Dr Sheppard is a director of Affiniti Research Products Ltd., vice president of BIOMOL International LP, and owns equity interest in BIOMOL International LP. All other authors have no conflict of interest.
Loss of Ubiquitin-Binding Associated With Paget's Disease of Bone p62 (SQSTM1) Mutations
Article first published online: 6 DEC 2004
Copyright © 2005 ASBMR
Journal of Bone and Mineral Research
Volume 20, Issue 4, pages 619–624, April 2005
How to Cite
Cavey, J. R., Ralston, S. H., Hocking, L. J., Sheppard, P. W., Ciani, B., Searle, M. S. and Layfield, R. (2005), Loss of Ubiquitin-Binding Associated With Paget's Disease of Bone p62 (SQSTM1) Mutations. J Bone Miner Res, 20: 619–624. doi: 10.1359/JBMR.041205
- Issue published online: 4 DEC 2009
- Article first published online: 6 DEC 2004
- Manuscript Accepted: 4 NOV 2004
- Manuscript Revised: 24 SEP 2004
- Manuscript Received: 5 AUG 2004
- Paget's disease of bone;
- ubiquitin-associated domain
We have studied the effects of various PDB-causing mutations of SQSTM1 on the in vitro ubiquitin-binding properties of the p62 protein. All mutations caused loss of monoubiquitin-binding and impaired K48-linked polyubiquitin-binding, which was only evident at physiological temperature. This suggests that SQSTM1 mutations predispose to PDB through a common mechanism that depends on loss of ubiquitin-binding by p62.
Introduction: Mutations in the SQSTM1 gene, which affect the ubiquitin-associated (UBA) domain of the p62 protein, are a common cause of Paget's disease of bone (PDB). We previously showed that the isolated UBA domain of p62 binds K48-linked polyubiquitin chains in vitro and that PDB-causing mutations in the UBA domain can be resolved in to those which retain (P392L and G411S) or lose (M404V and G425R) the ability to bind K48-linked polyubiquitin. To further clarify the mechanisms by which these mutations predispose to PDB, we have extended these analyses to study the ubiquitin-binding properties of the PDB-causing mutations in the context of the full-length p62 protein.
Materials and Methods: We studied the effects of various PDB-causing mutations on the interaction between glutathione S-transferase (GST)-tagged p62 proteins and monoubiquitin, as well as K48-linked polyubiquitin chains, using in vitro ubiquitin-binding assays.
Results: All of the PDB-causing mutations assessed (P392L, E396X, M404V, G411S, and G425R) caused loss of monoubiquitin binding and impaired K48-linked polyubiquitin-binding when introduced into the full-length p62 protein. However, these effects were only observed when the binding experiments were conducted at physiological temperature (37°C); they were not seen at room temperature or at 4°C.
Conclusions: Our in vitro findings suggest that PDB-causing mutations of SQSTM1 could predispose to disease through a common mechanism that is dependent on impaired binding of p62 to a ubiquitylated target and show that 5q35-linked PDB is the first example of a human disorder caused by loss of function mutations in a UBA domain.
PAGET'S DISEASE OF bone (PDB) is a common disorder, characterized by focal increases in bone turnover, affecting between 1% and 3% of individuals over the age of 55 years in white populations.(1,2) The cause of PDB is incompletely understood, but genetic factors play an important role.(3,4) Positional cloning studies have shown that mutations in the SQSTM1 (also known as p62) gene are an important cause of familial and sporadic PDB.(5,6) Moreover, all 11 PDB-causing mutations of SQSTM1 described so far affect the ubiquitin-associated (UBA) domain of the p62 protein,(7–10) an ubiquitin-binding element, implying that the disease may be caused by a common mechanism of action involving the ubiquitin-binding properties of p62. Previous studies have shown that p62 functions as a scaffold protein in signaling pathways downstream of the interleukin (IL)-1, TNF, and nerve growth factor (NGF) receptors, which ultimately lead to activation of NF-κB. Deletion of SQSTM1 in mice has been shown to inhibit IκB kinase activation and to impair osteoclastogenesis in vitro.(11) This raises the possibility that the SQSTM1 mutations that cause PDB activate RANK-NF-κB signaling, as has been shown for activating mutations of the RANK gene, which cause the PDB-like disorders familial expansile osteolysis, expansile skeletal hyperphosphatasia, and early-onset familial PDB.(12–14) While the precise functional significance of the UBA domain of p62 is unknown, we previously showed that introduction of the most common PDB-causing mutation (P392L) into the isolated p62 UBA domain did not affect its ability to bind K48-linked polyubiquitin chains in vitro.(15) We recently confirmed this observation(8) and showed that, when other PDB-causing mutations were introduced into the isolated UBA domain, they could be resolved into those which impaired K48-linked polyubiquitin-binding (M404V and G425R) and those which did not (P392L and G411S). Because there was no apparent correlation between the K48-linked polyubiquitin chain-binding properties of the different mutant UBA domains and disease occurrence or extent, we sought to further clarify the mechanisms by which these mutations predispose to PDB by studying the in vitro ubiquitin-binding properties of the PDB-causing mutations in the context of the full-length p62 protein.
MATERIALS AND METHODS
The coding region for full-length human p62 (residues 1-440) was PCR amplified from an IMAGE clone (2906264), and the resulting product cloned in to the EcoR1/Xho1 sites of plasmid pGEX-4T-1 (Amersham Biosciences), allowing expression of the p62 protein with an N-terminal glutathione-S-transferase (GST) tag. PDB-associated mutations were introduced by site-directed mutagenesis (QuikChange XL kit; Stratagene). All constructs were verified by DNA sequencing. GST-p62 fusions were expressed in E. coli, and pelleted cells were lysed by sonication in 10 mM Tris, 150 mM NaCl, 0.1% (vol/vol) Triton X-100 (TBST), pH 7.5, buffer. Wildtype and PDB-mutant recombinant proteins were found to express at identical levels, and the mutations did not affect GST-p62 solubility. Bacterial lysates were centrifuged, and supernatants containing equal amounts of recombinant proteins (typically from 250 μl of culture) were precipitated for 30 minutes with an excess of glutathione-Sepharose (G) or Sepharose-conjugated monoubiquitin (U; 10 mg/ml ubiquitin, immobilized on CNBr-activated Sepharose 4B). Control Sepharose with no ubiquitin (N) was prepared in parallel. All reagents were maintained at the indicated temperatures throughout the binding/washing stages. After extensive washing in TBST, precipitated proteins were detected by Western blotting with anti-p62 (N terminus). All experiments were repeated on at least three independent occasions, and representative examples of blots are presented.
Polyubiquitin chain-binding assays
Equal amounts (typically 20 μg) of recombinant GST-p62 fusion proteins expressed in E. coli were immobilized on 50 μl glutathione-Sepharose beads and used in K48-linked polyubiquitin chain-binding assays in 50 mM Tris, 0.1% (w/vol) BSA (TB buffer), pH 7.5, as described previously,(8,15) except using 5 μg of ubiquitin chains or in TBST buffer, with all reagents maintained at the indicated temperatures throughout the binding/washing stages. Control beads contained GST only. After washing, bound proteins were detected by Western blotting (anti-ubiquitin). Wildtype isolated UBA domain (p62 residues 387-436) and PDB missense mutants were prepared as described previously(8,15) and assayed using 1 μg of K48-linked polyubiquitin chains. Control beads (CON) were prepared in parallel without the addition of UBA domain polypeptides. All experiments were repeated on at least three independent occasions, and representative examples of blots are presented.
Homology models of the p62 UBA domain complexed with monoubiquitin were constructed based on the published nuclear magnetic resonance (NMR) structure of the yeast Cue2-1 CUE domain-ubiquitin complex.(16) The low root mean square division (RMSD) between backbone Cα atoms in the Cue2-1 CUE and p62 UBA structures (1.3 Å), together with the presence of a conserved hydrophobic ubiquitin-binding patch in both structures, justifies such an approach. Notably, the structural homology between the p62 UBA and Cue2-1 CUE domains is greater than that between the p62 UBA and human Rad23A UBA2 domains (B Ciani and M Searle, unpublished data, 2004).15N chemical shift perturbations to the p62 UBA domain confirm the involvement of the same hydrophobic binding surface involving residues on helices 1 and 3 in the interaction with di-ubiquitin (B Ciani, unpublished results, 2004). The structure in Fig. 3 was drawn using MolMol.(17)
Given that p62 was originally described as an ubiquitin-binding protein capable of interacting with Sepharose-conjugated monoubiquitin,(18) we first assessed the effects of PDB-causing UBA domain mutations on the in vitro interaction between full-length p62 and this model substrate. In this case, the ability of Sepharose-conjugated monoubiquitin to precipitate recombinant wildtype and PDB mutant GST-fusion proteins from bacterial extracts was assessed. At 4°C, we found that the P392L and G411S mutants were, like wildtype, able to bind immobilized monoubiquitin, whereas the M404V and G425R mutants were not (Fig. 1A), although at room temperature (∼20°C) the P392L and G411S mutations slightly reduced the ability of p62 to bind monoubiquitin, relative to wildtype (Fig. 1B). This is a similar effect to that previously reported for K48-linked polyubiquitin binding by the isolated UBA domains containing the mutations.(8) However, at physiological temperature (37°C), we found that none of the p62 missense mutants were able to bind monoubiquitin, whereas the wildtype protein retained this ability (Fig. 1C). Additionally, we found that the PDB-associated E396X mutant, which lacks the majority of the UBA domain, was unable to bind monoubiquitin. Therefore, under these in vitro conditions, all of the PDB-associated mutations we have assessed resulted in a loss of monoubiquitin-binding by p62.
We next examined the effects of the PDB-causing mutations on K48-linked polyubiquitin chain-binding by full-length p62. For these experiments, the ability of glutathione-Sepharose immobilized GST-p62 proteins to precipitate a mixture of K48-linked polyubiquitin chains, at 37°C, was assessed. Initially these experiments were performed in the buffer system (TB buffer) previously used for the polyubiquitin-binding studies with the UBA domain polypeptides.(8,15) Under these conditions, we found that K48-linked polyubiquitin-binding was significantly diminished, relative to wildtype, in the p62 proteins carrying each of the five mutations, in particular for Ub5-Ub7 chains (Fig. 2A, boxed). The effects of the mutations on the interaction of p62 with longer polyubiquitin chains were similar to our earlier findings with the isolated UBA domains,(8) with the P392L and G411S mutations retaining, the M404V mutation reducing, and the G425R mutation almost completely ablating binding to longer chains. A low level of residual binding of shorter polyubiquitin chains was noted for the E396X mutant (which lacks a functional UBA domain), which we attribute to either nonspecific interactions or the presence of a second uncharacterized ubiquitin-binding site in p62 intimated in a previous study,(18) which may account for some of the residual polyubiquitin-binding seen for each of the missense mutants. The reduced binding of Ub5-Ub7 chains associated with the P392L and G411S mutations was not seen in the isolated mutant UBA domains at 37°C (which in previous studies were only assayed at 20°C(8,15)), because the P392L and G411S UBA domains bound all polyubiquitin chains comparable with wildtype at this temperature (Fig. 2C).
The effects of the PDB-causing mutations on K48-linked polyubiquitin-binding by full-length p62 at 37°C were more pronounced when the experiments were carried out in the buffer system used for the monoubiquitin-binding studies described above (TBST buffer). In this case, polyubiquitin-binding by the wildtype p62 protein was much less efficient, and at 37°C, was undetectable with all of the mutants assessed (Fig. 2B).
These data indicate that, in full-length p62, all of the PDB-causing mutations assessed are capable of impairing not only interactions with monoubiquitin, but also with K48-linked polyubiquitin chains.
All 11 of the PDB-causing SQSTM1 mutations described so far affect the UBA domain of p62,(7–10) indicating that the disease may be caused by a common mechanism. Because the normal function of UBA domains is to mediate noncovalent interactions with ubiquitin,(19) we have previously studied the ubiquitin-binding properties of the isolated p62 UBA domain and found that PDB-causing missense mutations resolved into those which impaired (M404V and G425R) or did not affect (P392L and G411S) K48-linked polyubiquitin chain binding in vitro.(8,15) Here, we have extended our studies to the full-length p62 protein and have uncovered a loss of ubiquitin-binding common to all of the PDB-causing mutations assessed, which was not evident when we studied the effects of the mutations in the isolated UBA domain.
Initially, we assessed the effects of the PDB-causing mutations on the interaction between full-length GST-tagged p62 and monoubiquitin covalently immobilized on Sepharose beads. At 4°C and at room temperature, the wildtype protein and P392L and G411S mutants were able to bind to monoubiquitin, whereas the M404V, G425R, and E396X mutants were not, similar to the effects of the missense mutations on K48-linked polyubiquitin binding by the UBA domain in isolation.(8,15) In contrast to these previous findings, we found that, at 37°C, none of the full-length mutants were able to bind monoubiquitin. Because the three other PDB-causing SQSTM1 truncating mutations generate proteins that are shorter (390 and 393 residues) than the 395 residue E396X protein, we predict that all of the truncating mutations will also ablate this interaction. We therefore conclude that, under these in vitro conditions and at a physiological temperature, the P392L, M404V, G411S, G425R, E396X, 390X, and 394X mutations result in a common loss of monoubiquitin-binding in p62, whereas the wildtype protein retains this ability. The temperature-dependent loss of monoubiquitin-binding that specifically affects the P392L and G411S proteins presumably reflects the binding isotherms for these mutants and indicates that these may represent temperature-sensitive mutations that cause subtle changes in the p62 protein, altering its ubiquitin-binding affinity.
We went on to determine the effects of the PDB-causing mutations on K48-linked polyubiquitin-binding by full-length p62, at 37°C. Using glutathione-Sepharose immobilized full-length GST-p62 proteins to precipitate K48-linked polyubiquitin chains from solution, we found that, in a buffer system where polyubiquitin-binding by wildtype p62 was readily detectable (TB buffer), interaction of p62 with Ub5-Ub7 chains was substantially reduced by all of the mutations, and interaction with longer polyubiquitin chains was also reduced by the M404V, G425R, and E396X mutations. In an alternative buffer (that used for the monoubiquitin-binding experiments; TBST buffer) in which wildtype p62 was found to bind K48-linked polyubiquitin chains much less efficiently, we were unable to detect polyubiquitin binding by any of the mutants assessed. These data are consistent with overall weaker p62-ubiquitin interactions in TBST buffer, presumably because of the presence of the Triton X-100 detergent, weakening the hydrophobic interaction, and with all of the mutations further reducing the affinity of the p62-ubiquitin interaction.
The detrimental effects of the P392L and G411S mutations on the interaction of p62 with shorter polyubiquitin chains and monoubiquitin, rather than longer polyubiquitin chains, are likely to result from the higher relative affinity of UBA domains for the latter species.(19) Consistent with this proposal, we noted that, in the TB buffer where the more robust p62-polyubiquitin interactions were noted, the loss of monoubiquitin binding associated with the P392L and G411S mutations was overcome, and the mutants were still able to interact with Sepharose-conjugated monoubiquitin at 37°C (data not shown). It will be of particular interest to determine quantitatively the affinities of the wildtype and PDB mutant proteins for monoubiquitin and polyubiquitin in the future.
It is noteworthy that we did not see reduced binding of Ub5-Ub7 chains associated with the P392L and G411S mutations in the isolated UBA domain at 37°C, indicating the effects of these mutations on ubiquitin-binding by full-length p62 seen at 37°C are unlikely to be caused by simple thermal instability of the mutant UBA domains. This proposal is supported by folding studies that showed that, at 37°C, the wildtype and P392L UBA domain polypeptides were >90% folded, and the G411S peptide was >85% folded (B Ciani and M Searle, unpublished data, 2004). Instead, we propose that the mechanism of ubiquitin recognition may differ in the full-length protein compared with the isolated UBA domain. In support of this proposal, we note that the p62 UBA domain is necessary, but not sufficient, for monoubiquitin binding at 37°C (J Cavey and R Layfield, unpublished data, 2004), implicating other components of p62 in the subtleties of UBA domain-mediated recognition of monoubiquitin. We have previously determined the solution structure of the p62 UBA domain by NMR spectroscopy(15) and found it to contain a hydrophobic patch equivalent to the surface implicated in ubiquitin-binding by other UBA domains.(20) The effects of the missense mutations on polyubiquitin-binding in the isolated UBA domain would be consistent with a simple model that assumes the UBA domain functions as a compact monomer. Within this model, the M404V and G425R mutations both affect the hydrophobic patch of the p62 UBA domain, whereas P392 and G411 are more remote.(8,21) Such a model cannot, however, account for the detrimental effects of the P392L and G411S mutations on monoubiquitin or polyubiquitin-binding uncovered in the full-length p62 protein at 37°C. For example, modeling the p62 UBA-monoubiquitin interaction on the monomeric yeast CUE domain, Cue2-1-monoubiquitin complex(16) shows that neither P392 nor G411 are part of the proposed UBA-ubiquitin interface (Fig. 3; CUE domains are structurally related to UBA domains and are functional monoubiquitin-binding domains). Instead, we speculate that the mechanism of ubiquitin recognition in full-length p62 is likely to be different to that in the isolated UBA domain and may require structural rearrangements of the UBA domain (e.g., similar to those noted in the yeast Vps9p CUE domain, where a domain swapped dimer is involved in ubiquitin recognition(22)), which can be inhibited by subtle conformational changes associated with the P392L and G411S mutations.
Our in vitro findings suggest that UBA domain mutations in p62 may predispose to disease through a single mechanism, which depends on loss of binding of p62 to an ubiquitylated target protein(s). However, whether monoubiquitin-binding (which in other contexts regulates nondegradative processes such as endocytic trafficking and transcriptional regulation), K48-linked polyubiquitin-binding (which regulates protein degradation by the 26S proteasome), or binding to other polyubiquitin chains, for example, K63-linked (which regulates NF-κB signaling pathways(23)) is physiologically relevant for p62 is currently unclear. We speculate that the interaction of p62 with an ubiquitylated target(s) is likely to be central to the control of osteoclast NF-κB signaling and that 5q35-linked PBD p62 mutations affect this process with pathological consequences. To date, natural ubiquitylated substrates of the p62 UBA domain have not been identified, but could include TNF receptor-associated factor (TRAF)6, which relays the RANK signal in a ternary complex involving p62 and atypical protein kinase C (aPKC).(11) K63-linked autoubiquitylation of TRAF6 is required for TAK1 kinase-mediated phosphorylation and activation of IKKß by IL-1β and other proinflammatory cytokines,(23) and may also be a mechanism regulating RANK signaling.(24) p62 directly binds TRAF6 in a stimulation-dependent manner through a region not involving the UBA domain.(25) Other UBA domain proteins have been shown to regulate polyubiquitin chain extension(26) or disassembly.(27) We speculate that p62, through its UBA domain, could control osteoclast NF-κB signaling by either regulating K63-linked polyubiquitin chain extension of monubiquitylated TRAF6, leading to activation of TAK1 kinase, or by mediating TRAF6 proteasomal degradation,(28) presumably through regulating K48-linked polyubiquitin chain extension. It will be of particular importance to determine the relative affinities of p62 for different polyubiquitin chains, including those which are K63-linked, and to study the effects of the PDB-associated mutations on such interactions. Additionally, identification of the in vivo ubiquitylated substrates of p62 should be most informative in determining both the functional significance of the p62-ubiquitin interaction and consequences of the PDB-causing mutations.
Disruption of ubiquitin interactions seems to be a critical event in PDB, as supported by the recent observation that the progressive disorder inclusion body myopathy associated with PDB and frontotemporal dementia (IBMPFD) is caused by mutations that cluster in the N-terminal ubiquitin-binding region of the valosin-containing protein (VCP).(29) Significantly, VCP is known to be a regulator of NF-κB signaling,(30) further emphasizing the emerging link between dysfunction of ubiquitin-mediated processes linked to the RANK-NF-κB axis in PDB and related disorders.
In summary, regardless of the precise mechanism of action of the disease-associated SQSTM1 mutations, this work extends the raft of human conditions that represent disorders of the ubiquitin system(31) to include 5q35-linked PDB, which is the first example of a human disease caused by loss of function mutations in a UBA domain.
The authors thank the Wellcome Trust (RL and JC); the BBSRC (MSS and BC); and the Arthritis Research Campaign (SHR and LJH), for funding part of this work.
- 12002 Incidence and natural history of Paget's disease of bone in England and Wales. J Bone Miner Res 17: 465–471., , , , ,
- 22000 Prevalence of pelvic Paget's disease of bone in the United States. J Bone Miner Res 15: 461–465., , ,
- 31991 Familial aggregation of Paget's disease of bone. J Bone Miner Res 6: 495–500., , ,
- 41983 A family study of Paget's disease of bone. J Epidemiol Community Health 37: 226–231., ,
- 52002 Recurrent mutation of the gene encoding sequestosome 1 (SQSTM1/p62) in Paget disease of bone. Am J Hum Genet 70: 1582–1588., , ,
- 62002 Domain specific mutations in Sequestosome 1 (SQSTM1) cause familial and sporadic Paget's disease. Hum Mol Genet 11: 2735–2739., , , , , , , , , ,
- 72003 Three novel mutations in SQSTM1 identified in familial Paget's disease of bone. J Bone Miner Res 18: 1748–1753., , , , , ,
- 82004 Novel UBA domain mutations of SQSTM1 in Paget's disease of bone: Genotype phenotype correlation, functional analysis and structural consequences. J Bone Miner Res 19: 1122–1127., , , , , , , , , , , , ,
- 92004 Two novel mutations at exon 8 of Sequestosome 1 gene (SQSTM1) in an Italian serie of patients affected by Paget's disease of bone (PDB). J Bone Miner Res 19: 1013–1017., , , , , , , , , , , , , , , , , ,
- 102004 Familial Paget's disease in The Netherlands: Occurrence, identification of new mutations in the sequestosome 1 gene, and their clinical associations. Arthritis Rheum 50: 1650–1654., , , , , ,
- 112004 The atypical PKC-interacting protein p62 is an important mediator of RANK-activated osteoclastogenesis. Dev Cell 6: 303–309., , , , , , ,
- 122000 Mutations in TNFRSF11A, affecting the signal peptide of RANK, cause familial expansile osteolysis. Nat Genet 24: 45–48., , , , , , , , , ,
- 132002 Expansile skeletal hyperphosphatasia is caused by a 15-base pair tandem duplication in TNFRSF11A encoding RANK and is allelic to familial expansile osteolysis. J Bone Miner Res 17: 26–29.,
- 142003 Phenotypic characterization of early onset Paget's disease of bone caused by a 27-bp duplication in the TNFRSF11A gene. J Bone Miner Res 18: 1381–1385., ,
- 152003 Structure of the ubiquitin-associated domain of p62 (SQSTM1) and implications for mutations that cause Paget's disease of bone. J Biol Chem 278: 37409–37412., , , ,
- 162003 Solution structure of a CUE-ubiquitin complex reveals a conserved mode of ubiquitin binding. Cell 113: 621–630., , , , , ,
- 171996 MOLMOL: A program for display and analysis of macromolecular structures. J Mol Graph 14: 51–55., ,
- 181996 p62, a phosphotyrosine-independent ligand of the SH2 domain of p56lck, belongs to a new class of ubiquitin-binding proteins. J Biol Chem 271: 20235–20237., , ,
- 192001 Proteins containing the UBA domain are able to bind to multi-ubiquitin chains. Nat Cell Biol 3: 939–943., , , , , ,
- 202002 Solution structures of UBA domains reveal a conserved hydrophobic surface for protein-protein interactions. J Mol Biol 319: 1243–1255.,
- 212004 Structural and functional studies of mutations affecting the UBA domain of SQSTM1 (p62) which cause Paget's disease of bone. Biochem Soc Trans 32: 728–730., , , , , ,
- 222003 Mechanism of ubiquitin recognition by the CUE domain of Vps9p. Cell 113: 609–620., , , , , ,
- 232001 TAK1 is a ubiquitin-dependent kinase of MKK and IKK. Nature 412: 346–351., , , , ,
- 242004 SQSTM1 and Paget's disease of bone. Calcif Tissue Int 75: 347–357.,
- 252000 The atypical PKC-interacting protein p62 channels NF-kappaB activation by the IL-1-TRAF6 pathway. EMBO J 19: 1576–1586., , ,
- 262001 Ubiquitin-associated (UBA) domains in Rad23 bind ubiquitin and promote inhibition of multi-ubiquitin chain assembly. EMBO Rep 2: 933–938., , ,
- 272003 Ubiquitin-binding proteins protect ubiquitin conjugates from disassembly. FEBS Lett 535: 77–81., ,
- 282000 T-cell-mediated regulation of osteoclastogenesis by signalling cross-talk between RANKL and IFN-gamma. Nature 408: 600–605., , , , , , , , , , ,
- 292004 Inclusion body myopathy associated with Paget disease of bone and frontotemporal dementia is caused by mutant valosin-containing protein. Nat Genet 36: 377–381., , , , , , , ,
- 302002 VCP (p97) regulates NFkappaB signaling pathway, which is important for metastasis of osteosarcoma cell line. Jpn J Cancer Res 93: 296–304., , , , , ,
- 312001 The ubiquitin protein catabolic disorders. Neuropathol Appl Neurobiol 27: 171–179., , ,