.2 and 4.5, respectively, in both the cases the bands corresponding to yeast NAP1 proteins run atypically slow on 15% SDS PAGE.
This work was supported by grant from the National Institutes of Health (RGM067777). KL and YJP are supported by the Howard Hughes Medical Institute. The atomic coordinates and the structure factors have been deposited in the Protein Data Bank (accession code 2Z2R).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.