RHBDL4 is an active rhomboid that specifically recognizes and cleaves atypical, positively charged transmembrane endoplasmic reticulum-associated degradation (ERAD) substrates. Interaction of valosin-containing protein (p97/VCP) and RHBDL4 is crucial to retrotranslocate polyubiquitinated substrates for ERAD pathway. Here, we report the first complex structure of VCP-binding motif (VBM) with p97 N-terminal domain (p97N) at 1.88 Å resolution. Consistent with p97 adaptor proteins including p47-ubiquitin regulatory X (UBX), gp78-VCP-interacting motif (VIM), OTU1-UBX-like element, and FAF1-UBX, RHBDL4 VBM also binds at the interface between the two lobes of p97N. Notably, the RF residues in VBM are involved in the interaction with p97N, showing a similar interaction pattern with that of FPR signature motif in the UBX domain, although the directionality is opposite. Comparison of VBM interaction with VIM of gp78, another α-helical motif that interacts with p97N, revealed that the helix direction is inversed. Nevertheless, the conserved arginine residues in both motifs participate in the majority of the interface via extensive hydrogen bonds and ionic interactions with p97N. We identified novel VBM-binding mode to p97N that involves a combination of two types of p97–cofactor specificities observed in the UBX and VIM interactions. This highlights the induced fit model of p97N interdomain cleft upon cofactor binding to form stable p97–cofactor complexes. Our mutational and biochemical analyses in defining the specific interaction between VBM and p97N have elucidated the importance of the highly conserved VBM, applicable to other VBM-containing proteins. We also showed that RHBDL4, ubiquitins, and p97 co-operate for efficient substrate dislocation.

Introduction

Transport of misfolded secretory proteins is tightly controlled by the endoplasmic reticulum-associated degradation (ERAD) pathway to ensure only properly folded proteins are transported to their site of action [1]. Perturbation of such a quality control system may lead to severe diseases such as diabetes and neurodegenerative disorders, including Alzheimer's disease, Parkinson's disease, and bipolar disorder [1]. In most cases, the dislocation process is coupled to the energy driven by the p97 cytosolic ATPase and ERAD factors containing a binding site for p97 [2,3].

The fact that the rhomboid-like family constitutes a superfamily of conserved proteins that specifically bind to substrate proteins and direct them into different dislocation pathways in ERAD has been studied extensively [46]. Recently, RHBDL4 was found to be an active rhomboid that resides in the ER and plays a pivotal role in the ERAD pathway. Particularly, the key catalytic S144 and H195 residues of RHBDL4 were able to cleave substrate within the membrane [7,8]. Fleig et al. found that RHBDL4 specifically cleaves single-spanning and polytopic membrane proteins within unstable transmembrane helices such as pre-T-cell receptor α (pTα). pTα is a type I transmembrane protein that is cleaved by RHBDL4 to form multiple N-terminal proteolytic fragments [7]. In addition, RHBDL4 was found to promote the cleavage, followed by subsequent proteasomal degradation of the apoptotic Bcl-2 family member, Bcl-2-interacting killer (Bik) [9]. Mutagenesis studies at G142 and S144 of RHBDL4 showed that these residues are crucial for its catalytic activity in cleaving Bik at a site located in the transmembrane region. Overexpression or knockdown of RHBDL4 in HEK 293T cells can reduce or enhance Bik-mediated apoptosis, respectively [9]. Wan et al. [10] also showed the importance of RHBDL4 in inducing the proteolysis of a tumor suppressor-activated pathway-6, which plays a role in the regulation of nonclassical exosomal secretion.

Beyond the proteolytic cleavage capability of RHBDL4, the C-terminal region of RHBDL4 contains two intriguing features, the p97/VCP-binding motif (VBM) and ubiquitin-interacting motif (UIM), which distinguish RHBDL4 from other known intramembrane proteases [7]. UIM can bind ubiquitin that helps to recognize ubiquitinated substrates specifically. Meanwhile, VBM can recruit p97 to generate the driving force for proteasomal degradation of the ubiquitinated substrate. This is consistent with the finding by Lemberg's group in that overexpression of an inactive protease mutant of RHBDL4 led to accumulation of ubiquitinated substrates, suggesting that ubiquitinated substrates need to be recognized by UIM in RHBDL4 prior to proteasomal degradation [7].

p97 is a ubiquitous, well-conserved type II AAA+-ATPase that plays crucial roles in the ERAD pathway. p97 is composed of the N-terminal domain (p97N), two ATPase ring domains (called D1 and D2), and a C-terminal tail, and forms homohexamer [11,12]. In particular, p97N is subdivided into two subdomains, known as p97N (N-terminus) and p97N (C-terminus) lobes. p97N recruits substrates or cofactors prior to driving the substrate toward proteasome by using energy catalyzed by D1 and D2 ring domains.

A large number of regulatory cofactors are required to enhance p97's role in controlling substrates specificity and turnover. All p97N-interacting proteins contain ubiquitin regulatory X (UBX), ubiquitin-binding domain, UBX-like element, p97/VCP-binding motif (VBM), p97/VCP-interacting motif (VIM), or SHP motif (also known as binding site 1 (BS1)) [1317]. In addition, proteasomal degradation of ER substrates driven by p97 requires recognition of ubiquitin at the substrate which is formed by a Lys 48-linked ubiquitin chain. Malfunction of p97 in recognizing ubiquitinated substrates has been found to block the retrotranslocation of the substrate in the ER. Such accumulation eventually leads to neurodegenerative diseases such as Alzheimer's disease and polyglutamine disease [18].

Currently, the structures of UBX or UBX-like domain complexed with p97 have been determined including FAF1-UBX, p47-UBX, and OTU1-UBX-like element (UBXL) [1921]. In addition, the complex structure of VIM of gp78 E3 ligase with p97N has been determined [22]. All p97 adaptor proteins aforementioned bind at the canonical binding site at the interface between the two lobes of p97N, except SHP motif, which binds at the noncanonical binding site [23]. However, the structural information for VBM interaction with p97 is unknown. In the present study, we present the first complex structure of VBM with p97N and propose the biological significance of the p97N–VBM interaction in recruiting ubiquitinated substrates to p97 in the ERAD process.

Results

RHBDL4 VBM binds into interdomain cleft of p97N

To identify the minimal region of RHBDL4 for p97N interaction, we constructed the RHBDL4 C-terminal region with various truncations. In vitro pull-down analyses revealed that the C5 construct (residues 299–315) containing VBM alone was sufficient for p97N binding, whereas C6 construct (residues 255–299) with VBM deletion abrogated p97 interaction (Figure 1A–C). Subsequently, we used a 15 amino acid synthesized peptide comprising an RHBDL4 VBM (residues 300–314) to produce a complex crystal with p97N. The crystal structure of p97N–VBM complex was determined at 1.88 Å resolution by the molecular replacement method using the p97N structure as a template (PDB ID 3QQ7). Clear density for VBM peptide was visible in the initial maps (mFo − DFc and 2mFo − DFc) obtained before ligand modeling (Supplementary Figure S1). The 2mFo − DFc map is a sigma-A weighted map, showing where the model should go, and the mFo – DFc map is a difference map which shows where the model is missing (where this map is positive) and where the model should not be but is (where it is negative). The complex structure was refined to Rwork of 17.7% (Rfree 21.0%; Table 1).

Overall structure of RHBDL4 VBM and p97N.

Figure 1.
Overall structure of RHBDL4 VBM and p97N.

(A) Schematic diagram of RHBDL4 and p97 constructs. Various truncates of RHBDL4 C-terminus with GST-tagged (top) and p97 constructs, including p97N and p97ND1. (B) GST-tag in vitro pull-down assay of p97ND1 using GST-RHBDL4c variants, C1–C6 constructs. (C) GST-tag in vitro pull-down assay of p97N using C5 construct in a concentration-dependent manner. GST protein served as a negative control. (D) Cartoon representation of the p97N domain (gray) and RHBDL4 VBM box (green) in normal view and 90° rotation. RHBDL4 VBM peptide is displayed in α-helix on top of the p97N structure.

Figure 1.
Overall structure of RHBDL4 VBM and p97N.

(A) Schematic diagram of RHBDL4 and p97 constructs. Various truncates of RHBDL4 C-terminus with GST-tagged (top) and p97 constructs, including p97N and p97ND1. (B) GST-tag in vitro pull-down assay of p97ND1 using GST-RHBDL4c variants, C1–C6 constructs. (C) GST-tag in vitro pull-down assay of p97N using C5 construct in a concentration-dependent manner. GST protein served as a negative control. (D) Cartoon representation of the p97N domain (gray) and RHBDL4 VBM box (green) in normal view and 90° rotation. RHBDL4 VBM peptide is displayed in α-helix on top of the p97N structure.

Table 1
X-ray data collection and refinement statistics
Data collection p97N–VBM peptide 
X-ray source PAL-5C 
Wavelength (Å) 0.9796 
Resolution (Å) 50−1.88 (1.91−1.88) 
Space group P65 
Unit cell parameters (Å) a = 97.0, b = 97.0, c = 50.9
α = 90°, β = 90°, γ = 120° 
No. of observed reflections 1 267 454 
No. of unique reflections 22 279 
Completeness (%) 99.4 (99.4) 
R1merge (%) 7.3 (56.5) 
CC1/22 in outer shell (%) 87.0 
Mean I/σ(I8.2 (3.2) 
Multiplicity 5.8 (6.3) 
Refinement statistics 
 Resolution (Å) 42.0−1.88 
Rwork/Rfree3 (%) 17.7/21.0 
 No. of atoms 
  p97N 1354/170 
  VBM peptide 135/15 (aa) 
  Water 127 
 Average B-factors (Å2
  p97N 20.0 
  VBM peptide 18.2 
  Water 25.7 
 RMS deviation from ideal geometry 
  Bond length (Å) 0.008 
  Bond angle (°) 1.330 
 Ramachandran plot 
  Favored regions (%) 98.9 
  Allowed regions (%) 1.1 
PDB code 5EPP 
Data collection p97N–VBM peptide 
X-ray source PAL-5C 
Wavelength (Å) 0.9796 
Resolution (Å) 50−1.88 (1.91−1.88) 
Space group P65 
Unit cell parameters (Å) a = 97.0, b = 97.0, c = 50.9
α = 90°, β = 90°, γ = 120° 
No. of observed reflections 1 267 454 
No. of unique reflections 22 279 
Completeness (%) 99.4 (99.4) 
R1merge (%) 7.3 (56.5) 
CC1/22 in outer shell (%) 87.0 
Mean I/σ(I8.2 (3.2) 
Multiplicity 5.8 (6.3) 
Refinement statistics 
 Resolution (Å) 42.0−1.88 
Rwork/Rfree3 (%) 17.7/21.0 
 No. of atoms 
  p97N 1354/170 
  VBM peptide 135/15 (aa) 
  Water 127 
 Average B-factors (Å2
  p97N 20.0 
  VBM peptide 18.2 
  Water 25.7 
 RMS deviation from ideal geometry 
  Bond length (Å) 0.008 
  Bond angle (°) 1.330 
 Ramachandran plot 
  Favored regions (%) 98.9 
  Allowed regions (%) 1.1 
PDB code 5EPP 

Data collection statistics are from HKL2000; refinement statistics are from PHENIX [31] and CCP4 [39].

*

, where I(h) is the intensity of reflection of h, ∑h is the sum over all reflections and ∑i is the sum over i measurements of reflection h.

CC1/2 in the outer shell was calculated from HKL2000 [36].

, where Rfree is the R value calculated for 5% of the data set not included in the refinement.

The structure of the p97N in RHBDL4 VBM complex shows high structural similarity with its apo structure (PDB ID 3QQ7) with the root-mean-square deviation (RMSD) of 0.72 Å for Cα atoms of 167 residues in p97N. Nevertheless, conformational changes at p97N induced by VBM binding are obvious, including R53, I70, and E141 (Supplementary Figure S2). The structure of VBM encompassing residues 300–314 of RHBDL4 adopts an α-helix, which interacts with p97N at the interdomain cleft of Nn and Nc lobes (Figure 1D). Similar to all known p97 cofactors such as UBX domain of FAF1 (PDB ID 3QQ8), p47 (PDB ID 1S3S), and OTU1 (PDB ID 4KDL), and VIM (PDB ID 3TIW) of gp78 E3 ligase, which bind into the p97N interdomain cleft, VBM also binds into this cleft. The total surface area of the interface between p97N and VBM has been calculated to be 650.8 Å using the PDBePISA program [24], which is a characteristic value for interfaces of p97N and cofactors (Supplementary Figure S3). The interface of VBM and p97N interaction mainly composed of hydrogen bonding across the α-helix in VBM and hydrophobic interactions at the C-terminus of the helix as analyzed in Ligplot+ software [25] (Figure 2A and Supplementary Figure S3). The binding affinity between p97N and RHBDL4 C3 (residues 255–315) was measured by surface plasmon resonance (SPR) analysis with dissociated constant (Kd) of 68.9 ± 2.4 µM, which is compatible with other VBM-containing proteins, including Hrd1, E4B, and Atx3 [26] (Figure 2B and Supplementary Table S1).

The main chain of S300 at the very N-terminus of VBM forms hydrogen bonds with Y110 of p97N, along with a water molecule which further stabilizes both the residues. On the contrary, R305 side chain of VBM stretches out to the Nn lobe, and interacts with the main chain carbonyl oxygen of R53 and side chain of D55 of p97N through hydrogen bonding and salt bridge, respectively. R308, one of the most conserved residues of VBM, stretches downward into the binding cleft of the p97N interdomain, forming extensive hydrogen bonding with main chain carbonyl oxygens of G54 and Y143 and hydroxyl group of T56 of p97N. This arrangement therefore retains a pentagonal planar co-ordination of R308 essential for maintaining the RHBDL4 VBM–p97N complex. In contrast, R311 residues at the C-terminus of the VBM helix extend toward the Nc side and form hydrogen bonds with the main chain carboxyl oxygens of P137, Y138, and L140 of p97N. In addition, the D313 side chain of VBM exhibits ionic interaction and forms water-mediated hydrogen bonds with R53 and N33 of p97N, which stabilize the C-terminus of the VBM helix interaction with p97N. On the contrary, hydrophobic interactions are mainly derived from VBM C-terminus via the interaction between L309 of VBM and L72 of p97N. Moreover, the aromatic ring of F312 of VBM is positioned to be deeply buried into the cleft formed by Nn and Nc, enabling hydrophobic interactions with V38, I70, L72, and A142 residues of p97N. The central region of VBM α-helix (M304, Q307, and R308) was further stabilized by hydrophobic interaction formed by E141 (Cβ and Cγ), A142, and Y143 residues of p97N.

Multiple sequence alignment of RHBDL4 VBMs has delineated the conservation of VBM throughout the family, from higher eukaryotes to lower eukaryotes including algae and fungi (Figure 2C). To evaluate the molecular determinants that govern the unique selectivity of p97N adaptor proteins such as RHBDL4, we performed site-directed mutagenesis on the conserved interacting residues identified in the p97N and RHBDL4 VBM consensus sequence. In particular, replacement of arginine-rich residues in the VBM (R305, R308, and R311) with alanine or aspartate revealed the complete abolishment of binding to p97ND1 (Figure 2D). Other single mutations of conserved residues including L309A, F312A, and D313A of VBM also caused abrogation of p97ND1 recruitment. Consistent with structural analysis, the consensus sequence of RRhRL-RF motif in VBM is pivotal in maintaining a stable p97N–VBM complex, which also highlights the importance of this minimal signature motif for p97N recognition.

Detailed interaction of p97N in complex with RHBDL4 VBM.

Figure 2.
Detailed interaction of p97N in complex with RHBDL4 VBM.

(A) Stereo view of detailed interacting residues in p97N interdomain-binding groove in 25° rotation. RHBDL4 VBM peptide (green) is displayed in α-strand and stick representation, whereas p97N-interacting residues are in gray in stick form. Dashed lines indicate H bonds. Water molecule is highlighted in red color. (B) SPR assay of p97N–VBM interaction. The p97N protein was immobilized on the surface of CM5 sensor chip (GE Healthcare) using amine coupling. The RHBDL4 (C3) protein was used in a concentration-dependent manner at 30, 60, 90, 120, 150, 180, 210, 270, and 300 µM, and is represented in various colors in a sensorgram. The kinetics of p97N–VBM interaction (Kd = 68.9 ± 2.4 µM) were calculated from three independent experiments. (C) Multiple sequence alignment of the RHBDL4 VBMs. (D) GST-tag in vitro pull-down assay of p97ND1 using wild-type and mutant variants of RHBDL4_267−315 (C4 construct). GST served as a negative control.

Figure 2.
Detailed interaction of p97N in complex with RHBDL4 VBM.

(A) Stereo view of detailed interacting residues in p97N interdomain-binding groove in 25° rotation. RHBDL4 VBM peptide (green) is displayed in α-strand and stick representation, whereas p97N-interacting residues are in gray in stick form. Dashed lines indicate H bonds. Water molecule is highlighted in red color. (B) SPR assay of p97N–VBM interaction. The p97N protein was immobilized on the surface of CM5 sensor chip (GE Healthcare) using amine coupling. The RHBDL4 (C3) protein was used in a concentration-dependent manner at 30, 60, 90, 120, 150, 180, 210, 270, and 300 µM, and is represented in various colors in a sensorgram. The kinetics of p97N–VBM interaction (Kd = 68.9 ± 2.4 µM) were calculated from three independent experiments. (C) Multiple sequence alignment of the RHBDL4 VBMs. (D) GST-tag in vitro pull-down assay of p97ND1 using wild-type and mutant variants of RHBDL4_267−315 (C4 construct). GST served as a negative control.

Comparison of p97N–VBM (RHBDL4) and p97N–VIM (gp78) interactions

The overall spatial arrangement of VBM (RHBDL4) with respect to p97N is highly analogous to those of previously reported VIM (gp78) (PDB ID 3TIW) (Figure 3A,B). However, surprisingly, the directionality of the α-helices is opposite in two structures. The RMSD of p97N is calculated to be 1.03 Å for Cα atoms of 165 residues, whereas that of gp78 VIM and RHBDL4 VBM shows 2.54 Å for Cα atoms of 13 residues in the VBM and VIM helices. This implies the obvious structural differences between VBM and VIM when binding into p97N interdomain cleft. Highly conserved basic residues in VBMs (RRRRL) and VIMs (RX5AAX2R) are important to maintain the p97N interaction by contributing a majority of the ionic and hydrogen bond interactions in the interface (Figure 3A,C). R305 of VBM is located at a position similar to that of the conserved R626 of VIM, with its side chain protruding toward the Nn lobe of the interdomain groove. R305 of VBM is stabilized by a salt bridge with carboxyl side chain of D55 of p97N, whereas R626 of VIM hydrogen bonds to backbone carbonyl oxygen of R53 of p97N. Highly conserved R308 positioned at the center of helix in VBM bends downward to the p97-binding pocket and forming extensive hydrogen bonding through a pentagonal coordination. In contrast, R625 of VIM at the very N-terminus of the helix also protrudes into the p97N interdomain cleft and interacts with the V108 residue of p97N through hydrogen bonding. In addition, R311 of VBM and R636 of VIM are located at the C-terminal end of the helix and stretch their side chains toward the Nc lobe. In this case, R311 of VBM forms hydrogen bonding with P137, Y138, and L140 of Nc, whereas R636 of VIM is stabilized by a combination of salt bridge and hydrogen bond formation by D35 and A142, respectively. Consistent with the sequence alignment result, conserved positively charged residues in both the VBM and VIM exhibit a majority of interaction for p97 recognition.

Despite their close resemblance at the arginine-rich region, other similarities were noted between amino acids of VIM and VBM, located at the same position though their directionality was inverted. This indicates specificity of this interdomain cleft to recognize adaptor proteins. The M304 of VBM (M628 of VIM) makes hydrophobic interaction with Y143 of p97N (Figure 3C). Additionally, the conserved leucine residues of VBM (L309) were mapped in with conserved alanine residues in VIM (A630), buried in the hydrophobic pocket formed by G54 and L72 of p97N. In addition, F312 of VBM was found to be located at the exact position as A633 of VIM, emphasizing the importance of hydrophobic residue to be sheltered in the hydrophobic environment formed by p97N interdomain cleft (V38, I70, L72, and A142). Apart from hydrophobic residues, D313 of VBM corresponds to E634 of VIM, forming hydrogen bonds with R53 side chain of p97.

Comparison of p97N-binding motifs in RHBDL4 VBM and gp78 VIM complex.

Figure 3.
Comparison of p97N-binding motifs in RHBDL4 VBM and gp78 VIM complex.

(A) Multiple sequence alignment of the VBM- and VIM-containing proteins. (B) Surface and cartoon representation of overlaid p97N–RHBDL4 VBM (green) (PDB ID 5EPP) and p97N–gp78 VIM (orange) structure (PDB ID 3TIW). VIM and VBM helices were superimposed with a deviation angle of 12°. (C) Stereo view representation of overlay detailed interactions of p97N in complex with RHBDL4 VBM (green) and gp78 VIM (orange) stick form.

Figure 3.
Comparison of p97N-binding motifs in RHBDL4 VBM and gp78 VIM complex.

(A) Multiple sequence alignment of the VBM- and VIM-containing proteins. (B) Surface and cartoon representation of overlaid p97N–RHBDL4 VBM (green) (PDB ID 5EPP) and p97N–gp78 VIM (orange) structure (PDB ID 3TIW). VIM and VBM helices were superimposed with a deviation angle of 12°. (C) Stereo view representation of overlay detailed interactions of p97N in complex with RHBDL4 VBM (green) and gp78 VIM (orange) stick form.

Comparison of RF motif in RHBDL4 VBM and FPR motif in FAF1-UBX in p97N interaction

UBX constitutes the largest group of p97 adaptor proteins, and the majority of interaction with p97N involves conserved residues in the S3/S4 loop of UBX, so called FPR signature motif, as exemplified in FAF1-UBX (PDB ID 3QQ8) [20,27]. Structure-based multiple sequence alignment of UBX-containing protein using T-coffee server [28] and ESPript program [29] has revealed a highly conserved FPR motif (Supplementary Figure S4). In particular, the first residue in the FPR motif (a highly conserved phenylalanine or hydrophobic residue such as valine, leucine, or tyrosine) occupies the hydrophobic pocket (Figure 4C). The second residue is proline, absolutely conserved throughout the UBX family, which makes a tight turn in S3/S4 loop. The residue at the third position is less conserved with basic residues, probably because it forms hydrogen bonds with two backbone carbonyl oxygens.

Comparison of p97N in complex with RHBDL4 VBM and FAF1-UBX complex.

Figure 4.
Comparison of p97N in complex with RHBDL4 VBM and FAF1-UBX complex.

(A) Overall view of p97N (dark gray) in complex with FAF1-UBX (pink) (PDB ID 3QQ8) and p97N (light gray) in complex with RHBDL4 VBM (green). (B) Close view in ribbon and stick representation of p97N–cofactor-binding groove towards UBX signature loop and VBM α-helix. UBX ‘FPR’ motif and VBM ‘RF’ motif of inverse directionality are labeled and displayed in stick form. RMSDs were calculated using CCP4 suite. (C) Stereo view representation of p97N and cofactor-binding groove and the interacting residues in stick form.

Figure 4.
Comparison of p97N in complex with RHBDL4 VBM and FAF1-UBX complex.

(A) Overall view of p97N (dark gray) in complex with FAF1-UBX (pink) (PDB ID 3QQ8) and p97N (light gray) in complex with RHBDL4 VBM (green). (B) Close view in ribbon and stick representation of p97N–cofactor-binding groove towards UBX signature loop and VBM α-helix. UBX ‘FPR’ motif and VBM ‘RF’ motif of inverse directionality are labeled and displayed in stick form. RMSDs were calculated using CCP4 suite. (C) Stereo view representation of p97N and cofactor-binding groove and the interacting residues in stick form.

We investigated whether UBX S3/S4 loop interaction with p97N is conserved in RHBDL4 VBM. Superimposition of p97N–RHBDL4 VBM complex and other UBX-containing proteins, including FAF1-UBX and OTU1-UBXL, has revealed a considerably low average Cα RMSD of 0.76 and 0.83 Å, respectively, for the corresponding 167 and 168 residue-long p97N superimposing region (Figures 4A and 5A). Intriguingly, the RF motif (residues 311–312) of RHBDL4 VBM is located at the same position as the FAF1-UBX (PDB ID 3QQ8) FPR motif (residues 619–621; Figure 4B) with inverse directionality. Highly conserved FPR motif that constitutes the S3/S4 loop reveals the most critical interaction in p97N recognition of UBX-containing proteins, which occupies ∼27% of interface area, that is 222.1 Å2 out of total UBX and p97N interface area of 817.6 Å2, as calculated using PDBePISA [24]. Mutagenesis study of FAF1-UBX FPR motif to SGG or AG has found to completely abrogate its interaction with p97N, causing its functional failure to recruit polyubiquitinated substrates to the N-terminal UBA domain of FAF1; however, the direct evidence of UBA regulation by UBX domain in FAF1 remains to be known [20,30]. In particular, the aromatic side chain of F312 in the RF motif of RHBDL4 VBM, resembling F619 on UBX, inserts into a hydrophobic pocket created by the Nn and Nc lobes of p97, forming extensive hydrophobic interaction with D35, V38, I70, and A142 of p97N (Figure 4C). P620 on FAF1-UBX adopts the cis-isomer form and β-turn structure that makes FPR motif buried into the p97N interdomain groove [20]. In contrast, the corresponding residue in RHBDL4 VBM is missing. Instead, the local structure of the α-helix of VBM makes it a structure with kinked backbone causing the RF residues to overlap to the position of F619 and R621 in FPR motif FAF1-UBX. In addition to these hydrophobic interactions, R311 in RF motif of RHBDL4 VBM corresponding to the R621 of FAF1-UBX forms several hydrogen bonds with the carbonyl oxygen atoms of P137 and L140 on p97N, which apparently stabilizes the VBM α-helix as seen in the S3/S4 loop FAF1-UBX (Figure 4C). Interestingly, despite the inverse directionality in VBM helix compared with the S3/S4 loop of UBX, the highly conserved arginine and phenylalanine residues in VBM and UBX show the same pattern in their interaction with p97N.

Comparison of p97N dynamics in apo and cofactors' complexes.

Figure 5.
Comparison of p97N dynamics in apo and cofactors' complexes.

(A) Superposition of the p97N domain in complex with RHBDL4 VBM (green), VIM (PDB ID 3TIW, colored in orange), UBX domain of FAF1 (PDB ID 3QQ8, colored in pink), UBXL of OTU1 (PDB ID 4KDL, colored in blue). (B) Ensemble model of apo-p97N in RMSF is indicated as blue to green. Area I (VBM, VIM, and UBX-binding pocket) is shown in green, red, and orange, respectively; areas II (SHP-binding pocket) and III (D1 ring interface) are shown in gray. (C) Top view of (B) around area I. L2, L4, and L12 loops are highlighted in green, red, and orange, respectively. (D) Plots for RMSF difference (complex apo) obtained from RMSFs of ensemble refinement for apo (black) and adaptor-bound crystal structures [p97N–VBM (green; PDB ID 5EPP), p97N–VIM (orange; PDB ID 3TIW), p97N–UBX (pink; PDB ID 3QQ8), and p97N–UBXL (blue; PDB ID 4KDL)]. L2, L4, and L12 loops of interface area I are highlighted in green, red, and orange, respectively, and marked with red stars. Gray area represents other flexible loop regions of p97N, and marked with black stars. Negative RMSF difference value indicates more rigidity of p97N upon cofactor binding.

Figure 5.
Comparison of p97N dynamics in apo and cofactors' complexes.

(A) Superposition of the p97N domain in complex with RHBDL4 VBM (green), VIM (PDB ID 3TIW, colored in orange), UBX domain of FAF1 (PDB ID 3QQ8, colored in pink), UBXL of OTU1 (PDB ID 4KDL, colored in blue). (B) Ensemble model of apo-p97N in RMSF is indicated as blue to green. Area I (VBM, VIM, and UBX-binding pocket) is shown in green, red, and orange, respectively; areas II (SHP-binding pocket) and III (D1 ring interface) are shown in gray. (C) Top view of (B) around area I. L2, L4, and L12 loops are highlighted in green, red, and orange, respectively. (D) Plots for RMSF difference (complex apo) obtained from RMSFs of ensemble refinement for apo (black) and adaptor-bound crystal structures [p97N–VBM (green; PDB ID 5EPP), p97N–VIM (orange; PDB ID 3TIW), p97N–UBX (pink; PDB ID 3QQ8), and p97N–UBXL (blue; PDB ID 4KDL)]. L2, L4, and L12 loops of interface area I are highlighted in green, red, and orange, respectively, and marked with red stars. Gray area represents other flexible loop regions of p97N, and marked with black stars. Negative RMSF difference value indicates more rigidity of p97N upon cofactor binding.

Comparison of p97N dynamics in apo and cofactors' complexes

To evaluate the effects of the dynamics of p97N in cofactor binding, we performed ensemble refinement using Phenix [31]. Although the ensemble models revealed that p97N adopts an overall stable structure, we observed relatively dynamic fluctuations in three sites of p97N in apo (Figure 5B,C). When we superimposed the apo structure with complex structures, we noted that area I was involved in the cofactor-binding interfaces for VBM, VIM, and UBX. Area II was involved with another type of cofactor, SHP motif found in Ufd1 (PDB ID 5C1B) [23]. Meanwhile, area III was involved in the D1 ring domain interface in full-length p97 (Figure 5B).

To see the changes in dynamics occurring after binding of cofactors, we performed the root-means-square fluctuation (RMSF) analysis for apo and cofactor complexes (Figure 5D). Most of the changes in dynamics were observed in the loop regions of area I. The results suggest that cofactor binding reduces flexibility and introduces more rigidity to the p97N interdomain interface (Figure 5D, and Supplementary Figure S5 and Video S1). Loop region in area I showed higher fluctuation than the β-sheet or α-helix of apo-p97N structure. Flexible L2, L4, and L12 loops located in the interdomain cleft (area I) turned into rigid structure upon cofactor binding, which can be attributed to the stabilization effect of interactions between p97N and cofactors. It is interesting to note that L11 and L14 loops in p97N are also rigidified in the VBM complex structure (Figure 5D). Since L11 and L14 loops are involved in D1 ring interaction, it will be interesting to further analyze the allosteric effect of VBM interaction on the conformational changes in p97 structure [32]. Collectively, we found that flexible regions in apo-p97N are involved in the interaction with cofactors, and cofactor binding rigidifies the flexible loops in the VBM complex.

RHBDL4 assembles into a ternary complex with p97 and ubiquitin

To examine ternary complex formation of RHBDL4, p97, and ubiquitinated substrate, we performed in vitro cross-linking assay using p97N, RHBDL4 C-terminus (C3 construct containing both UIM and VBM), and ubiquitins (mono- and di-) to confirm the formation of a binary complex between p97N and ubiquitins, as well as ternary complex between RHBDL4 C-terminus, p97N, and ubiquitins (Figure 6A). RHBDL4 C3 construct was used, because we could successfully purify the protein after cleavage of GST-tag. Additionally, we used monomeric p97N instead of hexameric p97, because the complex cross-linked products obtained made the cross-linking data with p97 difficult to interpret. We confirmed that RHBDL4 C3 forms binary complexes with p97N via the VBM and ubiquitins via the UIM. In addition, we found noble binary interactions between p97N and ubiquitins, as well as the existence of a ternary complex between p97N and ubiquitins bridges via RHBDL4 C3.

Ternary complex assembly of RHBDL4 C-terminus, p97, and ubiquitins.

Figure 6.
Ternary complex assembly of RHBDL4 C-terminus, p97, and ubiquitins.

(A) Treatment of RHBDL4 C3, p97N, and mono-Ub/di-Ub with DSS (lanes 2–5) indicates individual oligomerization status. Binary cross-linked proteins (C3/mono-Ub, C3/di-Ub, C3/p97N, p97N/mono-Ub, and p97N/di-Ub) are circled in blue and marked as numbers 1–4 and 6. Ternary cross-linked p97N/C3/mono-Ub and p97N/C3/di-Ub are marked in red circles and denoted as 5 and 7, respectively. Schematic diagrams of binary and ternary cross-linked products are displayed and molecular weight of individual proteins is marked. (B) GST in vitro pull-down assay of RHBDL4 C5, p97, and ubiquitins (mono-Ub, di-Ub, and tri-Ub). RHBDL4 C5 construct contains only VBM. Blue area indicates RHBDL4 C5 pull-down of p97N, followed by mono-Ub, di-Ub, and tri-Ub, designated as 1, 2, and 3, respectively. Red area shows RHBDL4 C5 pull-down of hexameric p97, followed by mono-Ub, di-Ub, and tri-Ub, designated as 1, 2, and 3, respectively. Schematic diagrams of binary complex of RHDBL4 C5/p97N and ternary complex of RHDBL4 C5/p97/ubiquitins are shown. (C) GST in vitro pull-down assay of RHBDL4 C4, p97, and ubiquitins (mono-Ub, di-Ub, and tri-Ub). RHBDL4 C4 construct contains both UIM and VBM. Blue area indicates RHBDL4 C4 pull-down of p97N, followed by mono-Ub, di-Ub, and tri-Ub designated as 1, 2, and 3, respectively. Red area shows RHBDL4 C4 pull-down of hexameric p97, followed by mono-Ub, di-Ub, and tri-Ub, designated as 1, 2, and 3, respectively. Schematic diagrams of ternary complex of RHDBL4 C4/p97N/ubiquitins and RHDBL4 C4/p97/ubiquitins are shown.

Figure 6.
Ternary complex assembly of RHBDL4 C-terminus, p97, and ubiquitins.

(A) Treatment of RHBDL4 C3, p97N, and mono-Ub/di-Ub with DSS (lanes 2–5) indicates individual oligomerization status. Binary cross-linked proteins (C3/mono-Ub, C3/di-Ub, C3/p97N, p97N/mono-Ub, and p97N/di-Ub) are circled in blue and marked as numbers 1–4 and 6. Ternary cross-linked p97N/C3/mono-Ub and p97N/C3/di-Ub are marked in red circles and denoted as 5 and 7, respectively. Schematic diagrams of binary and ternary cross-linked products are displayed and molecular weight of individual proteins is marked. (B) GST in vitro pull-down assay of RHBDL4 C5, p97, and ubiquitins (mono-Ub, di-Ub, and tri-Ub). RHBDL4 C5 construct contains only VBM. Blue area indicates RHBDL4 C5 pull-down of p97N, followed by mono-Ub, di-Ub, and tri-Ub, designated as 1, 2, and 3, respectively. Red area shows RHBDL4 C5 pull-down of hexameric p97, followed by mono-Ub, di-Ub, and tri-Ub, designated as 1, 2, and 3, respectively. Schematic diagrams of binary complex of RHDBL4 C5/p97N and ternary complex of RHDBL4 C5/p97/ubiquitins are shown. (C) GST in vitro pull-down assay of RHBDL4 C4, p97, and ubiquitins (mono-Ub, di-Ub, and tri-Ub). RHBDL4 C4 construct contains both UIM and VBM. Blue area indicates RHBDL4 C4 pull-down of p97N, followed by mono-Ub, di-Ub, and tri-Ub designated as 1, 2, and 3, respectively. Red area shows RHBDL4 C4 pull-down of hexameric p97, followed by mono-Ub, di-Ub, and tri-Ub, designated as 1, 2, and 3, respectively. Schematic diagrams of ternary complex of RHDBL4 C4/p97N/ubiquitins and RHDBL4 C4/p97/ubiquitins are shown.

To further examine ternary complex formation of RHBDL4, hexameric p97, and ubiquitins, we performed an in vitro pull-down experiment. RHBDL4 C5 (containing VBM alone) and C4 (containing both UIM and VBM) were chosen to evaluate their effects on ternary pull-down (Figure 6B,C). Mono- (mono-Ub), di- (di-Ub), and tri-ubiquitins (tri-Ub) were used in the assay to examine the optimal length required for stable pull-down of the ternary complex. Our results show that RHBDL4 C5 (VBM) successfully pulls down hexameric p97, together with di-Ub or tri-Ub, forming a ternary complex. Whereas RHBDL4 C5 pulls down monomeric p97N without ubiquitins, forming a binary complex only. This is consistent with the cross-linking data (Figure 6A). This result implies the potential binding of ubiquitins to the adjacent unoccupied p97N domain in hexameric p97 (Figure 6B). Mono-Ub was not detected probably because of the weak interaction between mono-Ub and p97N. Direct interaction of p97 and ubiquitin has also been reported by Meyer et al. [33] and Dai and Li [34]. When we used RHBDL4 C4 (containing both UIM and VBM), we detected a ternary complex with ubiquitins (di-Ub and tri-Ub) and hexameric p97 as previously shown in Figure 6B (Figure 6C). In addition, we observed ternary complex formation with ubiquitins (di-Ub and tri-Ub) and monomeric p97N with the cross-linking result. It is tempting to propose that RHBDL4 forms closed ternary complex with p97 and polyubiquitinated substrate, because the ubiquitin molecule can interact with UIM and p97N (Figure 7).

Schematic model of RHBDL4 co-operates with p97 in retrotranslocation of substrate (ex. aberrant pTα).

Figure 7.
Schematic model of RHBDL4 co-operates with p97 in retrotranslocation of substrate (ex. aberrant pTα).

Graphic illustration of the bipartite module of adaptor binding to p97N to assist substrate movement with arginine collar-mediated denaturation, which has been postulated to promote substrate unfolding. Model showing the recognition of ubiquitinated substrate by RHBDL4 through UIM and serine histidine dyad active site. The step is followed by cleavage of aberrant substrate and recruitment of p97 simultaneously. The binding of polyubiquitin chain to unoccupied adjacent subunit of p97N sequentially enhances substrate-binding affinity to p97, which induces the release of RHBDL4 from anchoring to p97. Such co-operation in p97 subunits help to drive the substrate for proteasomal degradation. Key residues of serine and histidine in RHBDL4 active site are labeled in red stars. Interaction between subunit with the binding affinity in the milimolar range is highlighted in pink and micromolar range is highlighted in blue.

Figure 7.
Schematic model of RHBDL4 co-operates with p97 in retrotranslocation of substrate (ex. aberrant pTα).

Graphic illustration of the bipartite module of adaptor binding to p97N to assist substrate movement with arginine collar-mediated denaturation, which has been postulated to promote substrate unfolding. Model showing the recognition of ubiquitinated substrate by RHBDL4 through UIM and serine histidine dyad active site. The step is followed by cleavage of aberrant substrate and recruitment of p97 simultaneously. The binding of polyubiquitin chain to unoccupied adjacent subunit of p97N sequentially enhances substrate-binding affinity to p97, which induces the release of RHBDL4 from anchoring to p97. Such co-operation in p97 subunits help to drive the substrate for proteasomal degradation. Key residues of serine and histidine in RHBDL4 active site are labeled in red stars. Interaction between subunit with the binding affinity in the milimolar range is highlighted in pink and micromolar range is highlighted in blue.

Discussion

In the RHBDL4-mediated ERAD pathway, the cleavage of substrates comprising transmembrane domains in the ER membrane is accompanied by the energy generated by p97 to destine the substrate for proteasomal degradation. In the present study, we elucidated the crystal structure and interaction of RHBDL4 VBM. Our work demonstrated the novel binding mode of RHBDL4 VBM to p97N, which comprised the combination of two types of p97–cofactor specificity observed in UBX and VIM interactions. Intriguingly, the RF motif in VBM corresponded to the FPR motif in UBX, and the RRR motif in VBM corresponded to RRR motif in gp78 VIM clearly defined its uniqueness towards p97N recognition. Comparison of RMSF difference among four cofactor-bound p97N complexes with apo-p97N revealed reduced flexibility in the p97N–cofactor complex, which can be attributed to the stabilization effect of interaction between p97N and cofactors. Our data analysis of p97 and RHBDL4 structure provided an explanation as to how the binding pocket formed in Nn and Nc lobes provided a sterically unopposed interface for the interaction of the various p97 adaptor proteins.

Based on the pull-down experiments showing the coexistence of this ternary complex of RHBDL4, p97, and ubiquitins, we proposed a model of the closed ternary complex formation among RHBDL4, p97, and polyubiquitinated substrate (Figure 7). The step starts with ubiquitinated aberrant transmembrane substrate pTα recognized by RHBDL4 in which the active site recognizes exposing helix-destabilizing residues in the substrate transmembrane, as well as the C-terminus UIM binds to ubiquitin of the ubiquitinated substrate (Kd of ∼0.2−2 mM) [35]. Once the intramembrane proteolysis occurred, RHBDL4 recruits p97 via a small VBM at the C-terminus (Kd of 68.9 ± 2.4 µM). The recruitment of p97 serves as a scaffolding platform that allows the polyubiquitinated chain of cleaved substrate to bind to the unoccupied adjacent subunit of hexameric p97 to form a closed ternary complex. The co-operation of polyubiquitin binding induces stronger attachment towards p97 ATPase, which competes with RHBDL4's interaction for p97N, resulting in the release of RHBDL4 from p97 as well as ubiquitinated substrates that eventually promote efficient transport of cleaved ER substrates for proteasomal degradation. Taken together, our work provides clues in understanding the interplay of RHBDL4 to pass over the ubiquitinated substrate to p97 for dislocation.

Materials and methods

Preparation of p97 recombinant proteins

Human p97N (21–199), p97ND1 (1–480), and p97 full length (1–806) were cloned into the pET28a vector (Novagen) comprising of the N-terminal hexahistidine tag and the tobacco etch virus (TEV) site to the target protein. The EC number of p97 is 3.6.4.6. Human recombinant constructs were transformed into Escherichia coli (E. coli) strain BL21 (DE3) and the cells were grown in Luria-Bertani (LB) medium containing 50 µg/ml kanamycin at 37°C until OD600 reached ∼0.6. Target proteins were induced by adding of 0.5 mM isopropyl β-d-1-thiogalactopyranoside (IPTG) and grown for 20 h at 20°C. Induced cells were harvested by centrifugation at 5000 g for 15 min at 4°C. Cell pellet was then resuspended in lysis buffer A [50 mM NaH2PO4 (pH 7.5), 300 mM NaCl and 5 mM imidazole] and disrupted through sonication. After removal of cell debris by centrifugation at 16 000 g for 40 min, supernatant containing soluble proteins was loaded onto an equilibrated gravity-flow open column (Bio-Rad, Hercules, CA) packed with Ni-IDA agarose resin (Elpis, Daejon, South Korea). The Ni-IDA column was subsequently washed with buffer A containing 5–35 mM imidazole and target protein was eluted with buffer A containing 300 mM imidazole. Eluted target protein was then concentrated using an Amicon Ultra-15 10K (Milipore, Merck, Germany) ultrafiltration unit and incubated with TEV protease overnight at 4°C to remove the hexahistidine tag at the N-terminus. Cleaved protein was dialyzed and a second Ni-IDA affinity chromatography was performed to remove the hexahistidine tag and uncleaved proteins. Target proteins were further purified through size-exclusion chromatography on a HiLoad 16/60 Superdex 75 column (GE Healthcare Life Science, Uppsala, Sweden), which was pre-equilibrated with buffer B [20 mM Tris–HCl (pH 7.5), 150 mM KCl, and 4 mM dithiothreitol (DTT)]. Final purified proteins were concentrated and stored at −80°C. Target proteins were confirmed through sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE).

Preparation of RHBDL4 recombinant proteins

RHBDL4 VBM peptide (SPEEMRRQRLHRFDS without N- and C-terminal modifications) purchased from AnyGen Co., Ltd. (South Korea) was used for crystallization. The concentration of peptide was calculated by measuring the weight of dried peptide (10 mg). For in vitro pull-down assay, various RHBDL4 C-terminal regions including 219–315 (C1), 229–315 (C2), 255–315 (C3), 267–315 (C4), 299–315 (C5), 255–299 (C6), and C4 mutants were cloned into modified pET28a vector containing a GST-tag at the N-terminal region, followed by a TEV site and target protein. RHBDL4 gene (EC number: 3.4.21.105) was amplified from human cDNA that was obtained from Korea Cell Line Bank (KCLB) strain YD-10B (KCLB, Seoul, South Korea). Recombinant constructs were expressed in E. coli strain BL21 (DE3), and the cells were grown in LB medium containing 50 µg/ml kanamycin at 37°C until OD600 reached ∼0.6. Target proteins were induced by adding 0.5 mM IPTG and incubated for 20 h at 20°C. Harvested cell were lysed and passed through GST-bind agarose (Elpis, Daejon, South Korea) affinity chromatography. Target proteins were further purified through a size-exclusion chromatography on a HiLoad 16/60 Superdex 75 column (GE Healthcare Life Science, Uppsala, Sweden), which were pre-equilibrated with buffer B [20 mM Tris–HCl (pH 7.5) and 150 mM KCl]. Final purified proteins were concentrated and stored at −80°C. The purity of target proteins was confirmed through SDS–PAGE.

Preparation of ubiquitin, E1, and E2-25K proteins

For ubiquitin purification, ubiquitin was cloned into pET28b vector with His-tagging at N-terminal region with a thrombin cleavage site. A recombinant construct was expressed and purified with Ni-IDA affinity chromatography as mentioned above. The eluted target protein was then concentrated and incubated with thrombin protease overnight at 4°C to remove the hexahistidine tag at the N-terminus. Cleaved protein was dialyzed and a second performed Ni-IDA affinity chromatography in order to remove the hexahistidine tag and uncleaved proteins. Target proteins were further purified through a size-exclusion chromatography on a HiLoad 16/60 Superdex 75 column (GE Healthcare Life Science, Uppsala, Sweden), which was pre-equilibrated with buffer containing 20 mM Tris–HCl (pH 7.5) and 150 mM KCl. Final purified proteins were concentrated and stored at −80°C. Target protein was confirmed through SDS–PAGE.

E1 ubiquitin-activating enzyme was constructed into pETDuet vector. Recombinant constructs were expressed and purified with Ni-IDA affinity chromatography as mentioned above. Eluted E1 protein was further subjected to HiLoad 16/60 Superdex 200 column (GE Healthcare Life Science, Uppsala, Sweden), which was pre-equilibrated with buffer [20 mM Tris–HCl (pH 7.8) and 100 mM NaCl]. Final target protein was concentrated and stored in deep freezer (−80°C) prior the enzymatic reaction.

E2-25K ubiquitin-conjugating enzyme was constructed into pET28a vector with an N-term His-tagging. The recombinant construct was expressed and purified with Ni-IDA affinity chromatography as mentioned above. Eluted E1 protein was further subjected to HiLoad 16/60 Superdex 75 column (GE Healthcare Life Science, Uppsala, Sweden), which was pre-equilibrated with buffer [50 mM Tris–HCl (pH 8.0), 100 mM KCI, 2 mM MgCl2, and 5% glycerol]. Final target protein was concentrated up to 10 mg/ml and stored in deep freezer (−80°C) prior to enzymatic reaction.

Crystallization, data collection, and structure determination

For crystallization of p97 N-terminal domain (p97N) and human RHBDL4 VBM peptide complex, initial screening was performed by the sitting-drop vapor diffusion method. Drops formed from 0.2 µl protein solution plus 0.2 µl reservoir solutions. Initial hit of protein complex crystals was observed under condition of reservoir solution containing 0.1 M Bis-Tris propane (pH 6.5), 0.2 M sodium acetate, and 20% (w/v) polyethylene glycol (PEG) 3350. p97N concentration of 20 mg/ml was used in crystallization, and it was mixed with RHBDL4 VBM peptide in a molar ratio of 1 : 2. A micro-seeding technique was applied to the protein complex crystals due to appearance of multineedle crystals and their tiny size. After 3 days of micro-seeding, crystals of 0.2 mm size were obtained. For micro-seeding optimization, the p97N concentration used was 15 mg/ml. Complex crystals were eventually soak in the reservoir buffer containing 35% (w/v) PEG 3350 as a cryoprotectant prior to flash freezing in a stream of nitrogen gas at 100 K.

Diffraction data for p97N and RHBDL4 VBM peptide complex were collected on Beamline 5C at Pohang Accelerator Laboratory (Pohang, Korea) to 1.88 Å resolutions. Raw data integration and scaling were performed with the HKL2000 [36]. Complex proteins belong to space group P65. The crystal contains one protein complex in the asymmetric unit and the calculated Matthews coefficient is 3.01 Å3/Da, with an estimated solvent content of 59.1% [37]. Initial molecular replacement was performed with the molrep program of CCP4 suite using the apo structure of p97N (PDB ID 3QQ7) as a search model. The VBM residues were constructed into the post-MR electron density map beginning with a polyalanine backbone of VBM peptide. The model was then refined in cyclic rounds of manual model building in COOT [38] and refinement using refmac program of CCP4 suite [39]. In the final stages, solvent molecules were modeled into large positive difference electrodensity peaks that were within hydrogen bond distance or suitable partners. Final refinement was performed using Phenix [31] up to 18.0% (Rfree = 21.0%) at 1.88 Å resolutions. A composite omit map with refinement was created using partial models (final model with VBM ligand deleted) in the phenix.composite_omit_map. For structural analysis, superposition of structures and RMSD (Å) scores for the Cα atoms was performed using CCP4 program LSQKAB [39]. All molecular graphics were created using PyMoL [40]. Refinement statistics are summarized in Table 1.

Ensemble refinement of apo and cofactor-bound p97N crystal structures

Ensemble refinement for apo-p97N (PDB ID 3QQ7), FAF1-UBX-bound p97N (PDB ID 3QQ8), OTU1-UBXL-bound p97N (PDB ID 4KDL), p97N–VIM (PDB ID 3TIW), and p97N–VBM (PDB ID 5EPP) was carried out using phenix.ensemble_refinement. Structural models and structural factors of p97N and cofactor complexes were downloaded from PDB server and used as input models for ensemble refinement [41,42]. Default parameters in the phenix.ensemble_refinement were used including pTLS = 0.8 and Tbath = 5 K, and solvent updated every 25 cycles. The simulations have an equilibration phase (10τx) in which the temperature, X-ray weight, and averaged structure factors stabilize, followed by an acquisition phase (10τx). Ensemble structures were visualized using PyMOL with a command ‘set all_states, 1’. Structural fluctuation in the ensembles was quantified by calculating the RMSF using PyMOL with a script ‘ens_tool.py’ and the command ‘ens_rmsf,selection’. RMSF difference histogram for the p97N molecules was plotted using SigmaPlot 12.

In vitro binding assays

For pull-down assays, GST-RHBDL4c variants or mutants were immobilized on GST agarose beads. Approximately 40 µl of beads were incubated with 5 µM of purified GST-RHBDL4c variants in 400 µl of binding buffer [20 mM Tris–HCl (pH 7.5), 150 mM KCl, 4 mM DTT, and 0.1% (v/v) Triton X-100] at 4°C for 2 h. The beads were then centrifuged at 1250 g for 30 s and resuspended with 400 µl of binding buffer. Such washing step was repeated for four times. Purified p97N or p97ND1 (25 µM) in a total volume of 400 µl of binding buffer was then added to the immobilized GST-RHBDL4c variants and again incubated for 2 h at 4°C. Similar washing steps as mentioned above were performed. Immobilized proteins were directly analyzed by 15% (w/v) SDS–PAGE. GST protein without RHBDL4 served as a negative control.

Ternary complex pull-down of p97, RHBDL4c, and ubiquitin was performed using GST-tagged RHBDL4c as bait, immobilized on GST agarose resin. Beads (40 µl) were incubated with 5 µM purified GST-RHBDL4 variants in 400 µl of binding buffer [10 mM NaH2PO4 pH 7.4, 140 mM NaCl, 2.7 mM KCl, 1.8 mM KH2PO4, and 0.1% (v/v) Tween 20] at 4°C for 2 h. The beads were then centrifuged at 1250 g for 30 s and resuspended in 400 µl of binding buffer. This washing step was performed four times. Purified full-length p97 or p97N was then added to immobilized GST-tagged p97 and incubated for 2 h at 4°C. Similar washing steps as mentioned above were performed. Subsequently, mono-Ub, di-Ub, or tri-Ub was added to the immobilized RHBDL4 and p97 complex proteins and incubated at 4°C for 2 h followed by the washing steps. Immobilized proteins were directly analyzed by 17% (w/v) SDS–PAGE. Three replicates were performed in each independent experiment.

Preparation of K48 linked di-Ub and tri-Ub

The K48-linked di-Ub was synthesized via enzymatic ubiquitination reaction. E1, E2-25K, and ubiquitin were mixed in the ratio 1 : 15 : 1000 in a reaction buffer containing 50 mM Tris (pH 10), 150 mM NaCl, 4 mM ATP, 5 mM DTT, and 5 mM MgCl2 and incubated for 3 h at 37°C. The mixture was then injected onto HiLoad 16/60 Superdex 75 size-exclusion column (GE Healthcare Life Science, Uppsala, Sweden), which was pre-equilibrated with buffer containing 50 mM sodium acetate (pH 4.5). The fractions containing K48-linked di-Ub and tri-Ub were pooled and separated via an anion exchange column (Resource Q, GE health) with gradient elution using buffer A and B (buffer A: 50 mM sodium acetate; buffer B: 50 mM sodium acetate, 1 M NaCl). Separated di-Ub and tri-Ub were further subjected to analytical size-exclusion chromatography on a Superdex 10/300 GL column equilibrated with 20 mM Hepes–NaOH (pH 7.5) and 150 mM KCl. Final purified proteins were concentrated and stored at −80°C. The purity of target protein was confirmed using SDS–PAGE.

Chemical cross-linking

Purified RHBDL4 C3 (100 µM; GST-cleaved), p97N, mono-Ub, and di-Ub were mixed in 50 µl of reaction buffer containing 10 mM NaH2PO4 pH 7.4, 140 mM NaCl, 2.7 mM KCl, 1.8 mM KH2PO4, 0.5 µM disuccinimidyl suberate (DSS; Pierce) for 2 h at room temperature. The reaction was stopped by adding 100 mM Tris pH 8.0. The samples were mixed with 5× SDS sample buffer, boiled, and analyzed by 17% SDS–PAGE.

Surface plasmon resonance

SPR experiment was performed using the BIACORE X instrument (Biacore AB, Uppsala, Sweden) at 298 K. N-terminus 6His-tagging with p97N was used at a concentration of 1 mg/ml in 10 mM HEPES–NaOH (pH 7.4), 100 mM NaCl and immobilized on surface of CM5 sensor chip (GE Healthcare) using amine coupling based on the recommended Biacore instruction. SPR binding assay of RHBDL4 C-terminal protein (C3; residues 255–315) was carried out at a flow rate of 20 µl/min in 10 mM HEPES–NaOH (pH 7.4) and 100 mM NaCl. C3 was used in a concentration-dependent manner of 30, 60, 90, 120, 150, 180, 210, 270, and 300 µM, respectively. The dissociation constant Kd of 68.9 ± 2.4 µM (with kon = 5.57 ± 0.86 × 101 M−1 s−1 and koff = 1.49 ± 0.18 × 10−3 s−1) was calculated from three independent experiments. The kinetics of binding curves were analyzed using BIA evaluation software 4.1 for 1:1 interaction model and SigmaPlot 12 software.

Accession number

Atomic co-ordinates and structure factors of p97N and RHBDL4 VBM complex have been deposited in the Protein Data Bank at RCSB with accession code 5EPP.

Abbreviations

Bik, Bcl-2-interacting killer; di-Ub, di-ubiquitin; DSS, disuccinimidyl suberate; DTT, dithiothreitol; ER, endoplasmic reticulum; ERAD, endoplasmic reticulum-associated degradation; FPR, phenylalanine, proline, and arginine residues; IPTG, isopropyl β-d-1-thiogalactopyranoside; KCLB, Korea Cell Line Bank; LB, Luria-Bertani; mono-Ub, mono-ubiquitin; Nc, p97N N-terminus; Nn, p97N C-terminus; p97/VCP, valosin-containing protein; p97N, p97 N-terminal domain; PEG, polyethylene glycol; pTα, pre-T-cell receptor α; pTLS, percent of atoms that are included in the torsion libration screw (TLS) fitting procedure; RF, arginine and phenylalanine residues; RMSD, root-mean-square deviation; RMSF, root-means-square fluctuation; RRhRL, (R: arginine; F: phenylalanine; P: proline; L: Leucine; h: helical residues); SPR, surface plasmon resonance; TEV, tobacco etch virus; tri-Ub, tri-ubiquitin; UBX, ubiquitin regulatory X; UBXL, UBX-like element; UIM, ubiquitin-interacting motif; VBM, VCP-binding motif; VIM, VCP-interacting motif.

Author Contribution

J.J.L. and S.H.E. conceived the original concept and designed the experiment. J.J.L. carried out the experimental studies and solved the crystal structures. Y.L. supported the SPR experiment and structure determination. T.T.L. and J.Y.K. helped in constructs cloning. J.Y.A., J.-G.L., H.-S.Y., and K.R.P. supported the data collection. T.G.K., J.K.Y., Y.J., and S.H.E. revised the manuscript critically. All authors read and approved the final manuscript.

Funding

This work was supported by grants from National Research Foundation (NRF) Grants [20070056157, 2013M3A9A7046297, 2013R1A2A2A01068440, and 2015M2A2A4A03044653].

Acknowledgments

The authors thank Korea Basic Science Institute (KBSI, Ochang, South Korea) and expert assistance of staffs who helped in data collection at beamline BL-5C of the Pohang Accelerator Laboratory (PAL, Pohang, South Korea) and beamline NW12A at the Photon Factory (Tsukuba, Japan).

Competing Interests

The Authors declare that there are no competing interests associated with the manuscript.

References

References
1
Yoshida
,
H.
(
2007
)
ER stress and diseases
.
FEBS J.
274
,
630
658
doi:
2
Rabinovich
,
E.
,
Kerem
,
A.
,
Fröhlich
,
K.-U.
,
Diamant
,
N.
and
Bar-Nun
,
S.
(
2002
)
AAA-ATPase p97/Cdc48p, a cytosolic chaperone required for endoplasmic reticulum-associated protein degradation
.
Mol. Cell. Biol.
22
,
626
634
doi:
3
Jarosch
,
E.
,
Taxis
,
C.
,
Volkwein
,
C.
,
Bordallo
,
J.
,
Finley
,
D.
,
Wolf
,
D.H.
et al. 
(
2002
)
Protein dislocation from the ER requires polyubiquitination and the AAA-ATPase Cdc48
.
Nat. Cell Biol.
4
,
134
139
doi:
4
Lemberg
,
M.K.
and
Freeman
,
M.
(
2007
)
Functional and evolutionary implications of enhanced genomic analysis of rhomboid intramembrane proteases
.
Genome Res.
17
,
1634
1646
doi:
5
Bagola
,
K.
,
Mehnert
,
M.
,
Jarosch
,
E.
and
Sommer
,
T.
(
2011
)
Protein dislocation from the ER
.
Biochim. Biophys. Acta
1808
,
925
936
doi:
6
Christianson
,
J.C.
and
Ye
,
Y.
(
2014
)
Cleaning up in the endoplasmic reticulum: ubiquitin in charge
.
Nat. Struct. Mol. Biol.
21
,
325
335
doi:
7
Fleig
,
L.
,
Bergbold
,
N.
,
Sahasrabudhe
,
P.
,
Geiger
,
B.
,
Kaltak
,
L.
and
Lemberg
,
M.K.
(
2012
)
Ubiquitin-dependent intramembrane rhomboid protease promotes ERAD of membrane proteins
.
Mol. Cell.
47
,
558
–569| doi:
8
Tsai
,
Y.C.
and
Weissman
,
A.M.
(
2012
)
A ubiquitin-binding rhomboid protease aimed at ERADication
.
Dev. Cell.
23
,
454
456
doi:
9
Wang
,
Y.
,
Guan
,
X.
,
Fok
,
K.
,
Li
,
S.
,
Zhang
,
X.
,
Miao
,
S.
et al. 
(
2008
)
A novel member of the Rhomboid family, RHBDD1, regulates BIK-mediated apoptosis
.
Cell Mol. Biol. Sci.
65
,
3822
3829
doi:
10
Wan
,
C.
,
Fu
,
J.
,
Wang
,
Y.
,
Miao
,
S.
,
Song
,
W.
and
Wang
,
L.
(
2012
)
Exosome-related multi-pass transmembrane protein TSAP6 is a target of rhomboid protease RHBDD1-induced proteolysis
.
PLoS ONE
7
,
e37452
doi:
11
Zhang
,
X.
,
Shaw
,
A.
,
Bates
,
P.A.
,
Newman
,
R.H.
,
Gowen
,
B.
,
Orlova
,
E.
, et al. 
(
2000
)
Structure of the AAA ATPase p97
.
Mol. Cell.
6
,
1473
1484
doi:
12
Meyer
,
H.
,
Bug
,
M.
and
Bremer
,
S.
(
2012
)
Emerging functions of the VCP/p97 AAA-ATPase in the ubiquitin system
.
Nat. Cell Biol.
14
,
117
123
doi:
13
Jentsch
,
S.
and
Rumpf
,
S.
(
2007
)
Cdc48 (p97): a ‘molecular gearbox’ in the ubiquitin pathway?
Trends Biochem. Sci.
32
,
6
11
doi:
14
Baek
,
G.H.
,
Cheng
,
H.
,
Choe
,
V.
,
Bao
,
X.
,
Shao
,
J.
,
Luo
,
S.
et al. 
(
2013
)
Cdc48: a Swiss army knife of cell biology
.
J. Amino Acids
2013
,
12
doi:
15
Schuberth
,
C.
,
Richly
,
H.
,
Rumpf
,
S.
and
Buchberger
,
A.
(
2004
)
Shp1 and Ubx2 are adaptors of Cdc48 involved in ubiquitin-dependent protein degradation
.
EMBO Rep.
5
,
818
824
doi:
16
Wolf
,
D.H.
and
Stolz
,
A.
(
2012
)
The Cdc48 machine in endoplasmic reticulum associated protein degradation
.
Biochim. Biophys. Acta
1823
,
117
124
doi:
17
Madsen
,
L.
,
Seeger
,
M.
,
Semple
,
C.A.
and
Hartmann-Petersen
,
R.
(
2009
)
New ATPase regulators—p97 goes to the PUB
.
Int. J. Biochem. Cell Biol.
41
,
2380
2388
doi:
18
Hirabayashi
,
M.
,
Inoue
,
K.
,
Tanaka
,
K.
,
Nakadate
,
K.
,
Ohsawa
,
Y.
,
Kamei
,
Y.
et al. 
(
2001
)
VCP/p97 in abnormal protein aggregates, cytoplasmic vacuoles, and cell death, phenotypes relevant to neurodegeneration
.
Cell Death Differ.
8
,
977
984
doi:
19
Dreveny
,
I.
,
Kondo
,
H.
,
Uchiyama
,
K.
,
Shaw
,
A.
,
Zhang
,
X.
and
Freemont
,
P.S.
(
2004
)
Structural basis of the interaction between the AAA ATPase p97/VCP and its adaptor protein p47
.
EMBO J.
23
,
1030
1039
doi:
20
Hänzelmann
,
P.
,
Buchberger
,
A.
and
Schindelin
,
H.
(
2011
)
Hierarchical binding of cofactors to the AAA ATPase p97
.
Structure
19
,
833
843
doi:
21
Kim
,
S.J.
,
Cho
,
J.
,
Song
,
E.J.
,
Kim
,
S.J.
,
Kim
,
H.M.
,
Lee
,
K.E.
et al. 
(
2014
)
Structural basis for ovarian tumor domain-containing protein 1 (OTU1) binding to p97/Valosin-containing protein (VCP
).
J. Biol. Chem.
289
,
12264
12274
doi:
22
Hänzelmann
,
P.
and
Schindelin
,
H.
(
2011
)
The structural and functional basis of the p97/Valosin-containing protein (VCP)-interacting motif (VIM): mutually exclusive binding of cofactors to the N-terminal domain of p97
.
J. Biol. Chem.
286
,
38679
38690
doi:
23
Hänzelmann
,
P.
and
Schindelin
,
H.
(
2016
)
Characterization of an additional binding surface on the p97 N-Terminal domain involved in bipartite cofactor interactions
.
Structure
24
,
140
147
doi:
24
Krissinel
,
E.
and
Henrick
,
K.
(
2007
)
Inference of macromolecular assemblies from crystalline state
.
J. Mol. Biol.
372
,
774
797
doi:
25
Wallace
,
A.C.
,
Laskowski
,
R.A.
and
Thornton
,
J.M.
(
1995
)
LIGPLOT: a program to generate schematic diagrams of protein-ligand interactions
.
Protein Eng.
8
,
127
134
doi:
26
Liu
,
S.
,
Fu
,
Q.-S.
,
Zhao
,
J.
and
Hu
,
H.-Y.
(
2013
)
Structural and mechanistic insights into the arginine/lysine-rich peptide motifs that interact with P97/VCP
.
Biochim. Biophys. Acta, Proteins Proteomics
1834
,
2672
2678
doi:
27
Schuberth
,
C.
and
Buchberger
,
A.
(
2008
)
UBX domain proteins: major regulators of the AAA ATPase Cdc48/p97
.
Cell. Mol. Life Sci.
65
,
2360
2371
doi:
28
Di Tommaso
,
P.
,
Moretti
,
S.
,
Xenarios
,
I.
,
Orobitg
,
M.
,
Montanyola
,
A.
,
Chang
,
J.-M.
et al. 
(
2011
)
T-Coffee: a web server for the multiple sequence alignment of protein and RNA sequences using structural information and homology extension
.
Nucleic Acids Res.
39
,
W13
W17
doi:
29
Robert
,
X.
and
Gouet
,
P.
(
2014
)
Deciphering key features in protein structures with the new ENDscript server
.
Nucleic Acids Res.
42
,
W320
W324
doi:
30
Lee
,
J.-J.
,
Park
,
J.K.
,
Jeong
,
J.
,
Jeon
,
H.
,
Yoon
,
J.-B.
,
Kim
,
E.E.
et al. 
(
2013
)
Complex of Fas-associated factor 1 (FAF1) with valosin-containing protein (VCP)-Npl4-Ufd1 and polyubiquitinated proteins promotes endoplasmic reticulum-associated degradation (ERAD
).
J. Biol. Chem.
288
,
6998
7011
doi:
31
Adams
,
P.D.
,
Afonine
,
P.V.
,
Bunkóczi
,
G.
,
Chen
,
V.B.
,
Davis
,
I.W.
,
Echols
,
N.
, et al. 
(
2010
)
PHENIX: a comprehensive Python-based system for macromolecular structure solution
.
Acta Crystallogr. D Biol. Crystallogr.
66
,
213
221
doi:
32
Banerjee
,
S.
,
Bartesaghi
,
A.
,
Merk
,
A.
,
Rao
,
P.
,
Bulfer
,
S.L.
,
Yan
,
Y.
, et al. 
(
2016
)
2.3 A resolution cryo-EM structure of human p97 and mechanism of allosteric inhibition
.
Science
351
,
871
875
doi:
33
Meyer
,
H.H.
,
Wang
,
Y.
and
Warren
,
G.
(
2002
)
Direct binding of ubiquitin conjugates by the mammalian p97 adaptor complexes, p47 and Ufd1–Npl4
.
EMBO J.
21
,
5645
5652
doi:
34
Dai
,
R.M.
and
Li
,
C.-C.H.
(
2001
)
Valosin-containing protein is a multi-ubiquitin chain-targeting factor required in ubiquitin-proteasome degradation
.
Nat. Cell Biol.
3
,
740
744
doi:
35
Fisher
,
R.D.
,
Wang
,
B.
,
Alam
,
S.L.
,
Higginson
,
D.S.
,
Robinson
,
H.
,
Sundquist
,
W.I.
et al. 
(
2003
)
Structure and ubiquitin binding of the ubiquitin-interacting motif
.
J. Biol. Chem.
278
,
28976
28984
doi:
36
Otwinowski
,
Z.
and
Minor
,
W.
(
1997
) Processing of X-ray diffraction data collected in oscillation mode. In
Methods in Enzymology
(
Carter
, Jr,
C.W.
and
Sweet
,
R.M.
eds), pp.
307
326
,
Academic Press
37
Matthews
,
B.W.
(
1968
)
Solvent content of protein crystals
.
J. Mol. Biol.
33
,
491
497
doi:
38
Emsley
,
P.
and
Cowtan
,
K.
(
2004
)
Coot: model-building tools for molecular graphics
.
Acta Crystallogr. D Biol. Crystallogr.
60
,
2126
2132
doi:
39
Winn
,
M.D.
,
Ballard
,
C.C.
,
Cowtan
,
K.D.
,
Dodson
,
E.J.
,
Emsley
,
P.
,
Evans
,
P.R.
, et al. 
(
2011
)
Overview of the CCP4 suite and current developments
.
Acta Crystallogr. D Biol. Crystallogr.
67
,
235
242
doi:
40
The PyMOL Molecular Graphics System, Version 1.5.0.4 Schrödinger, LLC Available at www.pymol.org
41
Burnley
,
B.T.
,
Afonine
,
P.V.
,
Adams
,
P.D.
and
Gros
,
P.
(
2012
)
Modelling dynamics in protein crystal structures by ensemble refinement
.
eLife
1
,
e00311
doi:
42
Burnley
,
B.T.
, and
Gros
,
P.
(
2013
)
Phenix.ensemble_refinement: a test study of apo and holo BACE1
.
Comput. Crystallogr. Newsl.
4
,
51
–58