Biochemical Society Transactions

Seventh International Meeting on AAA Proteins

Insights into adaptor binding to the AAA protein p97

Heidi O. Yeung, Patrik Kloppsteck, Hajime Niwa, Rivka L. Isaacson, Steve Matthews, Xiaodong Zhang, Paul S. Freemont


The AAA (ATPase associated with various cellular activities) p97 [also known as VCP (valosin-containing protein)] participates in numerous biological activities and is an essential component of the ubiquitin signalling pathway. A plethora of adaptors have been reported for p97, and increasing evidence is suggesting that it is through adaptor binding that p97 is diverted into different cellular pathways. Studying the interaction between p97 and its adaptors is therefore crucial to our understanding of the physiological roles of the protein. The interactions between p97 and the PUB [PNGase (peptide N-glycosidase)/ubiquitin-associated] domain of PNGase, the UBX (ubiquitin regulatory X) domain of p47, and the UBD (ubiquitin D) domain of Npl4 have been structurally characterized. UBX and UBD are structural homologues that share similar p97-binding modes; it is plausible that other proteins that contain a UBX/UBX-like domain also interact with p97 via similar mechanisms. In addition, several short p97-interacting motifs, such as VBM (VCP-binding motif), VIM (VCP-interacting motif) and SHP, have been identified recently and are also shared between p97 adaptors, hinting that proteins possessing the same p97-binding motif might also share common p97-binding mechanisms. In this review, we aim to summarize our current knowledge on adaptor binding to p97.

  • ATPase associated with various cellular activities (AAA)
  • p47
  • p97
  • peptide N-glycosidase/ubiquitin-associated domain (PUB domain)
  • ubiquitin regulatory X (UBX)
  • Ufd1–Npl4


p97, also known as VCP (valosin-containing protein), is a Type II AAA (ATPase associated with various cellular activities) protein, characterized by the presence of two AAA domains, each containing a Walker A (P-loop) and Walker B (DEXX box) motif for ATP binding and hydrolysis respectively [1,2]. C-terminal of the walker B motif is a highly conserved region (termed the SRH for second region of homology), which distinguishes AAA proteins from other P-loop ATPases. p97 functions as a hexamer, which forms a barrel-like structure with two stacked rings [35]. The N domain of p97 comprises an N-terminal double Ψ subdomain (NN) and a C-terminal four-stranded β-barrel subdomain (NC), while the D1 and D2 domains adopt a typical AAA fold [69]. The C-terminal region of p97 is disordered and has not been structurally characterized.

p97 constitutes ∼1% of liver cytosol and has been demonstrated to participate in a wide range of biological activities, including ERAD (endoplasmic reticulum-associated degradation), membrane fusion, transcriptional activation, cell-cycle control and apoptosis [1016] (reviewed in [17]). These activities are at least in part regulated by the ubiquitin signalling pathway, in which p97 is proposed to serve as a molecular chaperone, guiding ubiquitin substrates to the 26S proteasome for degradation. p97 has been reported to interact with a large number of adaptors (reviewed in [18]) that probably regulate its functions by directing it into different cellular pathways [11,16,1929]. The association between p97 and such a wide range of proteins is still poorly understood. However, it is beginning to emerge that certain protein domains, such as UBX (ubiquitin regulatory X), PUB [PNGase (peptide N-glycosidase)/ubiquitin-associated] and PUL [PLAP (phospholipase A2-activating protein), Ufd3p and Lub1p], function as general p97-binding modules. Furthermore, several novel p97-interacting motifs have been identified, which extend our understanding of the recognition of p97 by many adaptors.

Major p97-interacting domains

UBX/UBX-like domains

The UBX domain was originally described as an 80-residue module with unknown function [30]. However, its presence in a variety of p97 adaptors, including p47, p37, Ubx2 and VCIP135 (VCP–p47 complex-interacting protein p135), suggests a role in p97 binding [31]. The UBX domain exhibits a β-grasp fold, with a β-β-α-β-β-α-β secondary structure organization, and the five β-strands are arranged into a mixed sheet [30]. The crystal structure of the p47 UBX domain in complex with p97 ND1 indicates that the interaction is primarily mediated by the central β-sheet through a series of hydrophobic contacts [31]. The most critical interaction is located at the highly conserved loop region between strands 3 and 4 (the S3/S4 loop), where the aromatic side chain of Phe343 inserts into the hydrophobic cleft formed between p97 NN and NC (Figure 1A).

Figure 1 Molecular detail of p97 interaction with three adaptors: p47 UBX domain, Npl4 UBD domain and PNGase PUB domain

(A) and (B) show binding to p97 N domain of p47 UBX and Npl4 UBD respectively, with the p97 N domain surface illustrating the vacuum electrostatic potential to show the hydrophobic binding of both p47 UBX and Npl4 UBD. (A) p47 UBX is shown in cartoon representation, showing the insertion of the aromatic side chain of Phe343 (ball and stick) into the hydrophobic cleft formed between NN and NC. Other relevant binding residues are also shown (ball and stick) for illustrative purposes. (B) Cartoon representation of the solution structure of the UBD domain with binding residues shown as ball and stick representation. The UBD 310 helix is close to the S3/S4 loop position, which provides specific interaction for the UBX domain of p47. (C) The PUB domain (electrostatic surface representation) binds to the p97 C-terminus interacting with the four C-terminal residues of p97 (ball and stick), showing the insertion of the aromatic side chain of p97 Tyr805 into a hydrophobic pocket on the PUB domain.

Ufd1–Npl4 (UN) is another extensively studied adaptor of p97 due to its central role in ERAD [3234]. The N-terminal region of Npl4 contains a UBD (ubiquitin D) domain, which competes with p47 for p97 binding [16]. Despite the lack of sequence similarity, NMR data indicates that the UBD and UBX domains are, in fact, structural homologues with strikingly similar binding mechanisms to p97 [35]. Npl4–UBD and p47–UBX utilize the same binding surface on p97, with critical interacting residues on both domains aligning structurally. The S3/S4 loop region on Npl4–UBD contains a 310-helix that provides specific contacts with p97 (Figure 1B) and is indispensable for maintaining the p97–UN complex. UBX/UBX-like domains are widespread among p97 adaptors, and it is of great interest to determine whether the above observations represent a general p97-binding mechanism for all UBX proteins.

The PUB domain

The PUB (also known as PUG) domain was first identified at the N-terminus of PNGase in higher eukaryotes [36]. ITC (isothermal titration calorimetry) data indicated that the final ten C-terminal residues of p97 (p97 C10) are sufficient to mediate binding with the PUB domain [37]. A crystal structure of the PNGase PUB domain in complex with the final four C-terminus residues of p97 has been published recently (Figure 1C) [38]. The PUB domain consists of a four-helix bundle with a twisted antiparallel β-sheet. The p97–PUB complex is stabilized by a combination of hydrophobic and ionic forces, with a key interaction mediated by the aromatic side chain of the penultimate tyrosine residue (Tyr805) in p97 by inserting into a hydrophobic pocket on the PUB domain (Figure 1C). Phosphorylation of Tyr805 completely abolishes the interaction between p97 and PUB, suggesting a regulatory role of phosphorylation in p97 function.

PUB proteins are present in a wide range of higher eukaryotes (summarized in Table 1). Intriguingly, a number of PUB proteins, such as UBXD1 (UBX domain-containing protein 1), also contain a UBA or UBX domain (Figure 2D) [37]. The co-presence of UBX and PUB is intriguing, as both domains are able to bind p97 simultaneously in vitro [37]. The function of PUB/UBX-containing proteins, is not known; however, given their obvious link with p97, further characterization of these proteins is certainly worthwhile.

View this table:
Table 1 A summarization of PUB domain-containing proteins from selected organisms

Information was extracted from the Pfam database (; the primary accession number and protein description was obtained from the UniProtKB/TrEMBL database ( Redundant sequences are not listed in the Table. UBXD1, UBX domain-containing protein 1.

Figure 2 Sequence alignment of the SHP, VBM and VIM motifs and domain organizations of selected p97 adaptors

Sequence alignment of (A) the SHP box, (B) the VBM motif and (C) the VIM motif, with conserved binding residues shaded in grey. Multiple sequence alignment was prepared using ClustalW and residues were coloured using Jalview according to the percentage of conservation, with darker shades of grey showing higher conservation. (D) Domain organizations of selected p97 adaptors as obtained from the Pfam database, showing the position of p97-binding modules relative to each other. schpo, Schizosaccharomyces pombe; Drome, Drosophila melanogaster.

The PUL domain

The PUL domain was first identified at the C-terminus of several members of the PPPDE [permuted papain fold peptidases of dsRNA (double-stranded RNA) viruses and eukaryotes] superfamily, which function as de-ubiquitinating enzymes in the ubiquitin signalling pathway [39]. Bioinformatics analysis has indicated that PUL is an α-helical domain, also present at the C-terminus of the human PLAP, and its homologues in Saccharomyces cerevisiae (Ufd3p/Doa1) and Schizosaccharomyces pombe (Lub1p), hence the term PUL [39]. The cellular function of Doa1 is not well defined; however, different studies have suggested a role in de-ubiquitination for this protein [24].

Pull-down experiments demonstrated that the PUL domain is responsible for the interaction between Doa1 and the yeast p97 homologue, CDC48 (cell division cycle 48) [40]. Like the PUB domain, p97 C10 is sufficient for Doa1 binding, and, interestingly, this interaction is also abolished by phosphorylation of Tyr805 [38]. On the basis of these findings, Zhao et al. [38] have proposed that the PUB domain and Doa1 might share very similar p97-binding mechamisms. Interestingly, a WD40 repeat region, present at the N-terminus of Doa1 and its homologues, has also been shown to interact with CDC48 by immunoprecipitation in Lub1p [41], this interaction is possibly mediated through an additional protein yet to be identified, suggesting multiple modes of interaction between PUL proteins and p97.

Novel p97-interacting motifs

The SHP box

The interaction between CDC48 and the yeast Derlin-1 homologue, Dfm1p, is mediated through its cytosolic region, which contains two homologous eight-residue motifs (FXGXGQRU; where X is any amino acid and U is a non-polar residue). This motif is also found in the p47 yeast homologue Shp1p, and has been termed the SHP box [25]. p47 possesses a second p97-binding site which shares a high degree of sequence homology with a region in Ufd1 [35]. Interestingly, both regions contain the SHP box with minor variations. Using bioinformatics tools, we have identified the SHP box in a number of other UBX-containing proteins (Figure 2A), and, intriguingly, all are located N-terminal of the UBX domain (Figure 2D), suggesting spatial constraints on the p97 interaction mode. In addition to the previously suggested profile for the SHP box (FXGXGQRU), sequence alignment reveals that a serine residue at position two and a leucine residue at position eight are also highly conserved (Figure 2A).

The SHP-binding site on p97 has been controversial. Pull-down experiments have demonstrated that SHP binding to p97 does not interfere with the binding of the UBX or UBD domain, suggesting an alternative binding site for this motif [32]. This is in agreement with the p47–UBX/p97–ND1 crystal structure, which shows that the SHP box is located above the ND1 plane and interacts with p97 NC of an adjacent hexamer via a crystal contact. NMR data, however, suggest that the region containing the SHP box on Ufd1 interacts with a region between the two p97 subdomains, which partially overlaps with the UBX- and UBD-binding sites [35]. Given the flexible nature of p47 and the UN complex, it is possible that the observed SHP-binding sites represent a snapshot of transient interactions during a series of conformational changes at certain stages during complex formation or the ATPase cycle. Further work is needed to clarify these discrepancies and to better define the role of this eight-residue motif in p97 binding.

The VIM (VCP-interacting motif)

During protein dislocation, p97 is recruited to the endoplasmic reticulum membrane through interacting with various membrane-anchored proteins, such as the membrane-spanning ubiquitin E3 ligases Hrd1 and gp78 (glycoprotein 78) [42,43]. Mouse gp78 interacts with p97 through its C-terminal cytosolic tail, which contains a sequence highly conserved amongst mammals. This sequence was termed the VIM (Figure 2C) [44]. Interestingly, a VIM was identified at the N-terminus of SVIP (small VCP-interacting protein), which also interacts with p97 [45]. The VIM motif is predicted to form an α-helix, and pull-down assays indicate that both the N and D1 domains of p97 are required for VIM binding. However, gp78, SVIP, p47 and UN have all been shown to bind p97 in a mutually exclusive manner [44], hinting that they probably share similar binding sites on p97, which resides in the N domain for p47 and Npl4 [31,35]. It is possible that the presence of D1 is important to induce a conformation in the N domain that is favourable for adaptor binding, leading to the above observation. Further structural work is needed to verify the VIM-binding site on p97.

The function of SVIP is not well understood; however, the simultaneous presence of VIM in gp78 and SVIP suggests a link between these two proteins. Indeed, a recent study suggested that SVIP plays a regulatory role in ERAD by directing p97 away from the gp78 pathway [46].

The VBM (VCP-binding motif)

The VBM was first identified in the polyQ (polyglutamine) tract-containing protein, Atx-3 (ataxin-3) [47]. Atx-3 is a de-ubiquitinating enzyme, and the expansion of the polyQ tract leads to the neurodegenerative disease SCA3 (spinocerebellar ataxia type III)/MJD (Machado Joseph disease). A p97-binding site on Atx-3 has been mapped to a region N-terminal to the polyQ tract, termed the VBM, with the following sequence profile: (L/I/V/Y)R(K/R/W)(R/K/L)RXX(Y/F)(F/K/R/Y). A distinctive feature of this motif is the four consecutive basic amino acid residues that were shown to be essential in mediating p97 interaction.

Using bioinformatics approaches, we have also identified a VBM in the ubiquitin E3 ligase, Hrd1, in higher eukaryotes (Figure 2B). Hrd1 has been shown to interact with p97 through its cytosolic tail, where the VBM is located, suggesting further that VBM is responsible for the interaction between p97 and Hrd1 [42]. This identification of VBM in Hrd1 is intriguing, as Hrd1 and Atx-3 have opposite activities, in ubiquitin ligation and de-ubiquitination respectively. It would be interesting to investigate whether, like gp78 and SVIP, a functional relationship exists between Hrd1 and Atx-3.


There are collectively over 40 p97/CDC48 adaptors identified to date. The ability of p97 to interact with so many different proteins has puzzled scientists for over a decade. In this review, we have discussed three protein domains, UBX, PUB and PUL, which are likely to serve as general p97-binding modules, and several p97-binding motifs that have been identified only recently. Structural and biochemical data strongly suggest that proteins possessing the same p97-binding domain interact with p97 via very similar modes, providing a partial explanation for the versatility of p97 in adaptor binding.

Intriguingly, many p97 adaptors contain more than one p97-binding site. The SHP box is present in a variety of UBX-containing adaptors, and a number of PUB proteins also possess a UBX domain. Furthermore, the WD40 repeat region on PUL proteins has been shown using immunoprecipitation to interact with CDC48. The presence of multiple binding modules on a single protein provides the possibility of adopting diverse interacting modes with p97 and adapting its various cellular functions.

A recently reported functional relationship between the VIM-containing proteins, gp78 and SVIP, has led us to hypothesizing that similar relationship also exists between the VBM-containing proteins, Atx-3 and Hrd1. Hrd1 mediates the ubiquitination of the polyQ tract protein, huntingtin [48], and is likely to also play a role in the ubiquitination of Atx-3. Although there is no evidence supporting this hypothesis at present, the presence of various p97-interacting motifs provide some hints to uncover the relationship between these seemingly unrelated p97 adaptors.

p97 is a central player in the ubiquitin signalling system and its participation in many cellular activities is dependent on its ability to associate with numerous adaptors. Studying the interactions between p97 and these proteins is fundamental to understanding its wide range of physiological roles. With the number of p97 adaptors still growing, the identification of further p97-interaction domains/motifs is likely. And this will no doubt bring us closer to unravelling the ever-expanding network that surrounds p97.


  • Seventh International Meeting on AAA Proteins: Independent Meeting held at the Royal Agricultural College, Cirencester, U.K., 9–13 September 2007. Organized and Edited by John Mayer (Nottingham, U.K.) and Paul Freemont (Imperial College London, U.K.).

Abbreviations: AAA, ATPase associated with various cellular activities; Atx-3, ataxin-3; CDC48, cell division cycle 48; ERAD, endoplasmic reticulum-associated degradation; gp78, glycoprotein 78; NC, C-terminus of N domain; NN, N-terminus of N domain; p97, C10, final ten C-terminal residues of p97; PLAP, phospholipase A2-activating protein; PNGase, peptide N-glycosidase; polyQ, polyglutamine; PUB, PNGase/ubiquitin-associated; PUL, PLAP, Ufd3p and Lub1p; UBD, ubiquitin D; UBX, ubiquitin regulatory X; UN, Ufd1–Npl4; VCP, valosin-containing protein; SVIP, small VCP-interacting protein; VBM, VCP-binding motif; VIM, VCP-interacting motif


View Abstract