The structure of nucleosome assembly protein 1

Young-Jun Park and Karolin Luger†

Department of Biochemistry and Molecular Biology, Colorado State University, Fort Collins, CO 80523-1870

Edited by Roger D. Kornberg, Stanford University School of Medicine, Stanford, CA, and approved November 29, 2005 (received for review September 14, 2005) Nucleosome assembly protein 1 (NAP-1) is an integral component the assembly function of yNAP-1 is reversible, and that yNAP-1 in the establishment, maintenance, and dynamics of eukaryotic may have a general role in promoting chromatin fluidity. chromatin. It shuttles histones into the nucleus, assembles nucleo- In addition to its nucleosome assembly and histone binding somes, and promotes chromatin fluidity, thereby affecting the activity, NAP-1 is implicated in transcriptional regulation (14–17), transcription of many genes. The 3.0 Å crystal structure of yeast and in the regulation of the cell cycle (18–20). In Saccharomyces NAP-1 reveals a previously uncharacterized fold with implications cerevisiae, a genomewide analysis showed that the expression of for histone binding and shuttling. A long ␣-helix is responsible for Ϸ10% of all genes was affected in a nap-1-deficient yeast strain. In homodimerization via a previously uncharacterized antiparallel mouse, knocking out the neuron-specific NAP-1͞2 gene resulted in non-coiled-coil, and an ␣͞␤ domain is implicated in protein–protein embryonic lethality at the mid-gestation stage (21). Yeast NAP-1 interaction. A nuclear export sequence that is embedded in the has been shown to interact specifically with B type cyclins (clb2), dimerization helix is almost completely masked by an accessory kinase Gin4, and NBP (NAP-1 binding protein) (18–20). This domain that contains several target sites for casein kinase II. The finding suggests that NAP-1 participates in the control of mitotic four-stranded antiparallel ␤-sheet that characterizes the ␣͞␤ do- events (18). main is found in all histone chaperones, despite the absence of Microscopic analyses of fluorescently tagged NAP-1 showed homology in sequence, structural context, or quaternary structure. that although it is located in the nucleus in S phase, it is found To our knowledge, this is the first structure of a member of the in the cytoplasm during the G2 phase of the cell cycle. Cyto- large NAP family of proteins and suggests a mechanism by which plasmic localization is observed in various species (22), suggest- the shuttling of histones to and from the nucleus is regulated. ing that NAP-1 may be involved in shuttling histones between the cytoplasm and the nucleus. This finding is further supported by chromatin ͉ histone chaperone ͉ nuclear transport ͉ x-ray crystallography the finding that NAP-1 interacts with Kap114p, a member of the karyopherin (importin) family of proteins that is responsible for hromatin assembly and disassembly are dynamic biological the nuclear import of H2A and H2B. Export of NAP depends on Cprocesses that increase chromatin fluidity and regulate the a nuclear export signal-like sequence (23). The mechanisms for accessibility of the eukaryotic genome to DNA replication, tran- regulating NAP-1 subcellular localization in normal cycling cells scription, repair, and cell cycle progression. The basic structural unit are currently unknown but may involve - of eukaryotic chromatin is the nucleosome in which 146 bp of DNA dependent regulation (24). are wrapped around a histone octamer that consists of two mole- Here, we report the crystal structure of NAP-1. Our study reveals cules each of the histones H2A, H2B, H3, and H4 (1). Chromatin that NAP-1 exhibits a previously uncharacterized fold consisting of assembly is a stepwise process that starts with the association of a a long ␣-helix that is mainly responsible for dimerization and a ␤ tetramer of histone (H3-H4)2 with the DNA, followed by the -sheet that bears some similarity with other known histone incorporation of H2A-H2B dimers to form the nucleosome. The chaperone proteins. The structure further suggests how nuclear nucleosome assembly process is facilitated by several partially import of histones may be regulated. The results of this study redundant pathways and is aided by histone chaperone proteins, represent a step toward mapping the interactions of yNAP-1 with such as nucleoplasmin, antisilencing factor 1 (Asf1), histone regu- histone and provide important information about the structural lator (HIR), chromatin assembly factor 1 (CAF-1), N1͞N2, and relationship between histone chaperones. nucleosome assembly protein 1 (NAP-1) (reviewed in ref. 2). NAP-1 is conserved among all eukaryotes from yeast to humans Results (3–6), and, in addition to its well characterized role in chromatin yNAP-1 Exhibits a Previously Uncharacterized Fold. The crystal struc- assembly, has multiple and ill defined functions in vivo. Yeast ture of NAP-1 from S. cerevisiae (yNAP-1) was determined at 3.0 NAP-1 (perhaps the best-characterized of all members of the Å resolution (Table 1, which is published as supporting information NAP-1 family) is a 48-kDa polypeptide that binds H2A-H2B and on the PNAS web site). Of a total of 417 aa, well defined electron H3-H4 and mediates nucleosome assembly in vitro (3, 7). Structural density was observed for the central region encompassing residues and functional analyses of the central domain of yeast NAP-1 70–370 (Fig. 1 A and B), whereas the N- and C- terminal regions (residues 74–365) have shown that this region retains a native-like were largely disordered. It has been shown earlier that residues structure and functions normally in nucleosome assembly (7, 8). 74–365 are sufficient for histone binding and nucleosome assembly. Sedimentation equilibrium experiments revealed that yNAP-1 ex- Additionally, limited has demonstrated that this is the ists as a stable dimer and self-associated oligomers in solution; most stable structural entity (7, 8). An N-terminally truncated monomers are never observed in the absence of denaturing agents version of the protein (encompassing amino acids 74–417) crys- (9, 10). tallized in the same space group, indicating that the N-terminal In HeLa cells, NAP-1 was shown to interact with newly synthe- sized histones H2A and H2B, suggesting a role in de novo chromatin assembly (11). NAP-1 was also identified as a component of the Conflict of interest statement: No conflicts declared. histone variant H2AZ-specific exchange complex SWR1 in yeast This paper was submitted directly (Track II) to the PNAS office. (12). Additionally, NAP-1 has the ability to remove H2A-H2B Freely available online through the PNAS open access option. dimers from nucleosomes and to replace them with either major- Abbreviations: NES, nuclear export sequence; NLS, nuclear localization sequence. type or variant H2A-H2B dimers in vitro (13). Thus, NAP-1 is Data deposition: The atomic coordinates and structure factors have been deposited in the involved in both replication-coupled and replication-independent Protein Data Bank, www.pdb.org (PDB ID code 2AYU). chromatin assembly. Nucleosome sliding occurs as a consequence †To whom correspondence should be addressed. E-mail: [email protected]. of dynamic histone dimer exchange (13). These results suggest that © 2006 by The National Academy of Sciences of the USA

1248–1253 ͉ PNAS ͉ January 31, 2006 ͉ vol. 103 ͉ no. 5 www.pnas.org͞cgi͞doi͞10.1073͞pnas.0508002103 Downloaded by guest on September 30, 2021 Fig. 1. Overall structure of yeast NAP-1. (A) sequences alignment of members of the NAP protein family. yNAP1, xNAP1, dNAP1, and hNAP1 are shown. y, S. cerevisiae;x,Xenopus laevis;d,Drosophila melanogaster;h,Homo sapiens nucleosome assembly protein 1. Red or blue color indicates identical or conserved amino acid residues from all of the NAP family proteins. PKE*P**K*EE (* is any amino acid) repeat sequences of NAP-1L3 are not shown in this alignment as indicated with parentheses. The alignment was generated with CLUSTALW. Subdomains A, B, C, and D are boxed in blue, yellow, green, and red, respectively. Secondary structure elements are indicated. Intervals of 10 aa for S. cerevisiae (black circles) are indicated. The two highly conserved hydrophobic sequence motifs 185IPSFWLT191 and 311SFFNFF316 are indicated with dotted lines. (B) The monomer structure of yeast NAP-1, shown as a ribbon diagram. Subdomains A (domain I), B (domain I), C (domain II), and D (domain II) are shown in blue, yellow, green, and red, respectively, as in A.(C) Topology diagram of the yeast NAP-1 monomer (helices are shown as circles, strands are shown as triangles, and loops are shown as pipes). Domains I and II are indicated.

region is not involved in crystal contacts. We have also determined and N-terminal regions (see above), residues 172–180 in the ␣3–␣4 the structure of a NAP-1 construct consisting only of the core loop and 293–301 in the ␤5–␤6 loop were disordered in our electron region (74–365) and, with one minor exception, found no differ- density maps and are not included in the model. ences when compared with the structure obtained with full-length In the crystal structure of full-length NAP-1, one NAP-1 mono- protein (see below). These observations suggest that the N- and mer represents the asymmetric unit (ASU), and the crystallo- C-terminal segments are normally disordered and do not contribute graphic twofold symmetry generates a homodimer with an exten- to the overall architecture of NAP-1, in agreement with in vitro sive interaction interface (Fig. 2). The dimer has an overall solution studies (7). ellipsoidal shape and approximate dimensions of 75 ϫ 57 ϫ 32 Å. Fig. 1B shows an overview of the yNAP-1 monomer; the corre- It consists of three layers: the first layer is defined by the dimer- sponding connectivity diagram is shown in Fig. 1C. The monomer ization helix ␣2 (domain I, shown in blue in Fig. 2), the second is can be divided into two domains: domain I (the dimerization formed by the ␤-sheets of domain II (green), and the bottom layer domain) consists of one long ␣-helix that spans 47 aa, flanked by is formed by ␣-helices (shown in red in Fig. 2) that span its entire two shorter ␣-helices. The main feature of domain II (the protein width. Truncated yNAP-1 (residues 74–365) crystallized in P42212 interaction domain) is a four-stranded antiparallel ␤-sheet that is and contains one NAP-1 dimer per ASU (data not shown). Except protected by ␣-helices at its underside. A small antiparallel ␤-hair- for ␤5, ␤6, and two of the flexible loops (which are involved in pin that functions as a nuclear localization sequence (NLS) extends different crystal contacts in the two lattices), the structures of the from it (34, 35). Each of these domains can be divided into two NAP-1 molecules in the noncrystallographic and crystallographic subdomains. Subdomain A (shown in blue) is mainly responsible for dimer are identical and superimpose with an rms deviation of 1.05 dimerization, whereas subdomain B (yellow) is implicated in reg- Å for 268 residues. ulating access to a NES embedded in subdomain A. Subdomain C (green) forms an amphipathic ␤-sheet that is protected at its NAP-1 Dimerizes via a Non-Coiled-Coil Motif. Domain I is mainly BIOPHYSICS underside by subdomain D (red). In addition to the disordered C- responsible for the dimerization of NAP-1 (Fig. 2) by close anti-

Park and Luger PNAS ͉ January 31, 2006 ͉ vol. 103 ͉ no. 5 ͉ 1249 Downloaded by guest on September 30, 2021 Fig. 2. The dimer structure of the yeast NAP-1. The two monomers are Fig. 3. The dimerization domain. (A) Top and side views of subdomain A related by a crystallographic twofold axis. A, top; B, bottom; C and D, two side show detailed interactions in the dimerization interface of yNAP-1. Val-109 views. (A and B) A view is obtained by a 180° rotation around the vertical axis and Pro-130 are indicated. The arrow denotes the location of Pro-130. (B) (as indicated). (C and D) The side views of the NAP-1 are obtained by rotation Stereo diagram of the final 2Fo-Fc electron density map (contoured at 2 sigma) of 90° around the axis of crystallographic symmetry. The unstructured region of residues 112–119 of subdomain A and 109–126 of subdomain AЈ. Dotted of ␣3–␣4 loop or acidic C-terminal domain that is not a part of this model is lines indicates hydrogen bonds. indicated with a yellow or red dotted line.

The Nuclear Export Sequence (NES) Is Masked by the Accessory parallel pairing of the long ␣2-helices along their entire length of 47 Domain. Residues 88–102 of yNAP-1 have been identified as a NES, aa (Ϸ69 Å). The arrangement is unusual in that the two helical axes and this activity has been confirmed for yeast NAP-1 in vivo (23, are aligned in an antiparallel ‘‘tram-track’’ fashion, rather than 34). Replacement of the hydrophobic residues Leu-96 and Leu-102 coiling around each other as in the widespread coiled-coil motif. that are characteristic of NES sequences to alanine caused a radical ␣ redistribution of the protein from the cytoplasm to the nucleus (23). The 2-helices are curved in both planes, with the helical axes ␣ ␣ following a sigmoidal path in one plane and forming a dome-shaped The NES of yNAP-1 initiates in the loop connecting 1 with 2 and includes the N-terminal 11 aa of ␣2, which are also involved in architecture in the other (Fig. 3A). More precisely, a Ϸ30° kink is forming the dimerization interface. The ␣-helical character and the introduced into ␣2 by the presence of a at position 130; a Ϸ distribution of hydrophobic and hydrophilic residues in the present second, 25° kink is observed at Val-109 (Fig. 3A). As a conse- structure have also been found in other published structures quence of the antiparallel pairing of the two helices, Pro-130 containing a functional NES (38). Ј juxtaposes Val-109 of the second NAP-1 molecule within the In yNAP-1, the NES is for the most part occluded by subdomain dimer. It is likely that the bend at Val-109 is caused by the need to B (residues 141–180; Figs. 2A and 4A). We have termed this maintain packing by keeping the two helical axes parallel. Fig. 1A domain the accessory domain because of variability in length demonstrates the complete amino acid conservation at these two between different NAP forms (Fig. 1A). In yeast, it consists of an positions within the entire NAP-1 family of proteins. The dimer is ␣-helix (␣3) and two loops and effectively links domain I to domain further stabilized by the ␣2–␣3 loop, the ␣3-helix, and the ␣3–␣4 II. By wrapping an extended ␣2–␣3 loop around the base of ␣2Ј of loop that wrap around the base of the ␣2-helix of the dimerization the dimerization partner, and by packing ␣3 next to ␣2Ј, it not only partner (Fig. 2 A and B). In addition, ␣1 has a key role in orienting contributes to dimer formation but also effectively covers the upper the C-terminal acidic domain of the dimerization partner by part of the NES (Fig. 4). A conserved proline (Pro-145) in the forming extensive hydrophobic interactions with ␣8Ј (Fig. 2A). ␣2–␣3 loop restricts its conformation and positions it over ␣2. The The dimer interface is characterized by exclusively hydrophobic loop connecting ␣3 and ␣4 (amino acids 161–180 in yNAP-1) is interactions over the entire length of the involved ␣-helices (see, for highly variable among proteins in the NAP family (Fig. 1A). In the example, Fig. 3B). It buries Ϸ22% of the overall surface of each present structure, it is rather flexible and characterized by poor subunit and involves 20 residues within ␣2 and four in ␣1. The electron density. Residues 172–180 were too disordered to be hydrophobic nature of these residues is overall maintained between included in the model, although weak electron density was observed humans, frog, flies, and yeast (Fig. 1A), suggesting that dimer that covers the N-terminal segment of the NES (Fig. 4B). formation is strictly conserved. Indeed, the majority of conserved ␣ Domain II Is Stabilized by an Extensive Network of Aromatic Residues. residues are clustered along the dimerization interface of 2 (Fig. Domain II spans residues 181–370 of yNAP-1 and consists of two 6A, which is published as supporting information on the PNAS web layers (subdomains C and D, shown in green and red, respectively, site). The extensive dimerization interface that spans the entire in Figs. 1 and 2). Subdomain C is a four-stranded antiparallel dimer diagonally (Fig. 6B) is consistent with a previous investigation ␤-sheet. A single bond between Cys-249 and Cys-272 showing that the NAP-1 dimer cannot be disrupted without dena- covalently links ␤3 and ␤4. Because the in ␤4 is only found turation and that this disruption is coincident with a global disrup- in the yeast homolog but not in any of the other NAP proteins, this tion in 2° structure (9). Domain II projects away from each side of disulfide link is not essential for maintaining the stability of the the dimer and is therefore not involved in dimer stabilization, with protein. The ␤-sheet has a pronounced amphiphilic character. The the exception of ␣8. solvent-exposed side is relatively hydrophilic with 53% of all

1250 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0508002103 Park and Luger Downloaded by guest on September 30, 2021 formed by ␤5 and ␤6, the 311SFFNFF316 motif. Ten aromatic residues, including those in the SFFNFF motif, are arranged in an unusual edge-to-face manner, forming an aromatic core that an- chors the ␤-sheet to the underlying ␣-helix (Fig. 7A). This extensive hydrophobic core confers a high degree of stability to the ␤-sheet and explains why truncated versions of yNAP-1 missing subdomain D are unable to fold upon expression (7) and are thus inactive (23). The second stretch of hydrophobic amino acids that is 100% conserved among the NAP family of proteins is the 185IPSF- WLT191 motif that is also located in domain II (Fig. 1A). Like the SFFNFF motif, its primary function appears to be the formation of a hydrophobic core, effectively connecting ␣2, ␣4, ␣7, and ␤1 (Fig. 7B). A database search revealed no structural homologues for the entire yNAP-1 structure. However, several proteins with antipar- allel ␤-sheets similar to subdomain C of NAP-1 were identified. Intriguingly, one of the closest matches to subdomain C was ASF1, despite the absence of amino acid homology. Subdomain C also exhibits some similarity to other histone chaperones, such as nucleoplasmin, ASF-1, and CAF-1 (Fig. 8, which is published as supporting information on the PNAS web site). Like yNAP-1, these proteins are implicated in histone binding and nucleosome assem- bly. The antiparallel ␤-sheet domain of the NAP-1 structure and the currently available structures of histone chaperones (or the homol- ogy model of dCAF1 p55) are compared in Table 2, which is Fig. 4. NES masking by the accessory domain. The accessory domain and NES published as supporting information on the PNAS web site. region are shown in yellow and blue, respectively. The color code for the other Residues 290–295 of yeast NAP have been identified as a NLS subdomains is as in Fig. 1. (A) Molecular surface of the molecule, shown in the ␤ same view as in Fig. 2A.(B) Detailed view of the masking of the NES by (35). These residues are part of a short antiparallel -sheet formed subdomain B, shown in two orientations. The region between amino acids 169 by ␤5 and ␤6 that protrudes from the main structure (Figs. 1 and and 181 is missing from the model; these amino acids are circled in red. 2). This region is highly basic (theoretical pI 10) as is characteristic for nuclear localization sequences.

residues being charged. In contrast, 87% of the amino acids that line yNAP-1 Exhibits a Distinct Charge Distribution. The dome-shaped the other side of the ␤-sheet are hydrophobic. This hydrophobic yNAP-1 dimer exhibits an uneven charge distribution: The convex surface is spanned in its entirety by ␣6 of subdomain D (Fig. 1B), surface of the NAP-1 dimer, mainly defined by the dimerization whose residues interact extensively with residues from all four helix (subdomain A) and the upper surface of the ␤-sheet and the ␤-strands (Fig. 7A, which is published as supporting information on ␤1–␤2 loop (subdomain C; Fig. 2B), exhibits a relatively nondistinct the PNAS web site). The hydrophobic core is reinforced by a highly charge distribution. The exception is the subdomain encompassing conserved sequence motif that folds back from the extended hairpin the NLS (␤5 and ␤6; Fig. 5A) (see above). In contrast, the concave

Fig. 5. Electrostatic potential distribution of NAP-1 dimer. (A–C) The scale of the surface potential (in KBT, where KB is Boltzmann constant) is shown at the top, with the blue color representing strong positive charge and the red color representing strong negative charge. The electrostatic potential on the underside of the yNAP-1 dimer is significantly negatively charged. Glu-370, the beginning of the acidic CTD, is indicated. (D) The salt-induced dissociation of NAP-1 with H2A- H2B dimer complex. Ten micromolar NAP-1 dimer was incubated with 10 ␮M fluorescently labeled [H2A-H2B CPM (CPM, 7-diethylamino-3-(4Ј-maleimidylphenyl)- 4-methylcoumarin)] dimer. Under these conditions, no free NAP-1 is observed (data not shown). was excited at 300 nm, and fluorescence of the com- plex was monitored. NAP-1 (1–417) with fluorescently labeled (H2A-H2B CPM) dimer complex (red square) or NAP-1 (74–365) with fluorescently labeled (H2A-H2B CPM) dimer complex (blue circle) was dissociated by increasing the amount of salt concentration. Ratios of fluorescence intensities (470 nm͞335 nm, acceptor͞ donor) are calculated, and then normalized data are plotted. All three independent measurements are plotted. The midpoints for the loss of FRET between NAP-1 and the (H2A-H2B) dimer are 494 mM (Ϯ8 mM) NaCl for full length NAP-1 and 414 mM (Ϯ11 mM) NaCl BIOPHYSICS for N- and C-terminally truncated NAP-1 (residue 74–365).

Park and Luger PNAS ͉ January 31, 2006 ͉ vol. 103 ͉ no. 5 ͉ 1251 Downloaded by guest on September 30, 2021 underside of the yNAP-1 dimer is highly acidic (Fig. 5B). Acidic mental evidence that this domain is actually involved in histone residues from ␣6 and from the loop connecting ␤3 and ␤4 delineate binding; rather, it seems to provide the scaffold onto which histone a cavity that is defined by conserved acidic residues from ␣2. This binding elements are grafted. cavity (Fig. 5C) is too small to accommodate an ␣-helix, let alone The dimerization motif found in yNAP-1 is completely different an entire histone dimer; however, it may be involved in neutralizing from the commonly found coiled-coil or helix bundle dimerization and binding the basic N-terminal histone tails. Histone tail binding motif. The coiled-coil is formed by component helices coming has been shown to contribute to the histone–yNAP-1 interaction. together to bury hydrophobic residues as they twist around each For example, the tails appear to be necessary to distinguish between helix (41). In contrast, the two long ␣-helices in yNAP-1 promote H3-H4 and H2A-H2B (7). The acidic character of this region is dimerization by an antiparallel, tram-track-like pairing, with hydro- enhanced by the presence of the acidic C-terminal domain that phobic residues interdigitating every helical turn. Coiled-coil for- extends from this region but that is too disordered in the current mation is likely prohibited by the strong bends in the dimerization structure to be included in the model. Although it is not essential helix (Fig. 3A). The large buried surface area and highly hydro- for histone binding, this increase in acidic residues contributes to the phobic nature of the amino acids between the two paired helices is stability of the NAP-1–histone complex. We show by fluorescence consistent with the finding that yNAP-1 exists as a dimer or quenching that the yNAP-1–H2A͞H2B complex is significantly multimer of dimers in solution (9). Sequence alignment suggests stabilized by the presence of the N- and C-terminal domain (Fig. that the dimerization motif is a highly conserved feature of all 5D), confirming the notion that histone binding utilizes the under- members of the NAP-1 family of proteins. For example, the side of the yNAP-1 dimer for interaction with histones. N-terminal region (residue 25–66; Fig. 1C) of TAF1͞SET was predicted to be important for dimerization via a coiled-coil inter- Discussion face, and dimerization was shown to be important for TAF1͞SET The structure of homodimeric yNAP-1 reveals a previously un- activity (42). characterized fold that is characterized by a long, non-coiled coil of Members of the NAP family have been shown to bind to a variety two ␣-helices forming an extensive dimerization interface and a of proteins involved in transcription, cell cycle regulation, protein slightly offset ␣͞␤ domain that is implicated in the interaction with shuttling, chromatin assembly, histone variant exchange, and chro- histones and other cellular proteins. An exhaustive search of all matin remodeling function (12, 16, 34, 43, 44). Drosophila NAP-1 available structures in the protein database identified only frag- coprecipitates with core histones H2A-H2B from embryos (5). ments of unrelated proteins containing ␤-strands or short helix Similarly, human NAP-1 coprecipitates with H2A from HeLa bundles. The overall structure of NAP-1 is unrelated to any cytoplasmic extract (11). Recombinant yeast NAP-1 binds to linker deposited structure, suggesting that NAP-1 represents a previously H1 and high mobility group (HMG) proteins (45). Several other uncharacterized fold. proteins, for example chromatin-remodeling factor SWR1 (12), Because of the pronounced curvature of the dimerization helices, importin kap114p (34), casein kinase 2 CK2 (46), transcription the structure assumes a dome-shaped architecture with an extended activator p300 (16), and cyclin (43), also have a functional interac- acidic region on its underside. This is the region from which the tion with NAP family members in a variety of systems. It appears disordered C-terminal acidic domain extends (indicated by the that the histone binding region of NAP-1 is spatially distinct from dotted line in Fig. 2 C and D). Although not essential for histone the region that interacts with other cellular factors. For example, the binding or chromatin assembly, we find that the acidic domain ability of NAP-1 to bind histones H2A-H2B was not compromised contributes to the stability of the yNAP-1–histone complex. We in the presence of p300 (16) or Kap114 (34). have shown previously that the C-terminal region is essential for NAP-1 participates in the active transport of histones from the yNAP-1 to promote reversible histone H2A-H2B dimer removal on cytoplasm to the nucleus by forming a ternary complex with a nucleosomal template, thereby promoting nucleosome sliding histones and karyopherins (47). The previously identified NLS (35) (13). A recent report suggested that the C-terminal region of NAP is bound specifically by Kap114p from yeast, a karyopherin with in HeLa cells exhibits a high degree of polymorphism due to multiple cargo binding domains. Our result that ␤5 (encompassing covalent posttranslational modifications by polyglutamylation. the NLS-forming residues) and ␤6 form a short ␤-sheet that is Roughly 50% of all NAP-1 molecules are glutamylated in HeLa slightly offset from the main ␤-sheet (see Fig. 1C) is consistent with cells. In contrast, only Ϸ4% of (the only other protein the in vivo finding that histones and karyopherin can be bound by known to be glutamylated) is found in glutamylated form in the yNAP-1 simultaneously. same cell (39). Two glutamylation sites (modified with up to 10 The majority of yNAP-1 is present in the cytoplasm because of glutamyl units) were identified in the C-terminal region of NAP-1 continuous NES-dependent export from the nucleus (23, 34). In the and NAP-2 in HeLa cells. This modification may lead to the present structure, most of the NES consensus sequence contained formation of a second or third acidic C-terminal moiety (39). within ␣2 is shielded from solvent by the accessory domain, whereas Accumulation of glutamyl moieties would allow for modulation of the ␣1–␣2 loop region is partially exposed. The ␣2–␣3 loop and the total negative charge of the C-terminal acidic domain, in a ␣3–␣4 loop have relatively high temperature factors compared with reversible manner, with the potential of switching between different the remainder of the accessory domain, indicating that these loop NAP functions. The presence of glutamylated NAP-1 is not re- regions are quite flexible and may be able to detach from the NES. stricted to HeLa cells but is also observed in different tissues and It is possible that the sequence and length variations in subdomain in different organisms. Interestingly, this modification seems to B may be related to differences in regulation of nuclear localization. occur in other acidic histone chaperones as well; for example, NES masking as a mechanism to regulate nuclear occupancy has nucleoplasmin in Xenopus (40). Together, these data implicate the been demonstrated for p53 and STAT-1 (48, 49). Inhibition of the underside of the ␤-sheet (subdomain C) as a histone interaction exportin Crm-1 (chromosome maintenance region 1) leads to the surface. accumulation of NAP-1 in the nucleus, suggesting that Crm-1 might The antiparallel ␤-sheet in subdomain C of yNAP-1 appears to be one of the nuclear exporters of yeast NAP-1 (34). be a recurring theme in histone chaperones, although the structural In many cases, nuclear transport is regulated by phosphorylation, motif is embedded within different architectures in the various as has been observed for p53 (48), STAT-1 (49), and Pho4 (50). In chaperones (Fig. 8). This structural similarity, despite the absence Drosophila, CKII phosphorylates NAP-1 (46), and there are con- of overall sequence homology and differences in tertiary and served sites among mammalian NAPs (21). NAP-2 under- quaternary structure, may be based on the conservation of function, goes cell-cycle-dependent changes in the phosphorylation state, and and the variation in overall sequence may allow for distinct roles in these sites are located in ␣3 and the ␣3–␣4 loop of the subdomain chromatin assembly. With one exception (51), there is no experi- B. These results support the hypothesis that histone chaperone

1252 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0508002103 Park and Luger Downloaded by guest on September 30, 2021 localization may be regulated by CKII-mediated phosphorylation was combined in a 1:1 ratio with reservoir solution containing (24). Putative phosphorylation sites in yNAP-1 were identified at 0.25 M mono-ammonium dihydrogen phosphate. Crystals were positions 140, 159, and 177, all in subdomain C. Phosphorylation of cryoprotected with 25% of glycerol and flash cooled in liquid these serine residues has the potential to alter the interaction of this nitrogen. Diffraction data were collected at beamlines 5.0.2 and subdomain with the NES, thus regulating NES accessibility and 4.2.2 of the Advanced Light Source (Lawrence Berkeley Na- subcellular trafficking of NAP-1. tional Laboratory). Data were processed with DENZO and re- No structure for any member of the NAP family has been duced with SCALEPACK (25). described to date; thus, the structure presented here represents The positions of heavy atoms were determined by using SOLVE, a paradigm for this class of proteins. The topology found for and RESOLVE was used for density modification (26). The model yeast NAP-1 represents a previously uncharacterized fold with was built with the program O (27). The structure model was refined unique features, although one subdomain exhibits similarities to against the native data set by using CNS (28). Most of the images other known histone chaperones. In particular, our structure were prepared by using PYMOL (29), expect for the electrostatic suggests a mechanism by which nuclear localization may be surface representation, which was created with GRASP (30). Struc- regulated. The transport of canonical or variant histones into or ture-superpositions were carried out by using LSQMAN (31). The out of the nucleus is an essential step in the activation of accessible surface area for each of the individual proteins and the chromatin assembly and the regulation of the gene expression in surface of the complex were calculated by using AREAIMOL (32). response to a cellular signal. The masking and unmasking of The area of interaction per monomer (buried surface area) was nuclear export and import signals may be a general mechanism calculated by ((Area of monomer ϩ Area of monomer) Ϫ Area of for regulating subcellular localization of NAP-1 and histones. dimer)͞2. Posttranslational modifications and masking͞unmasking of spe- cific signal sequences responsible for nuclear import and export Fluorescence Measurements. To test binding of the H2A-H2B dimer may be important for the coordinated control of the nucleocy- to NAP1, we measured FRET from intrinsic chromophores of toplasmic transport. yNAP1 (tryptophan) to an extrinsic chromophore of H2A-H2B Materials and Methods (CPM attached at H2B T112C). Both full-length and N- and C-terminally truncated NAP-1 have four . There are no Protein Expression and Purification. Full-length and truncated ver- ⌬ ⌬ tryptophan residues in the histone H2A-H2B dimer. Detailed sions of yNAP-1 (yNAP-1 N and yNAP-1 NC) were expressed methods of FRET experiments are described in a pervious study and purified as described in ref. 7. Selenomethionine (SeMet) was (13, 33). All samples were measured at a final concentration of 4 introduced by overexpression in minimal media with SeMet as the ␮M for the complex in Tris buffer (20 mM Tris⅐HCl,pH7.5͞1mM only source. Mutants of yeast NAP-1 in which selected EDTA͞1 mM DTT) at 20°C. To analyze the dissociation of the residues (Leu-102, Leu-218, and Leu-340) were replaced by me- complex, the ratio of donor emission and acceptor emission inten- thionine were prepared to improve phases. Activity of mutant sity was monitored with increasing NaCl concentration. NAP-1 proteins was verified by monitoring complex formation between the H2A-H2B dimer and NAP-1 using native gel electro- We thank A. Kikuchi for the yeast NAP-1 expression plasmid; S. phoresis (data not shown). McBryant, B. Eutzy, and D. Williams for assistance in the initial stages in this project; S. McBryant, J. Chodaparambil, and S. Chakravarthy for Structure Determination. Recombinant yeast NAP-1 was crystal- comments on the manuscript; and J. Nix for help with MAD data lized at 4°C by hanging-drop vapor diffusion. Protein (15 mg͞ml collection. This work was supported by National Institutes of Health protein, in 20 mM Tris⅐HCl, pH 7.5͞1mMDTT͞100 mM NaCl) Grant RGM067777A.

1. Luger, K., Maeder, A. W., Richmond, R. K., Sargent, D. F. & Richmond, T. J. (1997) Nature 26. Terwilliger, T. (2004) J. Synchrotron Radiat. 11, 49–52. 389, 251–259. 27. Jones, T. A., Zou, J. Y., Cowan, S. W. & Kjelgaard, M. (1991) Acta Crystallogr. A 47, 110–119. 2. Krude, T. (1999) Eur. J. Biochem. 263, 1–5. 28. Brunger, A. T., Adams, P. D., Clore, G. M., DeLano, W. L., Gros, P., Grosse-Kunstleve, R. W., 3. Ishimi, Y. & Kikuchi, A. (1991) J. Biol. Chem. 266, 7025–7029. Jiang, J. S., Kuszewski, J., Nilges, M., Pannu, N. S., et al. (1998) Acta Crystallogr. D 54, 905–921. 4. Ishimi, Y., Yasuda, H., Hirosumi, J., Hanaoka, F. & Yamada, M. (1983) J. Biochem. (Tokyo) 29. DeLano, W. L. (2002) The PyMOL User’s Manual (DeLano Scientific, San Carlos, CA). 94, 735–744. 30. Nicholls, A., Sharp, K. A. & Honig, B. (1991) Proteins 11, 281–296. 5. Ito, T., Bulger, M., Kobayashi, R. & Kadonaga, J. T. (1996) Mol. Cell. Biol. 16, 3112–3124. 31. Kleywegt, G. J. (1996) Acta Crystallogr. D 52, 842–857. 6. Steer, W. M., Abu-Daya, A., Brickwood, S. J., Mumford, K. L., Jordanaires, N., Mitchell, 32. Lee, B. & Richards, F. M. (1971) J. Mol. Biol. 55, 379–400. J., Robinson, C., Thorne, A. W. & Guille, M. J. (2003) Mech. Dev. 120, 1045–1057. 33. Park, Y. J., Dyer, P. N., Tremethick, D. J. & Luger, K. (2004) J. Biol. Chem. 279, 24274–24282. 7. McBryant, S. J., Park, Y. J., Abernathy, S. M., Laybourn, P. J., Nyborg, J. K. & Luger, K. 34. Mosammaparast, N., Ewart, C. S. & Pemberton, L. F. (2002) EMBO J. 21, 6527–6538. (2003) J. Biol. Chem. 278, 44574–44583. 35. Mosammaparast, N., Del Rosario, B. C. & Pemberton, L. F. (2005) Mol. Cell. Biol. 25, 8. Fujii-Nakata, T., Ishimi, Y., Okuda, A. & Kikuchi, A. (1992) J. Biol. Chem. 267, 20980–20986. 1764–1778. 9. McBryant, S. J. & Peersen, O. B. (2004) Biochemistry 43, 10592–10599. 36. Bahadur, R. P., Chakrabarti, P., Rodier, F. & Janin, J. (2004) J. Mol. Biol. 336, 943–955. 10. Toth, K. F., Mazurkiewicz, J. & Rippe, K. (2005) J. Biol. Chem. 280, 15690–15699. 37. Janin, J. (1997) Nat. Struct. Biol. 4, 973–974. 11. Chang, L., Loranger, S. S., Mizzen, C., Ernst, S. G., Allis, C. D. & Annunziato, A. T. (1997) 38. la Cour, T., Kiemer, L., Molgaard, A., Gupta, R., Skriver, K. & Brunak, S. (2004) Protein 36, Biochemistry 469–480. Eng. Des. Sel. 17, 527–536. 12. Mizuguchi, G., Shen, X., Landry, J., Wu, W. H., Sen, S. & Wu, C. (2004) Science 303, 343–348. 39. Regnard, C., Desbruyeres, E., Huet, J. C., Beauvallet, C., Pernollet, J. C. & Edde, B. (2000) 13. Park, Y. J., Chodaparambil, J. V., Bao, Y., McBryant, S. J. & Luger, K. (2005) J. Biol. Chem. J. Biol. Chem. 275, 15969–15976. 280, 1817–1825. 40. Regnard, C., Fesquet, D., Janke, C., Boucher, D., Desbruyeres, E., Koulakoff, A., Insina, 14. Kawase, H., Okuwaki, M., Miyaji, M., Ohba, R., Handa, H., Ishimi, Y., Fujii-Nakata, T., C., Travo, P. & Edde, B. (2003) J. Cell Sci. 116, 4181–4190. Kikuchi, A. & Nagata, K. (1996) Genes Cells 1, 1045–1056. 41. Burkhard, P., Stetefeld, J. & Strelkov, S. V. (2001) Trends Cell Biol. 11, 82–88. 15. Ito, T., Ikehara, T., Nakagawa, T., Kraus, W. L. & Muramatsu, M. (2000) Genes Dev. 14, 290, 1899–1907. 42. Miyaji-Yamaguchi, M., Okuwaki, M. & Nagata, K. (1999) J. Mol. Biol. 547–557. 16. Shikama, N., Chan, H. M., Krstic-Demonacos, M., Smith, L., Lee, C. W., Cairns, W. & La 43. Kellogg, D. R., Kikuchi, A., Fujii-Nakata, T., Turck, C. W. & Murray, A. W. (1995) J. Cell Thangue, N. B. (2000) Mol. Cell. Biol. 20, 8933–8943. Biol. 130, 661–673. 17. Chen, H., Li, B. & Workman, J. L. (1994) EMBO J. 13, 380–390. 44. Robinson, K. M. & Schultz, M. C. (2003) Mol. Cell. Biol. 23, 7937–7946. 18. Kellogg, D. R. & Murray, A. W. (1995) J. Cell Biol. 130, 675–685. 45. McQuibban, G. A., Commisso-Cappelli, C. N. & Lewis, P. N. (1998) Biol. Chem. 273, 6582–6590. 19. Altman, R. & Kellogg, D. (1997) J. Cell Biol. 138, 119–130. 46. Li, M., Strand, D., Krehan, A., Pyerin, W., Heid, H., Neumann, B. & Mechler, B. M. (1999) 20. Shimizu, Y., Akashi, T., Okuda, A., Kikuchi, A. & Fukui, K. (2000) Gene 246, 395–404. J. Mol. Biol. 293, 1067–1084. 21. Rodriguez, P., Munroe, D., Prawitt, D., Chu, L. L., Bric, E., Kim, J., Reid, L. H., Davies, 47. Mosammaparast, N., Jackson, K. R., Guo, Y., Brame, C. J., Shabanowitz, J., Hunt, D. F. & C., Nakagama, H., Loebbert, R., et al. (1997) Genomics 44, 253–265. Pemberton, L. F. (2001) J. Cell Biol. 153, 251–262. 22. Marheineke, K. & Krude, T. (1998) J. Biol. Chem. 273, 15279–15286. 48. Zhang, Y. & Xiong, Y. (2001) Science 292, 1910–1915. 23. Miyaji-Yamaguchi, M., Kato, K., Nakano, R., Akashi, T., Kikuchi, A. & Nagata, K. (2003) 49. Mowen, K. & David, M. (2000) Mol. Cell. Biol. 20, 7273–7281. Mol. Cell. Biol. 23, 6672–6684. 50. Kaffman, A., Rank, N. M., O’Neill, E. M., Huang, L. S. & O’Shea, E. K. (1998) Nature 396, 24. Rodriguez, P., Pelletier, J., Price, G. B. & Zannis-Hadjopoulos, M. (2000) J. Mol. Biol. 298, 482–486. 225–238. 51. Mousson, F., Lautrette, A., Thuret, J. Y., Agez, M., Courbeyrette, R., Amigues, B., Becker, 25. Otwinowski, Z. & Minor, W. (1997) Processing of X-Ray Diffraction Data Collected in E., Neumann, J. M., Guerois, R., Mann, C. & Ochsenbein, F. (2005) Proc. Natl. Acad. Sci. BIOPHYSICS Oscillation Mode (Academic, New York). USA 102, 5975–5980.

Park and Luger PNAS ͉ January 31, 2006 ͉ vol. 103 ͉ no. 5 ͉ 1253 Downloaded by guest on September 30, 2021