Implications for Proteasome Nuclear Localization Revealed by the Structure of the Nuclear Proteasome Tether Protein Cut8

Home , Helix bundle, Trichoplax

Implications for proteasome nuclear localization revealed by the structure of the nuclear proteasome tether protein Cut8

Kojiro Takedaa, Nam K. Tonthatb, Tiffany Gloverc, Weijun Xuc, Eugene V. Koonind, Mitsuhiro Yanagidaa, and Maria A. Schumacherb,1

aG0 Cell Unit; Okinawa Institute of Science and Technology (OIST); 1919-1 Tancha, Onna, Okinawa, Japan; bDepartment of Biochemistry, Duke University Medical Center, Room 243A Nanaline H. Duke Building, Box 3711, Durham, NC 27710; cDepartment of Biochemistry and Molecular Biology, University of Texas M. D. Anderson Cancer Center, Unit 1000, Houston, TX 77030; and dNational Center for Biotechnology Information, National Library of Medicine, National Institutes of Health, Bethesda, MD 20892

Edited by Axel T. Brunger, Stanford University, Stanford, CA, and approved August 29, 2011 (received for review March 7, 2011)

Degradation of nuclear proteins by the 26S proteasome is essential ultimately found to be a delay in the destruction of the mitotic for cell viability. In yeast, the nuclear envelope protein Cut8 med- cyclin and securin (13). However, the polyubiquitination of these iates nuclear proteasomal sequestration by an uncharacterized proteins was normal, suggesting a role for the proteasome. Sub- mechanism. Here we describe structures of Schizosaccharomyces sequent studies showed that Cut8 is responsible for sequestering pombe Cut8, which shows that it contains a unique, modular the 26S proteasome to the nucleus (10, 11). Cut8 was found to be fold composed of an extended N-terminal, lysine-rich segment localized to the nuclear envelope and overlapping the location that when ubiquitinated binds the proteasome, a dimer domain of the proteasome (17–21). An interaction between Cut8 and followed by a six-helix bundle connected to a flexible C tail. The the proteasome was subsequently demonstrated (11). Additional Cut8 six-helix bundle shows structural similarity to 14-3-3 phos- data showing that Cut8 is essential for nuclear localization is the phoprotein-binding domains, and binding assays show that this recent finding that a reduction in Cut8 levels coincides with the domain is necessary and sufficient for liposome and cholesterol altered localization of the proteasome from the nucleus to the binding. Moreover, specific mutations in the 14-3-3 regions corre- cytoplasm upon the transition from vegetative proliferation to sponding to putative cholesterol recognition/interaction amino the G0/quiescent phase (22). Nuclear sequestration of the protea- acid consensus motifs abrogate cholesterol binding. In vivo studies some by Cut8 has also been shown to be critical for double-strand confirmed that the 14-3-3 region is necessary for Cut8 membrane break repair (23). This is likely due to the requirement of the localization and that dimerization is critical for its function. Thus, proteasome for cohesion cleavage. Thus, not only is Cut8 crucial the data reveal the Cut8 organization at the nuclear envelope. for normal anaphase progression because it ensures essential Reconstruction of Cut8 evolution suggests that it was present in anaphase-promoting proteolytic events in the nucleus, but it is the last common ancestor of extant eukaryotes and accordingly that nuclear proteasomal sequestration is an ancestral eukaryotic also required for overall genome stability due to its function in feature. The importance of Cut8 for cell viability and its absence DNA repair events. in humans suggests it as a possible target for the development Interestingly, Cut8 tethers the proteasome in the nucleus of specific chemotherapeutics against invasive fungal infections. through ubiquitinated lysine residues within its N-terminal region. In addition to conferring tight binding of Cut8 to the 26S proteasome, ubiquitination also results in a short half-life of the he 26S proteasome is a large supramolecular machine that Tcarries out essential ubiquitin-mediated proteolysis events in Cut8 protein (approximately 3 min) as Cut8 itself is ultimately all eukaryotes (1–3). The importance of the proteasomal degra- a substrate of the proteasome. Takeda and Yanagida proposed dation of cytosolic proteins, such as antigens for presentation by that the short-lived nature of Cut8 is critical for feedback enrich- major histocompatibility complex molecules, is well known (4). ment of proteasome inside the nucleus, allowing Cut8 to act as a However, a large number of nuclear proteins also acquire the proteasome sensor (11). Given its central role in essentially all proteasomal-targeting polyubiquination tag as a consequence of cellular processes, the way(s) in which the proteasome is localized ubiquitin-activating, ubiquitin-conjugating, and ubiquitin-ligating within cells is of great importance. Cut8 represents the best char- enzymes (5). Nuclear proteins that are degraded in this manner acterized proteasome anchor protein. However, the molecular by the proteasome include the securin protein, which cleaves co- mechanism by which Cut8 binds and retains the proteasome is hesin to allow proper chromosome segregation (6–8). Although unknown, and Cut8 shows no homology to any protein of known ubiquitin-mediated proteasomal degradation of nuclear proteins structure. To gain insight into Cut8 function, we carried out struc- plays key roles in cell viability, the mechanisms by which the pro- tural, biochemical, and in vivo functional studies on the S. pombe teasome is sequestered in the nucleus is still not well understood. Cut8 protein. In the fungi Schizosaccharomyces pombe, data clearly demonstrated that the proteasome is enriched in the nucleus and, prin- Author contributions: K.T. and M.A.S. designed research; K.T., N.K.T., T.G., W.X., and M.A.S. cipally, the nuclear envelope (9). This localization was found to performed research; K.T., E.V.K., M.Y., and M.A.S. analyzed data; and M.A.S. wrote be mediated by the nuclear envelope protein, Cut8 (10, 11). the paper. Cut8 was originally identified in fission yeast as a tempera- The authors declare no conflict of interest. – ture-sensitive mutant, Cut8 563, that gives rise to the cut or cell This article is a PNAS Direct Submission. untimely torn phenotype (12, 13). cut8 mutants do not complete Freely available online through the PNAS open access option. mitosis and have hypercondensed chromosomes and a short Data deposition: The atomic coordinates for the P1 and C2 Cut8 structures have been spindle (13). This phenotype is similar to that displayed by yeast deposited in the Protein Data Bank, www.pdb.org (PDB ID codes 3Q5W and 3Q5X, that are defective in 26S proteasome components (14–16). In respectively). both cut8 and proteasome mutants, chromosome missegregation 1To whom correspondence should be addressed. E-mail: [email protected]. and aberrant spindle dynamics occur, but cytokinesis is normal, This article contains supporting information online at www.pnas.org/lookup/suppl/ leading to the cut phenotype. The cause of this phenotype was doi:10.1073/pnas.1103617108/-/DCSupplemental.

16950–16955 ∣ PNAS ∣ October 11, 2011 ∣ vol. 108 ∣ no. 41 www.pnas.org/cgi/doi/10.1073/pnas.1103617108 Downloaded by guest on September 26, 2021 Results and Discussion ubiquitination of four lysine residues in the N-terminal segment: Structure Determination of S. pombe Cut8. For structural and bio- Lys10, Lys11, Lys13, and Lys22. Whereas Lys10, Lys11, and Lys13 chemical studies an artificial S. pombe cut8 gene, codon-opti- form part of the disordered region in the Cut8 structure, Lys22 mized for expression in Escherichia coli, was utilized (SI Materials is visible and resides in the solvent exposed loop region of the and Methods). The full-length (FL) Cut8 protein was susceptible N-terminal arm (Fig. 1C). Interestingly, the ubiquitination of to proteolysis, and mass spectroscopic analyses revealed that it these lysines also serves as a degron tag for eventual destruction degraded to a stable fragment corresponding to residues 1–217. by the proteasome. Hence, the extended, flexible nature of the Therefore, Cut8(1–217) was produced for structural studies and Cut8 N-terminal region is consistent with its role as a proteasome two crystal forms, P1 and C2, were obtained. The P1 crystal form binding and sensor module. – was solved by multiple wavelength anomalous diffraction (MAD) The second domain of Cut8, residues 32 71, is composed of α (Fig. 1A). The structure contains two Cut8 molecules in the crys- three short -helices that form an intimate dimer connecting – tallographic asymmetric unit (ASU) and the final model includes regions 1 and 3. Domain 3, from residues 72 216, forms a large, residues 18–216 of one subunit, 19–216 of the second subunit, antiparallel six-helix bundle, which extends more than 40 Å below R ∕R the dimer domain. Cut8 is a domain-swapped dimer, whereby and has an work free of 23.1%/27.6% to 2.75-Å resolution (24–26). The C2 crystal form was solved by Molecular Replace- domain 3 of one subunit interacts with the N-terminal arm of ment using the P1 structure as a search model and contains the other subunit to specifically affix and orient the arm (Fig. 1C). one Cut8 subunit in the ASU (Fig. 1B). The C2 model includes Our proteolysis studies indicate that the C-terminal region, from – R ∕R residues 217–262, is disordered or highly flexible. Cut8 residues 25 216 and has been refined to an work free of 26.8%/29.4% to 2.98-Å resolution (Table S1). Cut8 Forms an Intertwined Dimer. Although previous studies using GST pull downs showed that Cut8 interacts with itself, the specific Cut8 Contains a Unique, Modular Fold. The structures show that Cut8 is an all-helical protein with the topology (α1; residues oligomeric state of Cut8 has been unknown (11). The crystal structure revealed the presence of an extensive dimer, which buries 35–42, α2; 46–57, α3; 60–69, α4; 75–92, α5; 105–125, 310; 126– 2 4;995 Å of protein surface from solvent (Fig. 1D). This is a unique 129, α6; 135–149, α7; 156–181, α8; 190–201, α9; 203–214). The dimerization module as no similar structures were identified in overall Cut8 structure contains a unique, modular fold that structural homology searches (27). In this dimer, helices α2and can be broken into four main regions, which are composed of α2′ (where ′ indicates other subunit in the dimer) interact to form either helical bundles or loop segments. The first region, consist- a parallel, coiled coil. The sides of the α2-α2′ coiled coil are sur- ing of residues 18–31, forms an extended arm. Residues 26–31 are rounded by α1andα3. α3 interacts with both α1′ and α2′ to seal the anchored against the helical body of the protein allowing residues “ ” α α ′ α ′ – top of the dimer, whereas 1 interacts with 2 and 3 to close 18 25 to project outward, more than 35 Å from the protein core “ ” – off the bottom portion of the dimer (Fig. 1D). The Cut8 dimer (Fig. 1C). Residues 1 17 are disordered and likely contribute interface is composed almost entirely of hydrophobic residues to the flexible nature of the N-terminal arm. Thus, when fully and includes Leu47, Phe48, Ile50, Leu51, Cys54, and Val55 from extended, the arm could project more than 80 Å. Previous studies α α – both 2 helices and Leu61, Ile65, Ile68, and Leu69 from helix 3 have shown that residues 1 72 bind the 26S proteasome (10, 11). and α3′. α1 residues Leu35 and Leu39 from both subunits com-

This tethering and sequestration function is dependent upon the plete the tight packing of the dimeric hydrophobic core. The dimer BIOCHEMISTRY is further stabilized by hydrogen bonds between the twofold related Lys57 and Asp60 residues and stacking interactions between the twofold related His58 side chains (Fig. 1D).

Chemical Cross-Linking and Gel Filtration Studies Indicate That Cut8 Is a Dimer. The structure suggests that Cut8 is a dimer, and this seems supported by the extensive buried surface area in the dimer and the fact that the identical dimer is present in two different crystal forms (the P1 and C2 dimers superimpose with an rmsd of only 0.70 Å for all Cα atoms) (Fig. 1B). To examine the Cut8 oligomeric state in solution, we carried out glutaraldehyde cross- linking and size exclusion chromatography experiments. Both studies support that Cut8 is a dimer (chromatography produced a calculated molecular mass ¼ 56 kDa compared to the expected value of 56 kDa) (Fig. 2A). Glutaraldehyde cross-linking also revealed the presence of higher-order oligomers, consistent with data from previous studies and our finding that Cut8 aggregates (11). To ascertain whether the dimer observed in the structure is that found in solution, we changed hydrophobic residues that the structures indicate are key for dimerization, to glutamate, to break up the hydrophobic interface core. Two mutants were constructed; L51E and the double mutant, L39E/I65E. We predicted that one problem with the construction of such mutations would Fig. 1. Structure of S. pombe Cut8. (A) Section of the experimental MAD be a significant loss in solubility as a result of exposing the hydro- electron density map contoured at 1σ.(B) Superimposition of the Cut8 dimers phobic interface to solvent. Indeed, both the L51E and L39E/ from the two crystal forms. (C) Structure of the Cut8 subunit. The three func- I65E mutant proteins were almost completely insoluble and were tional regions are colored differently; the proteasome binding arm is red, prone to degradation. Not enough stable L51E could be pro- the dimer domain is pink, and the helical bundle region is blue. The asterisk duced for studies, and several preparations of L39E/I65E were indicates the location of Lys22, which when ubiquitinated is involved in proteasome binding. (D) Cut8 dimer interface. Cut8 is shown in the same or- combined for size exclusion chromatography experiments. These ientation as C. One subunit is cyan and the other magenta. Below the ribbon studies revealed a clear peak consistent with a monomer diagram is a close-up view of the dimer interface. A–D were made using (molecular mass ¼ 27 kDa compared to the expected molecular PyMOL (43). mass ¼ 28 kDa) (Fig. 2B). Thus, the combined experiments

Takeda et al. PNAS ∣ October 11, 2011 ∣ vol. 108 ∣ no. 41 ∣ 16951 Downloaded by guest on September 26, 2021 mutants did not (Fig. 2C). In the absence of thiamine (when proteins were overexpressed at very high levels), WTand Cut8 mutants all rescued the temperature sensitivity of Δcut8 (Fig. 2C and Fig. S1). Thus, these data indicate that dimerization is functionally important as dimer mutations compromise Cut8 function, but the inefficiency of dimerization of the mutants may be partially alleviated by the significant overproduction of these proteins, which could allow Cut8 dimerization in vivo at sufficient levels for res- cuing the Δcut8 temperature sensitivity.

The Cut8 Six-Helix Bundle Is Structurally Similar to 14-3-3 Domains. Cut8 shows no sequence homology to any protein and the crystal structure shows it to contain an unusual organization of domains. Structure homology searches failed to uncover any protein that is structurally similar to the Cut8 monomer or dimer, indicating that the overall Cut8 structure contains a newly described fold (27). However, when individual domains were used to search the database, the Cut8 six-helix bundle was found to have weak similarity to 14-3-3 proteins (28, 29). The resulting rmsd for comparison of helices 4–8 from the Cut8 six-helix bundle with helices 3–7 of ζ 14-3-3 was 4.4 Å for 110 similar Cα atoms (Fig. 3A). Struc- tures of 14-3-3 proteins show that they are composed of nine antiparallel α-helices and form dimers. However, despite the resemblance of their antiparallel helical bundle regions, Cut8 and 14-3-3 proteins form distinct dimers (Fig. 3A). Specifically, in 14-3-3 proteins the first two and the last two helices are involved in dimerization, whereas in Cut8, the dimer domain is composed

Fig. 2. Cut8 dimerization. (A) Glutaraldehyde cross-linking of Cut8(1–217). Dimers are seen at the earliest time points (in minutes). (B) Size exclusion chromatography experiment. The y axis is the elution volume normalized for column volume (CV) and the x axis is the Log of the molecular weight (MW). Red circles indicate standards. The positions of WT Cut8(1–217) protein (magenta square) and L39E/I65E double Cut8(1–217) mutant (green circle) result in molecular mass of 56 kDa and 27 kDa, respectively. (C) Rescue assay examining affects of dimer mutations on Cut8 function. The null mutant, Δcut8, was transformed by plasmid expressing WT and dimer mutants of Cut8 and their colony formation examined at 26 and 36 °C. Unlike WT Cut8, the dimer mutants failed to rescue the temperature sensitivity in the presence of thiamine at 36 °C. Shown below are relative levels of protein expression as assayed by immunoblot. Asterisk shows a protein cross-reacted with anti-Cut8 antibody. Arrow indicates Cut8 protein expressed from plasmids.

support that the Cut8 dimer observed in our two crystal structures Fig. 3. (A) Cut8-14-3-3 homology and in vivo localization. Superimposition is found in solution. We next tested the effects of these dimer of Cut8 with ζ 14-3-3 protein. The Cut8 dimer is red and the ζ 14-3-3 dimer mutations in vivo. is cyan. Phosphopeptides that bind the 14-3-3 protein are blue surfaces. (B, Left) Model of Cut8 organization on the membrane. The location of truncation sites 202 and 217 are labeled. (Right) Electrostatic surface repre- Cut8 Dimerization Is Important for Function. To assay for the affects sentation of the Cut8 dimer shown modeled on the nuclear membrane. Blue of the dimer mutants in vivo, we took advantage of the fact that regions represent electropositive regions and red, electronegative. (C) Rescue the null mutant of Cut8 (Δcut8) is temperature sensitive (13). We assays examining the importance of the Cut8 14-3-3 similarity domain for introduced genes encoding WT Cut8, the artificial gene expressing function. The endogenous cut8 gene was replaced with FL Cut8, Cut8(1–225), – – WT Cut8, and our Cut8 dimer mutants into the S. pombe expres- Cut8(1 217), and Cut8(1 202), to which GFP gene was fused to the C terminus Δ and colony formation of strains examined at 26 and 36 °C. Although FL Cut8, sion vector Rep81 and transformed the plasmids into cut8 cells. Cut8(1–225), and Cut8(1–217) formed colonies at 36 °C, Cut8(1–202), which The genes were inserted after the nmt81 promoter such that the lacks helix 9 of the 14-3-3 domain, showed severe growth defects at 36 °C. encoded proteins were overexpressed upon withdrawal of thiamine. Below is immunoblot showing relative protein expression levels in cells. Even in the presence of thiamine (when they were repressed), (D) Examination of the cellular localization of FL Cut8, Cut8(1–225), Cut8 – – WT and mutant Cut8 proteins were expressed and at similar levels (1 217), and Cut8(1 202). (Left) GFP was fused to the C terminus of FL Cut8, Cut8(1–225), Cut8(1–217), and Cut8(1–202), and these fusion proteins were (Fig. 2C, Lower). The WT cut8 gene and the E. coli codon-opti- expressed from the native locus of cut8þ gene. (Right) Relative quantification mized cut8 gene both rescued the Δcut8 mutant in the absence of the GFP signals in the regions surrounded by hashed squares in photo- of thiamine. By contrast, L39E, L51E, I65E, and I65E/L39E Cut8 graphs. A and B were made using PyMOL (43).

16952 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1103617108 Takeda et al. Downloaded by guest on September 26, 2021 of N-terminal helices 1-3, which have no counterpart in 14-3-3 its location on helical regions. Specifically, CRAC motifs are proteins. It is known that 14-3-3 proteins bind a diverse array found on membrane juxtaposed helical segments of proteins and of serine and threonine phosphorylated proteins. Interestingly, are surface exposed (32). The putative CRAC motifs in Cut8 fit studies have also indicated that certain 14-3-3 proteins may inter- these criteria (Fig. 4A). act with phospholipids (30, 31). To determine if the Cut8 14-3-3 region and the CRAC-like motifs, in particular, are involved in lipid binding, we constructed Cut8 and Cut8 Mutant Localization Studies. The possibility that 14- maltose binding protein (MBP)-Cut8(1–202) and a CRAC mu- 3-3 domains may interact with membranes suggested the possibi- tant, Cut8(1–217)-Y94A/R99A/Y102A/R106, herein called Cut8 lity that the Cut8 14-3-3 homology region may be responsible for (1–217)-4A, in which the tyrosine and arginine residues in the membrane binding. Indeed, modeling shows that these regions predicted CRAC-like motifs were mutated to alanines (Fig. 4A). in the Cut8 dimer are suitably positioned to grip either side of a Multiple mutations were made as previous studies on CRAC- membrane (Fig. 3B). To test this idea, FL Cut8, Cut8(1–217), containing proteins have indicated that more than one residue Cut8(1–225), all of which contain the six-helix bundle, and of this motif must generally be mutated to observe an effect (32). Cut8(1–202), which is missing the last helix of the six-helix bun- Gel filtration showed that like WT Cut8(1–217), Cut8(1–217)-4A dle, were attached to GFP (11). The endogenous cut8þ gene of is dimeric (Fig. S2). We first analyzed the ability of FL Cut8 and S. pombe WT strain was replaced with the GFP fusions such Cut8(1–217) to interact with liposomes and found that both that expression of the GFP fusion Cut8 proteins was controlled bound avidly to liposomes (Fig. S3). By contrast, MBP-Cut8 by the native promoter. Immunoblots showed that the proteins (1–202), MBP, and Cut8(1–217)-4A showed essentially no bind- are expressed at comparable levels (Fig. 3C). The localization ing (Fig. 4B). To test the possibility that CRAC motifs may be patterns of these proteins were then analyzed by fluorescence mi- involved in cholesterol binding, we used a fluoresceinated choles- croscopy (Fig. 3 C and D). Examination of several cells revealed terol molecule in fluorescence polarization assays. Strikingly, – – K that although Cut8(1 202) is present as a diffuse stain in the Cut8(1 217) bound the fluoresceinated cholesterol with a d of nucleus, FL Cut8, Cut8(1–225), and Cut8(1–217) are localized to 2.1 0.5 μM, whereas MBP, MBP-Cut8(1–202), and Cut8 the nuclear envelope (Fig. 3D). Thus, the Cut8 six-helix bundle is (1–217)-4A showed no or only weak and nonsaturable binding important for membrane tethering. Indeed, although Cut8(1–217) (Fig. 4C). We next examined whether Cut8(1–217)-4A is func- rescued the Δcut8 phenotype, Cut8(1–202) did not (Fig. 3D). tional in vivo. Cut8(1–217) and Cut8(1–217)-4A genes were Although the in vivo localization studies show that the Cut8 cloned into the Rep81 expression vector, and Δcut8 cells were 14-3-3 region is required for membrane localization, they do not transformed by these plasmids. As shown in Fig. 4D, although eliminate the possibility that this could be an indirect, or protein- Cut8(1–217) rescued the temperature sensitivity of Δcut8, Cut8 mediated, interaction. Thus, we next tested the direct interaction (1–217)-4A did not, even when expressed at high levels. Taken of purified Cut8 proteins with lipid components found in the together, these results indicate that the 14-3-3 region of Cut8 nuclear envelope (e.g., cholesterol/phospholipid mixtures) (31). directly interacts with lipids. The fact that cholesterol is a key lipid in the nuclear membrane was particularly intriguing given that the Cut8 14-3-3 homology Location of Cut8-563 Mutant in the 14-3-3-like Domain Suggests Expla- region contains several aromatic and basic residues, which are nation for cut Phenotype. The cut8–563 mutant, which led to the found in cholesterol-binding motifs (32). In particular, Cut8 con- identification of the Cut8 protein, results in a serine to proline BIOCHEMISTRY tains two motifs with similarity to cholesterol recognition/interac- substitution at residue 201 (10, 13). The structure shows that tion amino acid consensus (CRAC)-like sequences (Fig. 4A). residue 201 is located in the 14-3-3 region (Fig. 5A), suggesting CRAC motifs are loosely conserved cholesterol-binding motifs that this mutation may affect Cut8 membrane binding. To test with the consensus L∕V-X1–5-Y-X1–5-R∕K. Although this is a this idea, we examined the localization of Cut8–563 at permissive very loose consensus and can be found in multiple, nonmem- (26 °C) and restrictive (36 °C) temperatures (Fig. 5B). For these brane-associated proteins, further support for its occurrence is experiments, the octa-Myc epitope was fused to the C terminus

Fig. 4. Cut8 14-3-3 domain interaction with liposomes and cholesterol. (A) Location of residues that are part of putative cholesterol interaction (CRAC) motifs in Cut8. A was made using PyMOL (43). (B) Liposome binding assays of Cut8(1–217), Cut8(1–217)-4A, MBP-Cut8(1–202), and MBP. (C) Fluorescence polarization binding assays testing the ability of Cut8 proteins to bind fluoresceinated cholesterol. (D) Rescue assay showing that Cut8(1–217)-4A does not rescue the temperature sensitive Δcut8 phenotype. Colony formation of Cut8(1–217) or Cut8(1–217)-4A expressing cells examined at 26 °C and 36 °C without thiamine. (Right) Immunoblot showing that Cut8(1–217) and Cut8(1–217)-4A were expressed at the same levels.

Takeda et al. PNAS ∣ October 11, 2011 ∣ vol. 108 ∣ no. 41 ∣ 16953 Downloaded by guest on September 26, 2021 Fig. 5. Cut8–563 localization. (A) The site of temperature sensitive mutation (S201) mapped onto the Cut8 structure. A was made using PyMOL (43). (B) WT Cut8-Myc8 and Cut8–563-Myc8 proteins were immunostained by anti-Myc monoclonal antibody at 26 °C and 36 °C. Chromosomal DNA was visualized by DAPI. Arrow indicates a cell showing defects in chromosome segregation. (C) Anti-Myc immunoblot showed that the Cut8–563 is expressed in vivo at levels comparable to WT.

of the cut8 genes, and the genes were under control of the native cut8+ promoter. Cut8 protein was visualized by immunofluores- cence against the Myc epitope (33). Immunoblot experiments showed that Cut8–563 and WT Cut8 were expressed at comparable levels (Fig. 5C). In WT and cut8–563 at 26 °C, Cut8 was localized to the nuclear envelope and the nucleus. However, Fig. 6. Sequence alignment of representative Cut8 homologs. (A) Align- Cut8–563 was not concentrated in the nucleus, but rather dis- ments of Cut8 orthologs from representative eukaryotes in which Cut8 persed throughout the cytoplasm at 2 h after the temperature proteins have been identified. Asterisks and colons denote residues that shift to 36 °C (Fig. 5B). Similar results were obtained from experi- are conserved or highly conserved, respectively. Conserved hydrophobic and ments using GFP-tagged, instead of Myc-tagged Cut8 (Fig. S4). hydrophilic residues are highlighted in yellow and green. The secondary These results are consistent with the idea that the mutation in structure is shown above the alignment (B) Mapping of conserved residues from the sequence alignment onto the Cut8 dimer structure. The conserved helix 8 in the 14-3-3 homology domain causes the mislocalization residues (hydrophobic are yellow and hydrophilic, green) are dotted sticks. of Cut8 at 36 °C, which prevents localization of the proteasome to Sites of mutations in Cut8(217)-4A are indicated by asterisks (red for tyrosine the nuclear envelope. and blue for basic residues). B was made using PyMOL (43). Thus, our combined structural, biochemical, and functional data suggest a model for Cut8 function at the nuclear envelope. tant for either fold or dimer formation (Fig. 6 A and B). Each In this model, the six-helix bundle (14-3-3) region docks on the of these genomes encodes a single Cut8 ortholog. nuclear envelope and acts as a platform to orient the Cut8 Further, extensive database searches with the PSI-BLAST N-terminal arms (Fig. 3B). This specific positioning of the program (34) using the detected Cut8 ortholog sequences as N-terminal arms allows the lysines to be flexibly displayed inside queries and additional searches using the highly sensitive struc- the nucleus where they can be ubiquitinated and act as protea- ture-based HHSearch program (35) failed to identify Cut8 some binding sensors. homologs in other groups of animals such as the basal animal trichoplax, nematodes, urochordates, and chordates. Further- Cut8 Homologues: Evolutionary Implications and Conservation of more, no homologs of Cut8 were found in plants or Chromalveo- Dimerization Residues. Studies have demonstrated the importance lata. Nevertheless, given that Cut8 orthologs were detected in of Cut8 in nuclear proteasome function in yeast and showed that a Cut8 homolog plays a similar role in Drosophila melanogaster. three of the five supergroups of eukaryotes (36, 37), namely Uni- However, the possible role of Cut8 in mammals has not been konts, Metazoa, Fungi, Amoebozoa, and several additional determined. Indeed, known Cut8 homologs show a low degree groups of protists, Archaeplastida (plants and green algae) and of sequence similarity, which has made their identification diffi- Excavata, these findings strongly suggest that Cut8 was present cult. Having the structure in hand allows us to use structural in the last common ancestor of the extant eukaryotes. constraints to aid in identification of Cut8 proteins. Using an Subsequent evolution apparently involved multiple, indepen- iterative Position-Specific Iterative Basic Local Alignment Search dent losses of cut8 genes in several lineages including, among Tool (PSI-BLAST) search (34) of the nonredundant protein others, the chordates, after their divergence from the common sequence database (National Center for Biotechnology Informa- ancestor with Cephalochordata (amphioxus), and plants, after tion, National Institutes of Health), we identified apparent ortho- the divergence from the common ancestor with the green algae. logs of Cut8 in all available fungal genomes, many metazoa, The alternative possibility, that the Cut8 sequences in chordates including the primitive radial animal Nematostella vectensis (sea and other groups diverged beyond recognition, cannot be ruled anemone), insects, sea urchin, and amphioxus, two amoebozoan out but seems unlikely given the highly statistically significant genomes (Dictyostelium discoideum and Polysphondylium palli- sequence similarity between the protist, fungal, and representa- dum), the green alga Micromonas pusilla, and the free-living tive animal Cut8 proteins (Fig. 6A). If extensive divergence is to excavate Naegleria gruberi. Although Cut8 shows relatively low explain the lack of detectable Cut8 homologs in these groups, this sequence identity to all these orthologous proteins (typically hypothesis would be conditioned on a drastic change of the <20%), the similarity was highly statistically significant and the function of this protein occurring on multiple occasions during conserved residues are those the structures indicate are impor- evolution.

16954 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1103617108 Takeda et al. Downloaded by guest on September 26, 2021 Notably, sequence alignments of the detected Cut8 homologs Studies indicate that in chordates and higher eukaryotes the show conservation only in residues encompassing helices 1–7 proteasome is localized differently depending on the cell type (Fig. 6A). The N- and C-terminal regions of these proteins reveal and cell cycle status (17–21). By contrast, in yeast, the 26S pro- essentially no sequence or length similarity. However, although teasome is localized predominantly in the nucleus (38–42). This the N-terminal regions are poorly conserved, they share an difference may reflect the fact that higher eukaryotes harbor an abundance of lysine residues that are the apparent targets of adaptive immune system and therefore depend on antigen degra- ubiquitination. A distinctive motif of KR, RKR, or K(R/Q)K dation in the cytosol compared to yeast and insects, which only is observed in this region (Fig. S5). By contrast, the C-terminal utilize innate immunity. Moreover, unlike animals and plants, tails are not only vastly different in lengths but show no sequence S. pombe performs closed mitosis whereby the nuclear envelope conservation (Fig. S6). When the conserved residues are mapped does not break down during cell division allowing it to retain the onto the S. pombe Cut8 structure, we find that these include proteasome. Thus, the observed differences in proteasome loca- all the hydrophobic residues (such as Leu39, Leu51, and Ile65) lization between yeast and higher eukaryotes are consonant that are located in the dimer interface (Fig. 6B). The remaining with the suggestion that structural homologs of Cut8 may not be conserved hydrophobic residues are involved in the formation of present in the latter organisms. This difference could be exploited the core of the six-helix bundle on helices 4–7. Only two polar residues are strongly conserved, His58 and Arg106. As noted, in the development of specific chemotherapeutics in the treat- His58 is important in dimerization and Arg106 forms part of a ment of emerging drug-resistant fungal infections. putative CRAC motif that we demonstrated is involved in lipo- – Materials and Methods some and cholesterol binding (Fig. 4 A C). CRAC motifs are not The P1 Cut8 structure was solved by MAD using crystals grown with seleno- highly conserved in terms of their locations, even among func- methionine-substituted protein. Model building was carried out using O and tional homologs. Indeed, more critical than the precise location refinement with Crystallography and NMR System. The C2 structure was of the motif within the protein sequence is that it is exposed on solved by MR (24–26). Details of structural, biochemical, and in vivo func- helical segments near the membrane. Consistent with this idea, tional studies are in SI Materials and Methods. Cut8 proteins are all extremely R/K- and Y-rich, and these residues are predicted to be surface exposed on the helical bundles ACKNOWLEDGMENTS. This work was supported by an M. D. Anderson Trust such that they could be involved in membrane binding (Fig. 6). Fellowship and National Institutes of Health Grant GM074815 (to M.A.S.).

1. Wilkinson CR, Perry M, McGurk G, Wallace M, Gordon C (1999) The 26S proteasome of 22. Takeda K, et al. (2010) Synergistic roles of the proteasome and autophagy for mito- the fission yeast Schizosaccharomyces pombe. Philos Trans R Soc Lond B Biol Sci chondrial maintenance and chronological lifespan in fission yeast. Proc Natl Acad Sci 354:1523–1532. USA 107:3540–3545. 2. Coux O, Tanaka K, Goldberg AL (1996) Structure and functions of the 20S and 26S 23. Kearsey SE, Stevenson AL, Toda T, Wang S-W (2007) Fission yeast Cut8 is required for proteasomes. Annu Rev Biochem 65:801–847. the repair of DNA double-strand breaks, ribosomal DNA maintenance and cell survival 3. Zwicki P, Voges D, Baumeister W (1999) The proteasome: A macromolecular assembly in the absence of Rqh1 helicase. Mol Cell Biol 27:1558–1567. designed for controlled proteolysis. Philos Trans R Soc Lond B Biol Sci 354:1501–1511. 24. Brünger AT, et al. (1998) Crystallography & NMR system: A new software suite 4. York IA, Rock KL (1996) Antigen processing and presentation by the class I major for macromolecular structure determination. Acta Crystallogr D Biol Crystallogr histocompatibility complex. Annu Rev Immunol 14:369–396. 54:905–921. 5. Hershko A (2005) The ubiquitin system for protein degradation and some of its roles in 25. Jones TA, Zou JY, Cowan SW, Kjeldgaard M (1991) Improved methods for building

the control of the cell division cycle. Cell Death Differ 12:1191–1197. protein models in electron density maps and the location of errors in these models. BIOCHEMISTRY – 6. Funabiki H, et al. (1996) Cut2 proteolysis required for sister chromatid separation in Acta Crystallogr A 47:110 119. fission yeast. Nature 381:438–441. 26. Terwilliger TC, Berendzen J (1999) Automated MAD and MIR structure solution. Acta – 7. Yanagida M (2000) Cell cycle mechanisms of sister chromatid separation: Roles of Cut1/ Crystallogr D Biol Crystallogr 55:849 861. separin and Cut2/securin. Genes Cells 5:1–8. 27. Krissinel E, Hendrick K (2005) Multiple alignment of protein structures in three dimen- – 8. Brooks CL, Gu W (2003) Ubuquitination, phosphorylation and acetylation: The mole- sions. Lect Notes Comput Sci 3695:67 78. cular basis for p53 regulation. Curr Opin Cell Biol 15:164–171. 28. Obsilova V, Silhan J, Boura E, Teisinger J, Obsil T (2008) 14-3-3 proteins: A family of – 9. Enenkel C, Lehmann A, Kloetzel PM (1998) Subcellular distribution of proteasomes versatile molecular regulators. Physiol Res 57:S11 S21. implicates a major location of protein degradation in the nuclear envelope-ER 29. Kostelecky B, Saurin AT, Purkiss A, Parker PJ, Mcdonald NQ (2009) Recognition of an intra-chain tandem 14-3-3 binding site within PKC epsilon. EMBO Rep 10:983–989. network in yeast. EMBO J 17:6144–6154. 30. Roth D, et al. (1994) Characterization of 14-3-3 proteins in adrenal chromaffin cells and 10. Tatebe H, Yanagida M (2000) Cut8, essential for anaphase, controls localization of demonstration of isoform specific phospholipids binding. Biochem J 301:305–310. 26S proteasome, facilitating destruction of cyclin and Cut2. Curr Biol 10:1329–1338. 31. Halskau O, et al. (2009) Three-way interaction between 14-3-3 proteins, the N-terminal 11. Takeda K, Yanagida M (2005) Regulation of nuclear proteasome by Rhp6/ region of tyrosine hydroxylase and negatively charged membranes. J Biol Chem Ubc2 through ubiquitination and destruction of the sensor and anchor Cut8. Cell 284:32758–32769. 122:393–405. 32. Expand RM (2006) Cholesterol and the interaction of proteins with membrane 12. Hirano T, Funahashi S, Uemura T, Yanagida M (1986) Isolation and characterization domains. Prog Lipid Res 45:279–294. of Schizosaccharomyces pombe cut mutants that block nuclear division but not 33. Funabiki H, Hagan I, Uzawa S, Yanagida M (1993) Cell cycle-dependent specific cytokinesis. EMBO J 5:2973–2979. positioning and clustering of centromeres and telomeres in fission yeast. J Cell Biol 13. Samejima I, Yanagida M (1994) Identification of cut8+ and cek1+, a novel protein 121:961–976. kinase gene, which complement a fission yeast mutation that blocks anaphase. 34. Altschul SF, et al. (1997) Gapped BLAST and PSI-BLAST: A new generation of protein Mol Cell Biol 14:6361–6371. database search programs. Nucleic Acids Res 25:3389–3402. 14. Gordon C, McGurk G, Dillon P, Rosen C, Hastie ND (1993) Defective mitosis due to a 35. Söding J (2005) Protein homology detection by HMM-HMM comparison. Bioinfor- – mutation in the gene for a fission yeast 26S proteasome subunit. Nature 366:355 357. matics 21:951–60. 15. Gordon C, McGurk G, Wallace M, Hastie ND (1996) A conditional lethal mutant in the 36. Keeling PJ, et al. (2005) The tree of eukaryotes. Trends Ecol Evol 20:670–676. fission yeast 26S proteasome subunit mts3+ is defective in metaphase to anaphase 37. Koonin EV (2010) The origin and early evolution of eukaryotes in the light of phylo- – transition. J Biol Chem 271:5704 5711. genomics. Genome Biol 11:1–12. 16. Wilkinson CR, Wallace M, Seeger M, Dubiel W, Gordon C (1997) Mts4, a non-ATPase 38. McDonald HB, Byers B (1997) A proteasome cap subunit required for spindle pole body subunit of the 26S proteasome in fission yeast is essential for mitosis and interacts duplication in yeast. J Cell Biol 137:539–553. – directly with the ATPase subunit Mts2. J Biol Chem 272:25768 25777. 39. Russell SJ, Steger KA, Johnson SA (1999) Subcellular localization, stoichiometry and 17. Palmer A, et al. (1996) Subpopulations of proteasomes in rat liver nuclei, microsomes protein levels of 26S proteasome subunits in yeast. J Biol Chem 274:21943–21952. and cytosol. Biochem J 316:401–407. 40. Laporte D, Salin B, Daignan-Fornier B, Sagot I (2008) Reversible cytoplasmic localiza- 18. Wilkinson CR, et al. (1998) Localization of the 26S proteasome during mitosis and tion of the proteasome in quiescent yeast cells. J Cell Biol 181:737–745. meiosis in fission yeast. EMBO J 17:6465–6476. 41. Wendler P, Lehmann A, Janek K, Baumgart S, Enenkel C (2004) The bipartite nuclear 19. Brooks P, et al. (2000) Subcellular localization of proteasomes and their regulatory localization sequence of Rpn2 is required for nuclear import of proteasomal base complexes in mammalian cells. Biochem J 346:155–161. complexes via karyopherin alphabeta and proteasome functions. J Biol Chem 20. Hartmann-Petersen R, Gordon C (2004) Proteins interacting with the 26S proteasome. 279:37751–37762. Cell Mol Life Sci 61:1589–1595. 42. Isono E, et al. (2007) The assembly of the 19S regulatory particle of the yeast 26S 21. Scharf A, Rockel TD, Mikecz A (2007) Localization of proteasomes and proteasomal proteasome. Mol Biol Cell 18:569–580. proteolysis in the mammalian interphase cell nucleus by systematic application of 43. Delano WL (2002) The PyMOL Molecular Graphics System (Delano Scientific, San immunocytochemistry. Histochem Cell Biol 127:591–601. Carlos, CA).

Takeda et al. PNAS ∣ October 11, 2011 ∣ vol. 108 ∣ no. 41 ∣ 16955 Downloaded by guest on September 26, 2021