Complex of Fas-associated Factor 1 (FAF1) with Valosin-containing Protein (VCP)-Npl4-Ufd1 and Polyubiquitinated Proteins Promotes Endoplasmic Reticulum-associated Degradation (ERAD)*

Background: FAF1, which has multiple ubiquitin-like domains, interacts with various proteins (VCP, Hsp70, and polyubiquitinated proteins). Results: Association of FAF1 UBX with VCP-Npl4-Ufd1 complex regulates ubiquitin binding to FAF1 UBA domain and promotes CD3δ degradation in ERAD. Conclusion: FAF1 is a ubiquitin receptor that promotes ERAD by delivering polyubiquitinated proteins from UBX domain to UBA domain. Significance: FAF1 plays a role in ERAD by modulating domain-domain interaction. Fas-associated factor 1 (FAF1) is a ubiquitin receptor containing multiple ubiquitin-related domains including ubiquitin-associated (UBA), ubiquitin-like (UBL) 1, UBL2, and ubiquitin regulatory X (UBX). We previously showed that N-terminal UBA domain recognizes Lys48-ubiquitin linkage to recruit polyubiquitinated proteins and that a C-terminal UBX domain interacts with valosin-containing protein (VCP). This study shows that FAF1 interacts only with VCP complexed with Npl4-Ufd1 heterodimer, a requirement for the recruitment of polyubiquitinated proteins to UBA domain. Intriguingly, VCP association to C-terminal UBX domain regulates ubiquitin binding to N-terminal UBA domain without direct interaction between UBA and UBX domains. These interactions are well characterized by structural and biochemical analysis. VCP-Npl4-Ufd1 complex is known as the machinery required for endoplasmic reticulum-associated degradation. We demonstrate here that FAF1 binds to VCP-Npl4-Ufd1 complex via UBX domain and polyubiquitinated proteins via UBA domain to promote endoplasmic reticulum-associated degradation.

destination. ERAD as a quality control for proteins eliminates misfolded proteins and prevents their accumulation. The ERAD process includes retrotranslocation of misfolded proteins from the ER membrane into the cytosol, polyubiquitination by E3 ligases, and delivery to the proteasome for final degradation (10,11). It is known that VCP uses nuclear protein localization protein 4 (Npl4)-Ufd1 heterodimer as the major ubiquitin-recruiting cofactor (12,13), and furthermore, VCP-Npl4-Ufd1 complex plays an essential role in retrotranslocation as well as delivery to the proteasome (13).
We previously demonstrated that UBX interacts with VCP in a stress-dependent manner and that FAF1 plays a key role as a ubiquitin receptor (1). However, how FAF1 functions as a ubiquitin receptor is not well understood. In this study, we examined the domain-domain interactions possibly involved in the regulation of FAF1-VCP interaction. We constructed various mutants in ubiquitin-related domains of FAF1 and examined their interactions with VCP. We found that recruitment of polyubiquitinated proteins to N-terminal UBA domain in FAF1 is regulated by the interaction of C-terminal UBX domain with VCP-Npl4-Ufd1 complex. Furthermore, because VCP-Npl4-Ufd1 complex is known to play a key role in ERAD, we investigated whether FAF1 participates in the ERAD process. We report here that FAF1 promotes the degradation of ERAD substrate CD3␦ in a VCP-Npl4-Ufd1-dependent manner.
Cell Extracts and Immunoprecipitation-HEK293T and HeLa cells were grown and maintained in DMEM (Eagle's minimal essential medium) supplemented with 10% fetal bovine serum (FBS) at 37°C in 5% CO 2 . Transfections were carried out with Effectene (Qiagen) transfection reagent in a 1:10 ratio as instructed by the manufacturer. For immunoprecipitation, cells (2-10 ϫ 10 6 ) were lysed with a hypotonic lysis buffer (10 mM HEPES, pH 7.4, 1.5 mM MgCl 2 , 0.5% Nonidet P-40) containing protease inhibitors (10 g/ml aprotinin, 10 g/ml leupeptin, 1 g/ml pepstatin, 100 g/ml phenylmethylsulfonyl fluoride) and phosphatase inhibitors (10 mM NaF, 10 mM Na 3 VO 4 ) for 30 min on ice and centrifuged at 16,000 ϫ g for 15 min. The supernatant was incubated for 3 h at 4°C with anti-FLAG-agarose affinity beads. The beads were washed five times with 1 ml of lysis buffer containing 0.5% Nonidet P-40 to remove nonspecific binding, and the immune complex was solubilized in SDS gel sample buffer, separated by 10% SDS-PAGE, and detected with silver staining or Western analysis.
Silencing RNAs-ON-TARGETplus SMARTpool siRNAs and siControl duplex siRNA (Dharmacon) were used to knock down Npl4 and FAF1, respectively, at a final concentration of 100 nM. Silencing was achieved using Dharma FECT1 transfection reagent (Dharmacon) according to the manufacturer's instructions. Changes in the expression of FAF1 in HEK293T and HeLa cells were analyzed 72 h after siRNA transfection.
Protein Identification Using UPLC-ESI-q-TOF Tandem MS-To identify the proteins and modifications, the gel bands were destained and digested with trypsin, and the resulting peptides were extracted as described previously (16). The peptide extracts were evaporated to dryness in a SpeedVac and dissolved in 10% acetonitrile solution containing 1.0% formic acid. The dissolved samples were desalted on line prior to separation using a trap column (5-m particle size; NanoEase TM dC 18 , Waters) cartridge. Peptides were separated using a C 18 reversed-phase 75-m-inner diameter ϫ 150-mm analytical column (3-m particle size; Atlantis TM dC 18 , Waters) with an integrated electrospray ionization SilicaTip TM (10-m inner diameter; New Objective). Chromatography was performed on line to a mass spectrometer (Q-Tof Ultima TM Global, Waters). Raw data obtained from the mass spectrometer were converted to .pkl files using ProteinLynx Global Server TM 2.3 data processing software (Waters). MS/MS spectra were matched against amino acid sequences in Swiss-Prot. Large numbers and types of potential post-translational modifications were considered. All reported assignments were verified by automatic and manual interpretation of spectra from Mascot and MOD i (17) in a blind mode.
Isothermal Titration Calorimetry (ITC)-ITC experiments were performed using VP-ITC and ITC200 instruments (MicroCal, Northampton, MA) at 298 K, and the data were analyzed using the program Origin 7.0. All samples (in 50 mM HEPES, pH 7.5, 150 mM NaCl) were centrifuged and degassed prior to the measurements at 298 K. The injectants were added at 150-s intervals to the sample solution in the cell.
Electron Microscopy (EM)-For EM, purified N-terminal His-tagged full-length FAF1 at 50 g/ml in 50 mM HEPES buffer, pH 7.5 containing 150 mM NaCl was incubated with a 10-fold molar excess of 1.5-nm Ni-NTA-Nanogold (Nano-probes Inc.) for 30 min, and the excess Ni-NTA-Nanogold was removed using a Superdex-200GL column (GE Healthcare). For generating FAF1 and VCP-Ufd1-Npl4 complex, gold-labeled FAF1 was incubated with purified VCP-Ufd1-Npl4 complex at a 1:1 molar ratio for 1 h at 4°C, and the product was further purified on a Superdex-200S column to ensure removal of unbound FAF1 in 50 mM HEPES, pH 7.5, 150 mM NaCl, 5 mM MgCl 2 , 1 mM AMP-PNP. The purified complex was diluted to 0.3 mg/ml concentration, then applied to glow-discharged carbon-coated grids, rinsed, and stained with 2% uranyl acetate. Images were recorded on a 2000 ϫ 2000 charge-coupled device camera using a Tecnai F20 field emission gun electron microscope operated at 200 kV (FEI Co.). For imaging Nanogold (ϳ1.5 nm in diameter) on the complex, a Tecnai Titan microscope was used at 300 kV without staining the sample (FEI Co.). To observe both gold particles and protein complex, we used a

FAF1 Regulates ERAD via VCP-Npl4-Ufd1 Complex
cryotransmission EM method. For cryoexperiments, we diluted purified samples to a concentration of ϳ0.02-0.1 mg/ml for VCP-Npl4-Ufd1 with Nanogold-labeled FAF1 complex and loaded samples onto holey carbon film-supported grids and plunge froze them. We recorded images on a charge-coupled device camera (2000 ϫ 2000 charge-coupled device camera, Gatan) using a Tecnai F20 field emission gun electron microscope operated at 200 kV. Particles were selected from the individual digital micrographs.
X-ray Crystallography of FAF1 UBX⅐VCP N-domain Complex-Diffraction quality crystals of FAF1 UBX domain and VCP N-terminal domain complex (see supplemental Fig. S1) were obtained from 100 mM sodium acetate, 1 M LiCl 2 , 30% PEG 4000. FAF1 UBX⅐VCP N-domain complex was methylated and crystallized following the procedure described earlier (18). The crystal structure was determined by the molecular replacement method (MOLREP from the CCP4 suite) (19) using the N-domain of the VCP ND1 crystal structure (Protein Data Bank code 1E32) (20) as a search model. The resulting electron density was good enough to orient the FAF1 UBX using CAPRA. Manual building using Coot (21) and refinement using CNS (22) gave the final models, which were assessed with PROCHECK (23). Crystallographic data collected and the refinement statistics are summarized in supplemental

RESULTS
Binding of VCP to C-terminal UBX Domain of FAF1 Is Necessary for the Recruitment of Polyubiquitinated Proteins to N-terminal UBA Domain-To investigate the effect of VCP binding on the biological function of FAF1, we designed a mutant of FAF1 defective in VCP binding. A previous structural study showed that UBX has a well conserved FPR motif in the S3/S4 loop that is a VCP binding motif (14,24) and that FAF1 also has an S3/S4 loop in UBX domain, 618 TFPR 621 , that is a possible site for VCP binding of FAF1 UBX domain. We produced a mutant of UBX domain, FAF1(TFPR 3 AG), that has a shorter S3/S4 loop ( 618 TFPR 621 mutated to AG). HEK293T cells were transfected with control FLAG, FLAG-FAF1, and FLAG-FAF1(TFPR 3 AG), and the cell lysates were immunoprecipitated with anti-FLAG antibody, respectively. Immune complexes were detected with Western analysis using anti-VCP antibody. As shown in Fig. 2A, FLAG-FAF1(TFPR 3 AG) with the deletion in the S3/S4 loop of UBX abrogated the interaction of FAF1 with VCP. Using the mutant defective in VCP binding, we investigated whether VCP binding to UBX domain affects the scaffolding ability of FAF1 to interact with Hsp70 through UBL1 domain and to recruit polyubiquitinated proteins through UBA domain. HEK293T cells overexpressing FLAG-FAF1 wild type (WT) and FLAG-FAF1(TFPR 3 AG) were lysed, and the cell lysates were immunoprecipitated with anti-FLAG antibody. The immune complexes were separated by 10% SDS-PAGE and detected with anti-Hsp70 and antiubiquitin antibodies. As shown in Fig. 2A, there were no discernible interaction changes of Hsp70 between FAF1 WT and FLAG-FAF1(TFPR 3 AG). However, FAF1(TFPR 3 AG) could not recruit polyubiquitinated substrates to N-terminal UBA domain of FAF1. These results indicate that VCP binding to FAF1 is required to recruit polyubiquitinated proteins to UBA domain of FAF1. We therefore hypothesize that VCP binding to FAF1-UBX domain is necessary for the biological function of FAF1.
UBX Domain Regulates UBA Domain through VCP Binding without Direct Interaction-It was previously reported that ubiquitin-like folds mimic ubiquitin to interact with ubiquitin binding motifs such as UBA, ubiquitin-interacting motif (UIM), and proteasome subunits (25)(26)(27). To elucidate how UBX domain regulates UBA domain in recruiting polyubiquitinated proteins, we examined whether there is a direct interaction between UBLs and UBX with UBA domain, similar to ubiquitin-UBA domain interaction, because they both contain ubiquitin-like folds. We constructed various deletion mutants of UBX, UBL1, and UBL2 domains of FLAG-FAF1 WT and FLAG-FAF1(TFPR 3 AG) ( Fig. 1) and examined whether ubiquitin-like folds inactivate UBA domain by direct interaction.
Because FLAG-FAF1(TFPR 3 AG) failed to recruit polyubiquitinated proteins, we examined whether ubiquitin-like folds of UBX domain directly inactivate UBA domain by assessing whether the FLAG-FAF1 UBX domain deletion mutant (FAF1⌬UBX) possesses ubiquitin recruiting function of UBA domain. HEK293T cells were transfected with FLAG-FAF1 WT, FLAG-FAF1(TFPR 3 AG), or FLAG-FAF1⌬UBX; cell lysates were immunoprecipitated with anti-FLAG antibody; and the immune complexes were detected by Western analysis using anti-ubiquitin, anti-FLAG, and anti-VCP antibodies. As shown in Fig. 2B, the recruitment of polyubiquitinated substrates through UBA domain disappeared in cells overexpressing FLAG-FAF1⌬UBX as well as FLAG-FAF1(TFPR 3 AG). This suggests that UBA domain is regulated by UBX domain through VCP binding rather than via direct UBX-UBA interaction mediated by ubiquitin-like folds of UBX domain. FAF1 has two UBL domains (UBL1 (aa 100 -176) and UBL2 (aa 195-270)) located next to UBA domain (aa 1-47). We examined whether these UBL domains directly interact with UBA domain. Such an interaction is possible based on structural considerations, similar to Rad23 and Dsk2 (26,28). We constructed UBL domain deletion mutants of FAF1 WT and FAF1(TFPR 3 AG) (Fig. 1A) and examined the interaction of UBA domain with polyubiquitinated proteins. HEK293 cells were transiently transfected with FLAG-FAF1 WT, FLAG-FAF1(TFPR 3 AG), and UBL domain deletion constructs ⌬UBL1 and ⌬UBL1-2. The cell lysates were immunoprecipitated with anti-FLAG antibody, and the immune complexes were separated by 10% SDS-PAGE and detected by Western analysis using anti-FLAG, anti-VCP, and anti-ubiquitin antibodies. As shown in Fig. 2C, recruitment of polyubiquitinated proteins to FAF1 WT and FAF1(TFPR 3 AG) was not affected by deletion of UBL domains. This indicates that UBL domains do not play any role in this interaction. We also confirmed domain-domain interaction using various domain proteins: FAF1 WT, FAF1(82-650) as a UBA deletion mutant, FAF1 I41N as a mutant defective in UBA domain, and FAF1(TFPR 3 MARCH 8, 2013 • VOLUME 288 • NUMBER 10  AG) as a mutant defective in UBX domain (Fig. 2D). These results indicate that the mutant FAF1(TFPR 3 AG), which does not interact with VCP, also failed to recruit polyubiquitinated proteins like UBA deletion mutants.

FAF1 Regulates ERAD via VCP-Npl4-Ufd1 Complex
Next, we determined the dissociation constants (K D ) between FAF1 UBA (aa 1-81) and UBX (aa 571-650) and between UBA and UBL1-2 (aa 100 -280) using ITC to examine direct domain interactions ( Thr 618 -Phe 619 -Pro 620 -Arg 621 forms a type VI ␤-turn (Fig. 3B) with Pro 620 , which is in the cis configuration, and it appears that this cis-Pro 620 -centered ␤-turn is important for UBX and VCP N-domain interaction. Recently, two independent studies also showed that the TFPR motif of FAF1 UBX domain adopts a cis-Pro 620 -centered ␤-turn (29) and touch-turn structures (which have a tighter turn than a ␤-turn) (30).
ITC studies showed that the FAF1 UBX domain binds the VCP N-domain at a 1:1 molar ratio with an apparent K D of 25.6 M (Fig. 3C). This is comparable with the K D value of 17.8 M obtained for Npl4 ubiquitin fold domain and VCP N-domain (supplemental Fig. S2A). On the other hand, TFPR 3 AG mutant showed no detectable binding (Fig. 3C). Therefore, these findings together with crystallographic results clearly show that the highly conserved motif 618 TFPR 621 in the S3/S4 loop of FAF1 UBX domain is crucial for binding to the N-domain of VCP.
FAF1 Interacts with the Substrate-recruiting Cofactor Npl4-Ufd1 Heterodimer via VCP-To characterize further the proteins that interact with FAF1, we performed immunoprecipitation on a large scale using anti-FLAG affinity beads from lysates of HEK293T cells overexpressing FLAG or FLAG-FAF1. Immune complexes were separated by 10% SDS-PAGE, and the FAF1-interacting protein was detected after silver staining (Fig.  4A). Using peptide sequencing analysis with UPLC-ESI-q-TOF tandem MS, we identified the FAF1-interacting protein as Npl4 (Fig. 4B). Because it is well known that Npl4 forms heterodimers with Ufd1 and binds to VCP through the ubiquitin fold domain (31), we confirmed the binding of Npl4-Ufd1 complex to FAF1 by Western analysis of FLAG-FAF1 immune complex with anti-Npl4, anti-Ufd1, and anti-VCP antibodies as reported previously (32). Interaction of FAF1 with Npl4-Ufd1 heterocomplex is shown in Fig. 4C. To further investigate whether FAF1 interacts directly with Npl4-Ufd1 heterodimer or through VCP, we performed immunoprecipitation of HEK293T cells overexpressing FLAG-FAF1 WT and FLAG-FAF1(TFPR 3 AG) with anti-FLAG antibody. As shown in Fig. 4D, FAF1(TFPR 3 AG), a VCP binding-deficient mutant did not interact with Npl4-Ufd1 heterodimer, suggesting that Npl4-Ufd1 heterodimer interacts with FAF1 through VCP.
We further examined the Ufd1-Npl4 dependence. The binding between full-length as well as various fragments of VCP (see Fig. 1B) and FAF1 was examined using ITC, and the results are summarized in Table 1. When VCP was presented as ND1, UBX domain bounds with an apparent K D of 20.1 M in the presence of ATP and 22.7 M in the absence of ATP, but the stoichiometry was no longer 1:1 (supplemental Fig. S2, B and  C). These values are in good agreement with value of 30 M reported previously (9). On the other hand, when the fulllength VCP (denoted as VCP ND1D2) was titrated with FAF1 UBX domain, there was no detectable binding whether ATP was present or not (Fig. 4E). However, if VCP ND1D2 was complexed with Npl4-Ufd1 heterodimer, FAF1 UBX domain bound with a slightly increased binding affinity with a K D value of 9.6 M (Fig. 4F) with one or two FAF1 UBX domains bound to one VCP-Npl4-Ufd1 complex. We tested this further by extending  Fig. S3). To confirm the stoichiometry of the VCP-Npl4-Ufd1-FAF1 complex observed in ITC data, we tried EM imaging after gold labeling on the purified FAF1. Examination of VCP alone revealed a closed form with a hole in the center, consistent with a top view of the hexameric ring (33). In contrast to the relatively uniform fields observed for VCP, the VCP-Npl4-Ufd1-FAF1 complex revealed more irregular particles and additional density at the periphery of the VCP ring (Fig. 5A). The irregularity of the complex likely stems from the flexible, elongated structure of FAF1, which may also promote a more variable orientation of the complex on the grid (Fig. 5B). For stoichiometry, we labeled FAF1 with Nanogold and examined the reconstituted complex by high resolution and cryoelectron microscopy. Nanogold particles were ready visualized in association with the complex at 300 kV, revealing a single gold particle and thus a single molecule of FAF1 per complex (Fig. 5C). We also used the cryomethod to confirm both densities from the gold particle and protein complex because the sample used for high resolution EM was not stained to show enough protein density (Fig. 5D). 1.4 -1.8-nm Nanogold particles were difficult to recognize in the conventional negatively stained EM but could be seen in the unstained complex with a ratio of one particle per complex when we used high resolution EM and cryotransmission EM. Arrows in Fig. 5 indicate the Nanogold-labeled FAF1 on the VCP-Npl4-Ufd1 complex. We tried the labeling with larger gold particles of 5-nm diameter, but the complex was not stable enough to go through the gel filtration step to get rid of excess gold particles.

FAF1 Regulates ERAD via VCP-Npl4-Ufd1 Complex
Npl4-Ufd1 Heterodimer Is Required for VCP-FAF1 Interaction-VCP functions as a molecular chaperone by interacting with diverse cofactors (9). We showed that VCP binding is crucial for the scaffolding ability of FAF1 and that prior binding of Npl4-Ufd1 heterodimer to VCP is crucial for FAF1 binding to VCP. To confirm this, we examined FAF1-VCP interaction in HeLa cells in which Npl4 was knocked down with its specific siRNA. These cells were transfected with FLAG or FLAG-FAF1, the cell lysates were immunoprecipitated with anti-FLAG antibody, and the immune complexes were analyzed by Western analysis using anti-FLAG, anti-VCP, anti-Npl4, anti-Ufd1, and anti-ubiquitin antibodies. We found that VCP could not bind to FAF1 in HeLa cells in which Npl4 was knocked down and that polyubiquitin binding to UBA domain was abol-ished as happened with mutant FAF1(TFPR 3 AG) (Fig. 6). This suggests that Npl4-Ufd1 heterodimer is crucial for the biological function of FAF1 conjugated to VCP.
FAF1 Promotes ER-associated Degradation via Ubiquitin Receptor Function-It is known that VCP-Npl4-Ufd1 complex plays a key role in ERAD. When ubiquitinated misfolded proteins are translocated through the ER membrane, VCP-Npl4-Ufd1 complex delivers them to the proteasome (12,13,34). Because FAF1 strongly binds to VCP and this binding is regulated by various stresses including heat shock (1) and by complex formation with Npl4-Ufd1, we examined the role of FAF1 in the ERAD pathway. Using the CD3␦ Tet-Off system (pYR-CD3␦-FLAG co-transfected with pTet-Off) as a classical ERAD substrate, we monitored the degradation rates of CD3␦ in HeLa  MARCH 8, 2013 • VOLUME 288 • NUMBER 10 cells overexpressing FLAG or FLAG-FAF1. We blocked the synthesis of CD3␦ by treating the cells with 50 g/ml doxycycline and then measured the degradation rates of CD3␦ through ERAD. To confirm that the Tet-Off system could be used to monitor CD3␦ degradation, we assessed CD3␦ degradation by treating the cells with 50 g/ml doxycycline alone or together with 10 g/ml cycloheximide. Supplemental Fig. S4 shows that the pattern of CD3␦ degradation promoted by FAF1 was the same whether the cells were treated with 50 g/ml doxycycline alone or with doxycycline plus 10 g/ml cycloheximide. Fig. 7A shows that FAF1 promoted CD3␦ degradation and is thus a component of the ERAD machinery. To further validate the function of FAF1 in ERAD, we examined ERAD in cells in which FAF1 was knocked down. Supplemental Fig. 5A shows that knocking down FAF1 did not affect protein levels of Npl4 and Ufd1. When HeLa cells in which FAF1 was knocked down or control cells treated with non-targeting siRNA were co-transfected with pYR-CD3␦-FLAG and pTet-Off and treated with doxycycline 24 h after transfection, the degradation of CD3␦ was attenuated in cells in which FAF1 was knocked down (Fig. 7B). Degradation of CD3␦, which was reduced when FAF1 was knocked down, was restored in cells overexpressing FLAG-FAF1, confirming that FAF1 plays a role in ERAD (Fig. 7C). We further investigated the role of FAF1 in ERAD as a ubiquitin receptor by examining the degradation rates of CD3␦ in HeLa cells respectively overexpressing the following mutants of FAF1: FLAG-FAF1 WT, FLAG-FAF1(82-650) (UBA-deleted mutant), or FLAG-FAF1(TFPR 3 AG) (VCP binding-defective mutant). As shown in Fig. 7D, neither FLAG-FAF1(82-650) nor FLAG-FAF1(TFPR 3 AG) promoted CD3␦ degradation like FAF1 WT. FLAG-FAF1(82-650), the UBA deletion mutant having intact UBX to bind to VCP, and FAF1(TFPR 3 AG), the mutant without the ability to bind to VCP and consequently its UBA function, did not affect the rate of ERAD. This suggests that FAF1 as a ubiquitin receptor plays a role in ERAD and that both UBA and UBX domains are required for promoting ERAD.

FAF1 Regulates ERAD via VCP-Npl4-Ufd1 Complex
FAF1 Promotes ER-associated Degradation in a VCP-Npl4-Ufd1-dependent Manner-We further investigated the VCP-Npl4-Ufd1-related function of FAF1 in the ERAD pathway by examining the connection of CD3␦ degradation to VCP-Npl4-Ufd1 complex. We monitored CD3␦ degradation in HeLa cells in which Npl4 was knocked down with siRNA. Knocking down Npl4 did not affect the expression level of FAF1 but did decrease the level of Ufd1 (supplemental Fig. 5A) as reported previously (35). HeLa cells treated with non-Npl4-targeting siRNA (control) or Npl4 siRNA were co-transfected with pYR-CD3␦-FLAG, pTet-Off, and FLAG-FAF1 or FLAG-FAF1(82- 650) and treated with doxycycline to monitor CD3␦ degradation. FAF1-overexpressing cells in which Np14 was knocked down did not recover the ability to promote the degradation rate of CD3␦, whereas FAF1-overexpressing control cells did (Fig. 8). These results provide further evidence that FAF1 plays a role in ERAD in a VCP-Npl4-Ufd1-dependent manner. Fig. 9 depicts our model showing how VCP-Npl4-Ufd1 complex selectively interacts with FAF1 through UBX domain to form FAF1-(VCP-Npl4-Ufd1) complex and that this complex formation is required for recruiting polyubiquitinated substrates to FAF1 UBA (Fig. 9A). Only intact FAF1 complex interacting with polyubiquitinated substrates via UBA domain and with VCP-Npl4-Ufd1 via UBX domain can promote ERAD by targeting the proteasome. Interactions between FAF1 and various proteasome subunits were identified (data not shown) (Fig.  9B). This model describes our hypothesis that FAF1 is a ubiquitin receptor that delivers ERAD substrates to the proteasome through a cooperative mechanism between UBX and UBA domains.

DISCUSSION
We demonstrated here that binding of FAF1 to VCP-Npl4-Ufd1 complex via UBX domain is a prerequisite to the binding of polyubiquitinated proteins to UBA domain of FAF1 and that only intact FAF1 complex plays a role in ERAD. Our study shows that VCP N-domain interacts with C-terminal UBX domain of FAF1, thereby facilitating the recruitment of polyubiquitinated proteins to N-terminal UBA domain of FAF1. Fig.  3A, which depicts the crystal structure of the FAF1 UBX⅐VCP N-domain complex, shows that the major interaction between the two involves the S3/S4 loop of the FAF1 UBX domain. ITC results also confirmed that 618 TFPR 621 is critical for VCP binding because the 618 TFPR 621 3 AG mutation showed no detectable binding between the two. Furthermore, the crystal structure studies revealed that the cis-Pro 620 -centered ␤-turn of 618 TFPR 621 is important in the interaction.
We also showed that the Npl4-Ufd1 heterodimer is essential for VCP-FAF1 interaction. No direct binding of FAF1 UBX domain to full-length VCP was observed in ITC studies. This was true even when UBX domain was extended further to include a UAS domain. However, when VCP and Npl4-Ufd1 complex were used, the FAF1 UBX and UAS-UBX domain titration showed binding at a molar ratio of about 1:6 with K D values of 9.6 and 19.5 M, respectively. The EM study (Fig. 5) also clearly showed a single gold bead per VCP-Npl4-Ufd1-FAF1 complex, suggesting one FAF1 bound to the VCP-Npl4-Ufd1 complex. These results demonstrate that full-length FAF1 binds at a 1:6 molar ratio with full-length VCP but not at a 1:1 molar ratio, i.e. one FAF1 binding to a hexameric unit of VCP. Furthermore, FAF1 binding to VCP required prior binding of Npl4-Ufd1 to VCP.
Intriguingly, FAF1(TFPR 3 AG), a mutant deficient in VCP binding via C-terminal UBX, could not recruit polyubiquitinated proteins to N-terminal UBA domain (Fig. 2). We also showed by monitoring direct domain-domain interactions of various FAF1 mutants such as UBL and UBX deletion mutants by ITC that FAF1 UBA domain does not directly interact with UBL1 and UBL2 or UBX domains. Together with our previous findings that UBA domain is crucial for FAF1-mediated apoptosis and proteasomal inhibition (1) and that UBA domain facilitates Lys 48 -linked polyubiquitinated protein recruitment (2), we can conclude that UBA is the main functional domain of FAF1 and that UBA is regulated by UBX binding to VCP already bound to Npl4-Ufd1 heterodimer. However, there is no direct interaction between UBA and UBX domains.
Because UBX domain is module that binds to VCP, many UBX-containing proteins can promote VCP-mediated processes by changing the cofactors. In yeast, membrane-bound UBA-UBX protein Ubx2 membrane-bound UBA-UBX protein, recruits Cdc48-Npl4-Ufd1 to the ER membrane to perform ERAD (35)(36)(37). p47, another UBX protein, acts as a cofactor for p97-mediated membrane fusion by forming a p97-p47 complex (38). A recent study showed that various UBA-UBX proteins bind to ubiquitin ligases and that UBXD7 promotes binding of HIF1␣ to p97 by selectively interacting with von Hippel-Lindau tumor suppressor (pVHL) (32). How VCP performs its specific functions together with these various cofactors should be further studied. This concept of cooperative regulation between domains (UBX-UBA) in a ubiquitin receptor has not been reported before.
We examined the role of FAF1 in ERAD because FAF1 interacts with VCP-Npl4-Ufd1 complex, which is known to be involved in ERAD. ERAD was measured in cells that either overexpressed FAF1 or in which FAF1 was knocked down by monitoring degradation of CD3␦, an ERAD substrate. Overexpression of FAF1 promoted ERAD, whereas depletion of FAF1 retarded it (Fig. 7, A and B). These effects were examined again in cells in which FAF1 was knocked down. Wild type FAF1, FAF1(82-650) (UBA domain deletion mutant), and FAF1(TFPR 3 AG) (a mutant deficient in the interaction with VCP and polyubiquitinated proteins) were reintroduced in these cells. Although FAF1 WT accelerated CD3␦ degradation,

FAF1 Regulates ERAD via VCP-Npl4-Ufd1 Complex
FAF1 UBA domain mutants FAF1(82-650) and FAF1 I41N lost the ability to promote ERAD because they could not recruit polyubiquitinated substrates even though they could interact with VCP. FAF1 UBX domain mutants FAF1(TFPR 3 AG) and FAF1⌬UBX failed to interact with VCP, resulting in no recruitment of polyubiquitinated substrates to UBA domain, and could not promote ERAD ( Figs. 2A and 7D). This suggests that FAF1 UBX-(VCP-Npl4-Ufd1) interaction is required for recruiting polyubiquitinated proteins to FAF1 UBA domain and that UBX domain ensures polyubiquitinated substrate recruitment to UBA domain through VCP-Npl4-Ufd1 complex rather than by inactivating UBA domain through direct interaction. This was confirmed in cells in which Npl4 was knocked down (Fig. 8). Because most cellular Npl4 interacted with FAF1 in immunoprecipitation with anti-FAF1 antibody (data not shown), Npl4 appears to be a major regulator of FAF1. We examined the effect of FAF1 on ERAD in cells in which Npl4 was knocked down. Effects of FAF1 on ERAD were completely abolished in these cells (Fig. 8). This finding implies that FAF1 exerts its biological function in ERAD by first interacting with VCP-Npl4-Ufd1 complex and then recruiting polyubiquitinated proteins to UBA domain. Specific scaffolding characteristics of FAF1, which are necessary for its biological role in ERAD, are finely regulated via changes in protein-protein interactions. The molecular mechanisms underlying this regulation should be studied further.
So far, UBL-UBA proteins Rad23 and Dsk2 have been proposed as proteasome-targeting factors that mediate ERAD (39,40). It has been shown that UBA domain recruits polyubiquitinated ERAD substrates and that UBL domain interacts with proteasome subunits to deliver substrates. FAF1 also interacts with 26 S proteasome subunits including 20 S core particles (proteasome subunit ␣-3,4,5,6) and parts of 19 S regulatory particles (Rpn1, -2, -5, -12, and Rpt2,3) (data not shown). The present study proposes that FAF1 is a scaffolding ubiquitin receptor involved in ERAD by regulating protein-protein interaction through UBA and UBX domains and delivering ERAD substrates to the proteasome. However, the proteasome-interacting motif has yet to be identified.
FAF1 was reported to suppress NF-B activity by interfering with nuclear translocation of the RelA subunit of NF-B and to inhibit IB kinase activation by interacting with p65 and IB kinase (41). A Drosophila homolog of FAF1 called Caspar was also found to inhibit NF-B. Caspar negatively regulates immune deficiency (Imd) responses by blocking nuclear translocation of NF-B (42). However, no role for FAF1 as a ubiquitin receptor in NF-B inactivation has been described, although such a role for FAF1 was demonstrated in other signaling pathways through interaction with various binding partners. Using the artificial ubiquitin-proteasome system substrate ubiquitin-X-GFP, we previously found that FAF1 inhibits proteasomal degradation through its UBA domain (1). In this study, we demonstrated that FAF1 UBA domain specificity is ensured by UBX domain through VCP-Npl4-Ufd1 complexation. FAF1 seems to promote ERAD, whereas nonspecific ubiquitin-proteasome system substrates retard degradation. Nonetheless, the substrate specificity of FAF1 UBA domain is hard to explain because many types of polyubiquitinated proteins including Hsp70 and ␤-catenin have been shown to interact with FAF1 UBA, and their degradation is regulated by FAF1 (4). One possibility is that the specific UBX-interacting complex provides the substrate specificity of UBA domain by delivering specific polyubiquitinated proteins to FAF1.
In summary, both in vivo and in vitro data suggest that VCP-Npl4-Ufd1 complex selectively interacts with UBX domain of FAF1, which in turn regulates the recruitment of polyubiquitinated substrates to FAF1 UBA domain. We propose that FAF1-UBA domain is a ubiquitin receptor for ERAD regulated by the interaction between UBX and VCP-Npl4-Ufd1 complex. The multiple functions of FAF1 should be investigated in further studies by examining the interaction between UBX and VCP- Npl4-Ufd1 complex and characterizing the various complexes of FAF1 formed inside cells in response to various stresses.