|Home | About | Journals | Submit | Contact Us | Français|
RFP2, a gene frequently lost in various malignancies, encodes a protein with RING finger, B-box, and coiled-coil domains that belongs to the RBCC/TRIM family of proteins. Here we demonstrate that Rfp2 is an unstable protein with auto-polyubiquitination activity in vivo and in vitro, implying that Rfp2 acts as a RING E3 ubiquitin ligase. Consequently, Rfp2 ubiquitin ligase activity is dependent on an intact RING domain, as RING deficient mutants fail to drive polyubiquitination in vitro and are stabilized in vivo. Immunopurification and tandem mass spectrometry enabled the identification of several putative Rfp2 interacting proteins localized to the endoplasmic reticulum (ER), including valosin-containing protein (VCP), a protein indispensable for ER-associated degradation (ERAD). Importantly, we also show that Rfp2 regulates the degradation of the known ER proteolytic substrate CD3-δ, but not the N-end rule substrate Ub-R-YFP (yellow fluorescent protein), establishing Rfp2 as a novel E3 ligase involved in ERAD. Finally, we show that Rfp2 contains a C-terminal transmembrane domain indispensable for its localization to the ER and that Rfp2 colocalizes with several ER-resident proteins as analyzed by high-resolution immunostaining. In summary, these data are all consistent with a function for Rfp2 as an ERAD E3 ubiquitin ligase.
The RBCC/TRIM (RING finger, B-box, coiled-coil/tripartite motif) group of proteins is an evolutionarily conserved family with more than 60 members. RBCC proteins are involved in diverse cellular processes, such as apoptosis (Kimura et al., 2003 ), proliferation (Urano et al., 2002 ), differentiation (Lalonde et al., 2004 ), and transcriptional regulation (Wang et al., 2004 ). One functionally uncharacterized RBCC member is the Rfp2 protein (also termed Leu5 and Trim13). The RFP2 gene was initially cloned during the search for genes located in the commonly deleted 13q14 locus in B-CLL (B-cell chronic lymphocytic leukemia). 13q14 deletions or translocations are present in ~50% of all CLL cases (Fitchett et al., 1987 ; Liu et al., 1997 ; Stilgenbauer et al., 2000 ; Mertens et al., 2002 ). Considering all genes in the minimal deleted region, RFP2 is the only gene with protein encoding potential and might therefore be regarded as a prime tumor suppressor gene candidate within the CLL deletion locus at 13q14 (Corcoran et al., 2004 ). Of note is that this locus is also targeted for deletion in other hematological malignancies such as myeloma (Shaughnessy et al., 2000 ; Elnenaei et al., 2003 ; Schmidt-Wolf et al., 2006 ), mantle cell lymphoma (Bigoni et al., 2001 ), and diffuse large cell lymphoma (Wada et al., 2000 ). Deletions have also been reported in several solid tumors such as head and neck tumors (Maestro et al., 1996 ) and oral tumors (Ogawara et al., 1998 ).
The RFP2 gene was recently shown to have coding potential for two separate open reading frames (ORFs), the RBCC protein Rfp2/Leu5 and Dltet (Corcoran et al., 2004 ), a protein with high similarity to BTB/POZ domain–containing proteins. Structurally, RBCC proteins generally contain a RING (Really Interesting New Gene) finger domain, one or two B boxes, and a coiled-coil domain (Reymond et al., 2001 ; Torok and Etkin, 2001 ). RING finger domains are cysteine- and histidine-rich motifs that bind two zinc ions. They mediate protein–protein interactions (Borden, 2000 ) and are frequently involved in proteolysis (Pickart, 2001 ). B-boxes are domains with as of yet unknown function with the ability to bind one zinc ion (Borden, 1998 ). Like RING fingers, coiled-coil domains also mediate protein–protein interactions and have furthermore been shown to be important for homo- and heterodimerization of RBCC family members (Jensen et al., 2001 ). Often, it seems that the entire RBCC domain is needed for proper function and intracellular localization of proteins containing these motifs, indicating that this multidomain structure acts as one functional unit (Reymond et al., 2001 ).
A limited number of the RBCC proteins have been implicated in proteolysis, acting as E3 ubiquitin ligases by promoting the attachment of the conserved polypeptide ubiquitin to specific substrates targeted for degradation (Trockenbacher et al., 2001 ; Urano et al., 2002 ; Vichi et al., 2005 ). E3 ligases can be divided into two major families: HECT (Homologue of E6-AP C-terminus) domain E3s and RING finger domain E3s (Pickart, 2001 ). RBCC proteins with E3 ligase activity accordingly form a subset of the latter family.
One important role of the ubiquitin-proteasome system is to mediate the retrotranslocation and turnover of membrane and secretory proteins from the ER through a set of processes collectively termed ERAD (Meusser et al., 2005 ). As a general surveillance mechanism, cells utilize the ERAD machinery to degrade proteins that are misfolded (McCracken and Brodsky, 2005 ; Denic et al., 2006 ). Additionally, proteins usually part of multisubunit complexes are similarly degraded when not assembled together with their normal oligomerization partner(s). An example of the latter is the degradation of individual subunits of the T-cell receptor (Yang et al., 1998 ). Once a protein is targeted for ERAD, E2-conjugating enzymes specific to the ER and ER-associated E3 ubiquitin ligases, such as Hrd1 and gp78, cooperate to attach ubiquitin molecules to the target (Fang et al., 2001 ; Botero et al., 2002 ; Kikkert et al., 2004 ). With the help of the Cdc48/VCP-Ufd1-Npl4 complex of proteins, the ubiquitinated substrate is finally extracted from the ER membrane and degraded by the 26S proteasome (Ye et al., 2001 ; Jarosch et al., 2002 ). Protein degradation at the level of the ER has attracted increased attention because ERAD has been implicated in a wide variety of conformational diseases, such as Parkinson's disease and Alzheimer's disease, as well as in cancer (Takahashi and Imai, 2003 ; Khan et al., 2004 ; Lindholm et al., 2006 ; Nakamura et al., 2006 ). Still, the molecular mechanisms underlying this process are poorly understood and a limited number of E3 ubiquitin ligases of the ER have been described.
Given the important biological function of other RBCC proteins and the frequent loss of the RFP2 gene in CLL and other malignancies, a functional characterization of the gene is highly warranted. In the present study we present evidence that Rfp2 belongs to a previously uncharacterized subgroup of RBCC proteins containing a transmembrane domain, a domain necessary for Rfp2 to properly associate with the ER. We further demonstrate that Rfp2 can regulate the turnover of a heterologous ERAD substrate, CD3-δ, and that wild-type Rfp2 has auto-polyubiquitination activity in vitro. In conclusion, these results together define Rfp2 as a novel ER E3 ubiquitin ligase and link Rfp2 to the very limited set of proteins currently known to function as effectors of ERAD.
HEK293 and U2OS cells were maintained in DMEM containing 10% FBS, 2 mM glutamine, 100 U/ml penicillin, and 100 g/ml streptomycin.
For transfection experiments, HEK293 and U2OS cells were transfected with TransIT-LT1 Transfection Reagent (Mirus, Madison, WI) according to the manufacturer's protocol. RNA interference (RNAi) experiments were performed with TransIT-TKO Transfection reagent according to the manufacturer's instructions (Mirus). Small interfering RNA (siRNA) duplexes targeting either a region in exon 3 of Rfp2 or a region in exon 1 of Dltet were designed and purchased from Dharmacon (Boulder, CO; http://www.dharmacon.com/sidesign). As controls, scrambled siRNAs or siRNA duplexes against green fluorescent protein were used. siRNA sequences can be obtained from the authors upon request.
If not otherwise indicated, cells were harvested 48 h after transfection and subsequently lysed in M-RIPA buffer (150 mM NaCl, 1% Nonidet P-40, 0.5% deoxycholic acid, 0.05% SDS, 50 mM Tris, pH 7.4). Lysis buffer was supplied with Complete protease inhibitor cocktail (Roche, Indianapolis, IN), 1 mM phenylmethylsulphonylfluoride (PMSF), 1 mM sodium orthovanadate, 5 mM sodium fluoride (Sigma, St. Louis, MO). Lysates were subjected to SDS-PAGE and transferred to PVDF membranes, and the proteins were detected by Western blot analysis using an enhanced chemiluminescence system (ECL, Amersham Biosciences, Piscataway, NJ).
For immunoprecipitation experiments, cells were lysed in M-RIPA or NP-40 (150 mM NaCl, 1% Nonidet P-40, 50 mM Tris, pH 7.4) containing 10 mM N-ethylmaleimide (Pierce, Rockford, IL). One milligram of precleared lysate was incubated with Rfp2 antibody (500 ng) at 4°C overnight, followed by incubation with Gammabind G Sepharose beads (Amersham Biosciences) for 1 h. The beads were washed five times with lysis buffer and immunoprecipitated proteins were subjected to SDS-PAGE and Western blot analysis. For glutathione S-transferase (GST) pulldown, 1 mg of total protein lysate was incubated with glutathione Sepharose 4B (Amersham Biosciences) at 4°C overnight. The resin was washed five times and analyzed as described above. For GST-pulldown in denaturing conditions, cells were lysed in NP-40 buffer containing 1% SDS to disrupt noncovalent interactions. The lysate was subsequently diluted 10 times with NP-40 buffer without SDS, and pulldown was performed as described above.
For protein stability experiments, cells were treated with 30 μM MG-132 (Biomol, Plymouth Meeting, PA) for the indicated time points. For cycloheximide chase experiments, cells were treated with 100 μg/ml cycloheximide (Biomol) in the presence or absence of MG-132. Cells were harvested at the indicated time points.
Cell cycle distribution of Rfp2 expression was examined by flow cytometry analysis as described previously (Panaretakis et al., 2003 ). DNA was stained by addition of propidium iodide staining solution (Nordic BioSite, Täby, Sweden) 20 min before flow analysis.
The full-length coding regions of the RFP2 and DLTET ORFs were separately PCR amplified with Advantage enzyme (BD Biosciences) and cloned into pEBG GST- and Tag2B FLAG-tagged vectors (Invitrogen, Carlsbad, CA). The sequence of the cloned cDNA was verified in its entire length by sequencing. The RFP2[C13A] mutant was constructed using the QuikChange II XL Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA). Primers used for cloning can be obtained from the authors upon request. The RFP2-ΔRING deletion mutant was generated by cutting the RFP2 construct with PstI and subsequently religating. The resulting ORF encodes 317 amino acids and completely lacks the RING domain. RFP2-ΔTM was constructed by PCR amplification using the reverse primer deltaTMR (ttttccatcgatgggcttgcaggcaaattagagg) together with a forward primer in frame with the ATG, resulting in the 3′-deletion of 131 amino acids, including the transmembrane domain.
Untagged pME18S-FL3-RFP2 expression constructs were kindly provided by Dr. Akio Matsuda (Laboratory for Biology, Institute for Life Science Research, Health Care Company, ASAH1 KASEI Corporation, Shizuoka, Japan). ECFP-VCP plasmid was kindly provided by Dr. Florian Salomons (Department of Cellular and Molecular Biology, Karolinska Institute, Solna, Sweden). CD3-δ-YFP and Ub-R-YFP expression constructs were kindly provided by Dr. Nico Dantuma (Department of Cellular and Molecular Biology, Karolinska Institute, Solna, Sweden).
The following primary antibodies were used: rabbit polyclonal against Rfp2 (see below), mouse monoclonal against GST (B-14, Santa Cruz Biotechnology, Santa Cruz, CA), rabbit polyclonal against GST (Z5, Santa Cruz Biotechnology), mouse monoclonal against FLAG (M2, Stratagene), rabbit polyclonal against hemaglutinin (HA) (Y11, Santa Cruz Biotechnology), mouse monoclonal against β-tubulin (Sigma), mouse monoclonal against β-actin (Sigma), rabbit polyclonal against GAPDH (Abcam, Cambridge, MA), mouse monoclonal against Bcl-2 (124, DAKO, Carpinteria, CA), mouse monoclonal against c-Myc (9E10, Santa Cruz Biotechnology), mouse monoclonal against green fluorescent protein (GFP; Roche), rabbit polyclonal against GFP (Sigma), and horseradish peroxidase (HRP)-conjugated goat antibody against biotin (Cell Signaling). As secondary antibodies HRP-conjugated anti-rabbit and anti-mouse antibodies (Cell Signaling, Beverly, MA) were used.
To generate a novel mono-specific antibody recognizing Rfp2, a protein fragment from the coiled-coil region of Rfp2 (amino acids 158–300) was chosen because of the low sequence similarity to all other human proteins. Subsequently, primers were designed and used in RT-PCR, resulting in a gene encoding the protein fragment (Agaton et al., 2003 ). Production, purification, and characterization of the protein fragment were performed as previously described (Agaton et al., 2003 ). Antibody purification was performed by a depletion chromatography step, where antibodies directed toward the His-ABP-part of the fusion protein were removed and a second purification allowed for capture of the specific antibodies (Agaton et al., 2004 ). Characterization of the antibodies was done according to a previous study (Nilsson et al., 2005 ).
Cytosolic, membranous, and nuclear extracts were prepared using the Qproteome Cell Compartment Kit according to the manufacturer's protocol (Qiagen, Chatsworth, CA). Briefly, sequential addition of different buffers to cell pellets followed by incubation and centrifugation results in the isolation of different cellular compartments. First buffer selectively disrupts the plasma membrane without solubilizing it, resulting in the isolation of cytosolic proteins. Plasma membranes and compartmentalized organelles remain intact and are pelleted by centrifugation. The pellet from the first step is resuspended in the second buffer, which solubilizes the plasma membrane as well as all organelle membranes except the nuclear membrane. This fraction contains membrane proteins (including proteins from the ER membrane) and proteins from the lumen of organelles (e.g., the ER and mitochondria). In the final step nuclei are solubilized using the third buffer in which all soluble and most membrane-bound nuclear proteins are extracted. Compartment separation was analyzed by immunodetection of stated marker proteins.
HEK293 cells were transfected with CD3-δ-YFP or Ub-R-YFP together with indicated constructs or siRNAs. After 24 h, transfectants were incubated for 30 min in methionine-/cysteine-free medium and then metabolically pulsed with 70 μCi 35S cell-labeling mix (Redivue ProMix 35S, Amersham Biosciences) at 37°C for 30 min. After the pulse, cells were washed, chased in complete medium supplemented with 10 mM methionine and 1 mM cysteine (Sigma), and harvested at the indicated time points in M-RIPA lysis buffer. CD3-δ-YFP and Ub-R-YFP were immunoprecipitated using rabbit polyclonal GFP antibody (Sigma) and the level of metabolically labeled YFP construct was subsequently analyzed by SDS-PAGE and autoradiography.
To prepare protein extracts for subsequent mass spectrometric analysis, ~5 × 107 HEK293 cells were used. Because Rfp2 was shown to be a strictly membrane-associated protein, soluble cytosolic proteins were discarded in order to reduce complexity of the sample. Washed cell pellets were resuspended in hypotonic lysis buffer (10 mM NaCl, 3 mM MgCl2, 20 mM Tris, pH 7.5), incubated for 10 min with shaking, and centrifuged at 3600 rpm for 10 min. Supernatants containing soluble cytosolic proteins were discarded, and cell pellets were lysed in NP-40 buffer. Endogenous Rfp2 protein was immunoprecipitated as described above using 2 μg of Rfp2 antibody and proteins were subjected to SDS-PAGE with subsequent staining with SimplyBlue Safestain (Invitrogen). In parallel, other control proteins were similarly immunoprecipitated and analyzed. Samples were run in different gels or some distance apart to minimize cross-contamination. Entire lanes were cut into 10-kDa pieces, and proteins were identified by nanoflow liquid chromatography tandem mass spectrometry as described below. Only peptides exclusively present in Rfp2 immunoprecipitated samples were regarded as possible Rfp2 interacting proteins.
All experiments were performed on a 7-tesla LTQ-FT mass spectrometer (Thermo Electron, Bremen, Germany), modified with a nanoelectrospray ion source (Proxeon Biosystems, Odense, Denmark). The high-performance liquid chromatography setup used in conjunction with the mass spectrometer (LC-MS) consisted of a solvent degasser, nanoflow pump, and thermostated microautosampler (Agilent 1100 nanoflow system, Wilmington, DE). Chromatographic separation of peptides was achieved on a 15-cm fused silica emitter (75-μm inner diameter, Proxeon Biosystems) packed in-house with a methanol slurry of reverse-phased, fully end-capped Reprosil-Pur C18-AQ 3 μm resin (Dr. Maisch GmbH, Ammerbuch-Entrigen). Briefly, the tryptic peptides were autosampled onto the packed column at a flow rate of 500 nL/min and then eluted at a flow rate of 200 nl/min using a linear gradient of 4.5–40.5% acetonitrile in 0.5% acetic acid over 90 min and ionized by an applied voltage of 1.8 kV to the emitter.
Analysis was performed using unattended data-dependent acquisition mode, in which the mass spectrometer automatically switches between a high-resolution survey scan (resolution = 100,000 at m/z 400), followed by acquisition of both an electron capture dissociation (ECD; Zubarev et al., 2000 ) and collision activated dissociation (CAD) tandem mass spectra (resolution = 25,000) of the two most abundant peptides eluting at this moment from the nano-LC column. Complementary fragmentation of the same peptide yielded different fragment ions that increased the specificity of the sequence information. Using this complementary information (Horn et al., 2000 ) for protein ID not only improves the confidence in protein identification performed by search engines, but for a fixed confidence level also identifies a larger number of peptides and proteins than when only one fragmentation technique is used (Nielsen et al., 2005 ).
All data from the acquired MS/MS spectra was extracted into so-called dta-files using TurboSequest software (Thermo), and an in-house written Java program was used for extraction of complementary fragment masses before database searching, as described previously (Nielsen et al., 2005 ; Savitski et al., 2005 ). The new dta-files containing complementary fragment masses was merged into a single file, which was searched using the Mascot Search Engine (Matrix Science, Boston, MA; Perkins et al., 1999 ) against the full IPI human database (version 3.15; downloaded February 2006) with carbamidomethyl cysteine as fixed modification and oxidized methionine and ubiquitination on lysine as variable modifications. Peptide searches were performed with an initial tolerance on mass measurement of 3 ppm in MS mode and 0.02 Da in MS/MS mode. Only proteins uniquely identified by a minimum of two significant peptides were taken into consideration for this study.
HEK293 cells were transfected with GST-tagged pEBG, DLTET, RFP2, or RFP2[C13A] expression plasmids. Forty-eight hours after transfection, cells were harvested in LSLD buffer (50 mM NaCl, 0.1% Tween-20, 10% glycerol, 50 mM HEPES) with protease inhibitor cocktail. Two milligrams of total protein was used for GST pulldown as described above. Precipitates were extensively washed (five times each in LSLD buffer and 20 mM Tris, pH 7.5), and in vitro reactions were performed directly on the beads. The in vitro reactions contained 20 ng human recombinant E1, 200 ng E2, 2 μg biotinylated ubiquitin (BioMol), 5 mM ATP, 5 mM MgCl2, 1 mM DTT, 1 U inorganic pyrophosphatase, 10 mM creatine phosphate, 2.5 U creatine phosphokinase from rabbit muscle (Sigma) in 20 mM Tris, pH 7.5, to a final volume of 50 μl. As E2s, UbcH5b and [C85A]UbcH5b were used. [C85A]UbcH5b is a catalytically inactive mutant incapable of forming thiol ester conjugates. The reactions were incubated at 37°C for 60–120 min and stopped by boiling in 1× NuPAGE LDS Sample Buffer (Invitrogen). Ubiquitinated proteins were detected by Western blot analysis using biotin antibodies.
U2OS cells and HEK293 were transfected as described. Cells on coverslips were fixed in 4% formaldehyde for 10 min, permeabilized with 0.2% Triton X-100 in PBS for 10 min, and blocked with blocking buffer (2% BSA, 0.2% Tween-20, 5% glycerol in PBS) for 30 min at room temperature. They were incubated with primary antibodies (anti-Rfp2 diluted 1:100, anti-Myc diluted 1:50) for 1 h at room temperature followed by appropriate secondary antibodies conjugated with fluorescein isothiocyanate (FITC; 1:40; DAKO), Texas Red (1:200; Vector Laboratories, Burlingame, CA) or Alexa Fluor 594 (1:500; Molecular Probes, Eugene, OR). Slides were mounted using Vectashield with DAPI (Vector Laboratories), and images were acquired on a Zeiss Axioplan 2 imaging microscope with Axiovision software (Thornwood, NY) and processed either as gray scale or dual color TIFF images in Adobe Photoshop (Adobe Systems, San Jose, CA). Mock-transfected cells were used as negative controls.
As a first step in characterizing the Rfp2 protein, epitope-tagged constructs of RFP2 were expressed in HEK293 cells by transient transfection. HEK293 cells were also transfected with expression constructs encoding Dltet, the second ORF present within the RFP2 gene locus (Corcoran et al., 2004 ). The expression of transfected GST-tagged Dltet was robust, and the detected amount was roughly equal to that of the GST tag alone. In contrast, the amount of GST-tagged Rfp2 protein was significantly lower (>10 times lower as assessed by quantification of band intensity; Figure 1A). The same phenomenon with low amount of Rfp2 protein was observed with other expression vectors, and in another cell type (HeLa cells, data not shown). One reason for the reduced amount of Rfp2 protein in cells could be that only a small subset of cells expresses Rfp2. However, this possibility was ruled out because flow cytometric DNA histogram analysis in HEK293 cells together with staining for endogenous Rfp2 expression using a specific Rfp2 antibody (see below) revealed that the protein is similarly expressed in all cells, regardless of cell cycle phase (data not shown). To address whether the low levels of Rfp2 protein could be due to high turnover of the protein, we next examined whether Rfp2 levels are increased in cells treated with proteasomal inhibitors. As can be seen in Figure 1B, treatment with the proteasomal inhibitor MG-132 markedly increased Rfp2 protein levels. Together these data show that Rfp2 is an inherently unstable protein regulated by the proteasomal system. In the following experiments, we focused our analysis on characterizing the Rfp2 protein.
A polyclonal antibody directed against the coiled-coil domain of Rfp2 (amino acids 158–300) was generated. The predicted size of endogenous Rfp2 protein is ~47 kDa, and a band of approximately that size was observed in untransfected HEK293 lysates after long exposure, suggesting that the Rfp2 antibody recognizes the low levels of endogenous Rfp2 protein (data not shown). We next silenced Rfp2 expression in HEK293 cells with siRNA oligos specific for Rfp2. As can be seen in Figure 2A, significant reduction of ectopically expressed FLAG-tagged Rfp2 was only observed in cells transfected with Rfp2-specific siRNA oligos, whereas control oligos had no effect on Rfp2 protein levels. Finally, we knocked down endogenous Rfp2 protein in HEK293 cells, confirming the predicted size of Rfp2 and demonstrating the prominent affinity of the antibody for the Rfp2 protein (Figure 2B).
Interestingly, MG-132 treatment or long exposure of filters probed with Rfp2 antibodies revealed the existence of additional distinct bands, of higher molecular weight, extending upward from the Rfp2 band (Figure 1B). In most cells, endogenous Rfp2 protein was detected as two major bands, but in some cases up to three bands were visible (Figures 1B and and3A).3A). This, together with the fact that the Rfp2 protein is regulated by the proteasomal system, prompted us to investigate whether these additional bands represent ubiquitinated conjugates of Rfp2. To this end, we ectopically expressed GST-Rfp2 in parallel with an empty GST-vector in HEK293 cells. Lysates were prepared and Rfp2 protein was immunoprecipitated with Rfp2 antibodies, separated on SDS-PAGE gels and subsequently stained with Coomassie (Figure 3B). Bands of expected and higher molecular weight were observed exclusively in the lane containing transfected Rfp2 protein (indicated by arrowheads). Distinct bands were excised and analyzed by nanoflow liquid chromatography tandem mass spectrometry (LC-MS/MS) as previously described (Shevchenko et al., 2000 ; Nielsen et al., 2005 ). Although the band corresponding to the predicted size of GST-tagged full-length Rfp2 (73 kDa, band 1) only contained tryptic peptides from Rfp2 and GST, both the slower migrating bands (bands 2 and 3) were additionally found to contain peptides of ubiquitin (see table in Figure 3B). These results suggest that Rfp2 is covalently conjugated with single ubiquitin molecules (mono- and/or oligoubiquitination). From these mass spectrometric analyses, however, we were unable to detect any diagnostic branched ubiquitin peptides between Rfp2 and the C-terminus of ubiquitin and could therefore not determine the exact Rfp2 lysine residue(s) conjugated to ubiquitin.
To directly demonstrate the occurrence of mono-ubiquitinated Rfp2 species, cells were cotransfected with GST-tagged Rfp2 and an HA-tagged ubiquitin expression plasmid. Pulldowns of Rfp2 were separated in high-resolution gels and immunoblotted in parallel with antibodies for tagged ubiquitin and Rfp2. As can be seen in Figure 3C, the Rfp2 antibody detected two bands of approximately 8-kDa difference in size. Conversely, the HA-antibody only showed immunoreactivity for the upper, slower migrating band, identifying this band as Rfp2 conjugated to a single ubiquitin molecule.
A common feature of E3 ubiquitin ligases is their ability to auto-polyubiquitinate and degrade themselves, possibly as a precautionary mechanism to limit their levels in the lack of specific substrates (Nuber et al., 1998 ; Canning et al., 2004 ). Because such an activity is an indication of E3 ligase function; we investigated whether Rfp2 is polyubiquitinated in cells. To detect only Rfp2 conjugated with ubiquitin, cells were transfected with GST-tagged Rfp2 together with HA-ubiquitin and lysed under denaturing conditions to disrupt noncovalent interactions. Samples were diluted with lysis buffer, and GST-Rfp2 was precipitated with glutathione beads and resolved on SDS-PAGE, and blots were subsequently hybridized with HA antibodies. As can be seen in Figure 4A, high-molecular-weight smears representing polyubiquitinated Rfp2 were observed under these conditions (Figure 4A, left).
To determine whether the RING domain of Rfp2 is essential for auto-polyubiquitination activity in vivo, we generated a deletion mutant, Rfp2-ΔRING, lacking this domain. Cells were cotransfected with HA-ubiquitin together with untagged full-length Rfp2 or Rfp2-ΔRING. Rfp2 was immunoprecipitated with Rfp2 antibodies, and HA antibodies were used to detect ubiquitinated Rfp2. As can be seen in Figure 4B, full-length Rfp2 was heavily polyubiquitinated, in contrast to the Rfp2-ΔRING mutant that virtually lacked the high-molecular-weight smear (Figure 4B, top panel). These results strongly suggest that the RING domain is required for auto-polyubiquitination activity of Rfp2. In line with the polyubiquitination deficiency of Rfp2-ΔRING, total protein levels of this mutant accumulated several magnitudes above the levels of wild-type Rfp2 protein (Figure 4B, bottom panel). Similar stabilization was observed when a single conserved cysteine in the Rfp2 RING domain was mutated to alanine (Rfp2[C13A]; Figure 4C, left). Additionally, cycloheximide chase experiments demonstrated that the Rfp2[C13A] mutant was stable throughout the incubation period, in contrast to wild-type Rfp2, which was rapidly degraded (Figure 4C, right).
Interestingly, even though the Rfp2-ΔRING mutant (as well as the Rfp2[C13A] point mutant) is deficient in polyubiquitination, it can still be modified with single ubiquitin molecules (Figure 4B, bottom panel, and data not shown). Together, these results indicate that although modification of Rfp2 with single ubiquitin molecules occurs in a RING-independent manner, polyubiquitination and subsequent degradation of Rfp2 requires an intact RING domain.
Rfp2 E3 ubiquitin ligase activity was further examined by a series of in vitro reactions. GST-tagged Rfp2 was incubated in vitro at 37°C together with recombinant human E1, the E2 enzyme UbcH5b, biotinylated ubiquitin, and ATP. As can be seen in Figure 5A, clear accumulation of polyubiquitinated Rfp2 products was observed. In reactions where wild-type UbcH5b was substituted with a mutated, catalytically inactive variant, UbcH5b[C85A], this auto-polyubiquitination was abolished. Moreover, when GST-Rfp2 was replaced with GST-Dltet or GST alone, polyubiquitination did not occur (Figure 5A). Additionally, no polyubiquitination was observed when ATP was omitted, demonstrating the specific ATP-dependent nature of the reaction (data not shown).
To rule out the possibility that in vitro polyubiquitination of Rfp2 is due to E2-only mediated modification of a misfolded Rfp2 protein, we also performed in vitro reactions with the Rfp2[C13A] mutant, lacking one conserved cysteine essential for RING-mediated polyubiquitination. As opposed to wild-type Rfp2, both the RING deletion and the point mutant were found to be deficient in generating polyubiquitinated Rfp2 species (Figure 5B and data not shown), Together, these experiments demonstrate that Rfp2 has E3 ubiquitin ligase activity in vitro, and we conclude that the RFP2 gene encodes an RBCC E3 ubiquitin ligase.
Bioinformatic analysis of the RBCC family reveals that although the N-terminal domains are very similar between different RBCC family members, the C-terminal domains are diverse and include SPRY, NHL, BROMO, and FN3 domains (Figure 6A; Jensen et al., 2001 ; Reymond et al., 2001 ; Meyer et al., 2003 ; Balint et al., 2004 ). Interestingly, bioinformatic analysis of the structure of Rfp2 showed that it lacks these commonly described C-terminal domains, but instead contains a single C-terminal transmembrane domain (present at amino acid positions 319–341) as predicted by TMHMM analysis (http://www.cbs.dtu.dk/services/TMHMM). This particular domain structure (RING, B-box, coiled-coil, transmembrane) has not yet been described in any other RBCC protein, and detailed analysis of all RBCCs (70 in total) shows that this structure is also found in one additional family member, namely TRIM59, the protein most homologous to Rfp2 (Figure 6A). Notably, this particular domain structure was not mentioned in a recent study that subclassified RBCCs according to C-terminal domain composition (Short and Cox, 2006 ). The functionally uncharacterized TRIM59 gene is located at chromosome 3q25, within a >1 Mb region paralogous to 13q14 (Chang et al., 2002 ). The Rfp2 and TRIM59 proteins are evolutionary conserved to the level of Fugu and can thus be considered to form a novel RBCC-transmembrane subgroup.
To examine in detail where Rfp2 is expressed in cells and if it colocalizes with any particular membranous structures, deletion constructs were generated and ectopically expressed in HEK293 and U2OS cells. Although full-length Rfp2 protein localized to the perinuclear area of the cell in a reticular, speckled manner (Figure 6B, top left), a mutant lacking the C-terminal portion including the transmembrane domain, Rfp2-ΔTM, was expressed throughout the cytoplasm (Figure 6B, top right). The localization of Rfp2 was further studied by fractionation of HEK293 cell lysates into cytosolic, membranous, and nuclear compartments. Full-length Rfp2 was strictly present in the membranous and nuclear fractions (Figure 6B, bottom left), in contrast to the Rfp2-ΔTM mutant that was relocalized to the soluble cytosolic fraction (Figure 6B, bottom right), a finding in accordance with the immunofluorescence stainings. Collection of multiple focal planes (z-stacks) obtained from immunofluorescence microscopy clearly demonstrated that the Rfp2 protein is restricted to the nuclear membrane and is absent from the interior of the nucleus (Figure 6B and data not shown). The same membrane-associated pattern was observed when analyzing endogenous Rfp2 protein by fractionation (Figure 6C). These experiments show that Rfp2 is localized to cytoplasmic and nuclear membranes and that this membrane association is dependent on the C-terminus of Rfp2 containing the transmembrane domain.
The speckled cytoplasmic staining observed for Rfp2 has previously not been characterized. In a recent study of more than 30 RBCCs by Reymond et al. (2001) , colocalization studies failed to identify any specific organelles associated with this expression pattern. On the basis of our fractionation results and the fact that the Rfp2 staining pattern is reminiscent of ER localization, we ectopically expressed Rfp2 together with the ER-localized mammalian E2-conjugating enzyme mmUbc6 and stained for both these proteins in U2OS cells. As can be seen in Figure 7A, top panel, Ubc6 was found to colocalize with Rfp2. Specificity of staining was ascertained by using Rfp2 antibodies as well as two different tag antibodies (GST and FLAG) to detect Rfp2 (data not shown). Finally, immunofluorescence analysis of endogenous Rfp2 protein in HEK293 confirmed that it is distributed in a similar manner as ectopically expressed Rfp2 protein in U2OS cells (Figure 7, bottom panel).
Together these results define Rfp2 as a membrane-associated protein that localizes to the ER.
Because Rfp2 localizes to the ER and has ubiquitin ligase activity, we decided to determine whether Rfp2 is involved in ERAD by testing whether it can mediate the turnover of a known ERAD substrate. The CD3-δ subunit of the T-cell receptor complex is one such well-established substrate when expressed in the absence of the other receptor subunits (Klausner et al., 1990 ). HEK293 cells were transiently transfected with CD3-δ-YFP together with wild-type Rfp2 and the Rfp2[C13A] mutant, respectively. Next, cycloheximide chase experiments were performed to examine the turnover of CD3-δ in vivo. As expected, CD3-δ was rapidly degraded in control cells transfected with an empty vector. In contrast, cotransfection of the Rfp2[C13A] mutant significantly stabilized CD3-δ protein (Figure 8A). These results suggest that the Rfp2[C13A] mutant exerts a dominant negative effect, preventing the degradation of CD3-δ. Ectopic expression of wild-type Rfp2 had a slight effect on the total levels of CD3-δ, as evident from reduced levels at time point zero, but the rate of degradation did not seem to be significantly altered. In accordance with a putative role for Rfp2 in CD3-δ degradation, high-resolution immunostaining showed clear colocalization of Rfp2 and CD3-δ in U2OS cells ectopically expressing these proteins (Figure 8B).
Inhibition of protein synthesis by cycloheximide will lead to substantial degradation of Rfp2 itself during the chase period. To circumvent the problem with possible depletion of the Rfp2 pool in the presence of cycloheximide, pulse-chase experiments using metabolically labeled proteins were also performed. As can be seen in Figure 9A, expression of the Rfp2[C13A] mutant stabilized CD3-δ compared with an empty vector control, a finding in line with the cycloheximide experiments. Again, overexpression of wild-type Rfp2 did not appear to have any significant effect on CD3-δ turnover (data not shown). In experiments where CD3-δ was replaced by the N-end rule substrate Ub-R-YFP (Dantuma et al., 2000 ), significant stabilization of the substrate could not be observed in the presence of the Rfp2[C13A] mutant (Figure 9B).
To examine the role of endogenous Rfp2 in CD3-δ turnover, Rfp2 expression was knocked down using siRNA. To avoid possible off-target effects, two different siRNAs targeting Rfp2 were used. Pulse-chase analysis showed that silencing of Rfp2, using either oligo, led to significant stabilization of CD3-δ compared with cells transfected with a scrambled siRNA oligo (Figure 9C). In summary, these experiments demonstrate that Rfp2 functions as a RING-dependent ERAD E3 ubiquitin ligase and regulates the degradation of the ER substrate, CD3-δ.
In an attempt to further define the function of Rfp2 in cells, a mass spectrometric approach was undertaken to identify putative interacting partners for Rfp2. Lysates were prepared from untransfected HEK293 cells and Rfp2-associated protein complexes were immunoprecipitated with Rfp2 antibodies, separated on SDS-PAGE gels, and stained with Coomassie (Figure 10A). Entire lanes were cut into smaller pieces and LC-MS/MS was used to obtain sequence information of putative Rfp2-interacting proteins. In parallel, immunoprecipitates obtained with a set of irrelevant antibodies were also analyzed as controls. Peptides unique for Rfp2 immunoprecipitates were regarded as putative Rfp2-interacting proteins (Figure 10A).
In line with the results described above, Rfp2 immunoprecipitates were found to contain a significant number of peptides from several proteins residing in the ER, including VCP, sarcoplasmic/endoplasmic reticulum calcium ATPase 2 (SERCA2A), dolichol-diphosphooligosaccharide-protein glycosyltransferase (DDOST), and calmodulin (see table in Figure 10A).
VCP, a protein directly linked to ERAD was identified with a very high accuracy due to the detection of a peptide unique for this protein (MDELQLFR with oxidized methionine; data not shown). To confirm the interaction between Rfp2 and VCP, human ECFP-tagged VCP was cotransfected with different GST-tagged Rfp2 constructs in HEK293 cells. GST pulldowns were resolved on SDS-PAGE gels and probed with GFP antibodies (recognizing the ECFP tag). As can be seen in Figure 10B, VCP copurified with wild-type Rfp2 but not with Rfp2 mutants lacking the transmembrane and coiled-coil domains. From this experiment, it appears that the C-terminal part of Rfp2 including the transmembrane domain is necessary for its interaction with VCP. To rule out the possibility of spurious interactions due to overexpression artifacts, the interaction between endogenous Rfp2 and VCP was also tested. ECFP-VCP was transfected alone, endogenous Rfp2 was immunoprecipitated, and the blot was probed with GFP antibodies. The appearance of a distinct VCP band after Western blot analysis showed that VCP protein interacts with endogenous Rfp2, a finding in line with the mass spectrometric analysis, where endogenous VCP coimmunoprecipitated with endogenous Rfp2 (Figure 10C). In summary, these results demonstrate that the Rfp2 protein interacts with the ER-resident protein VCP, an ATPase indispensable for ERAD.
Members of the RBCC family of proteins have been shown to be involved in a number of critical intracellular processes including differentiation, apoptosis, and tumorigenesis. In many instances both the specific enzymatic function and the intracellular localization of the protein was found to be crucial for appropriate biological function. In this study we show that Rfp2, a member of the RBCC/TRIM family of RING finger proteins, is a novel ERAD E3 ubiquitin ligase.
To date, only a few mammalian ER-associated E3 ubiquitin ligases, most containing RING and transmembrane domains, have been identified including MARCH-VI, Hrd1, and gp78/AMFR (Fang et al., 2001 ; Kikkert et al., 2004 ; Hassink et al., 2005 ; Kreft et al., 2006 ). Rfp2 is thus added to the limited, but growing, number of RBCC E3s (Trockenbacher et al., 2001 ; Urano et al., 2002 ; Horn et al., 2004 ; Kallijarvi et al., 2005 ; Vichi et al., 2005 ). Until now, none of the characterized RBCCs have been described as ER-resident proteins. The finding that Rfp2 is an E3 ligase localized to this compartment is substantiated by a series of observations. First, Rfp2 is demonstrated to be a membrane-bound protein (Figures 6, B and C). Second, it colocalizes with several ER-resident proteins, including Ubc6 and the T-cell receptor subunit CD3-δ. Both of these features are dependent on the C-terminal part of the protein harboring a single putative transmembrane domain (Figures 6B, B,7,7, and and8B).8B). Third, Rfp2 has the ability to mediate the degradation of a well-known ERAD substrate, CD3-δ (Figures 8A and and9,9, A and C). Finally, Rfp2 is found to coimmunoprecipitate with a number of ER-resident proteins, including VCP, a protein with a central role in ERAD (Figures 10, A–C).
Central to the function of the characterized E3 ligases of the ER is their involvement in the unfolded protein response (UPR) and ERAD (Menendez-Benito et al., 2005 ; Meusser et al., 2005 ). Previously, these proteins were mainly implicated in general surveillance processes. However, recent evidence shows that the ERAD pathway is also required for the regulated degradation of specific ER substrates, such as the elimination of HMG-CoA reductase by gp78 (Song et al., 2005 ). It is evident from our results that Rfp2, in common with gp78 and Hrd1, has the ability to target CD3-δ for degradation (Fang et al., 2001 ; Kikkert et al., 2004 ). In contrast to many of the other known ERAD substrates, such as TCRα and CFTR, CD3-δ is a protein of native conformation that is degraded in the ER when it lacks its oligomerization partners (Klausner et al., 1990 ). It is well established that single substrates can be degraded by several different E3s (Brooks and Gu, 2006 ), and it is noteworthy that CD3-δ can be degraded by Rfp2, as well as by both gp78 and Hrd1. This points to a considerable overlap in substrate specificity at the ER. One explanation for this could be that a certain amount of redundancy is required for the cell to cope with substrate overload during times of ER stress. Whether Rfp2 shares other ERAD substrates, such as TCRα or apolipoprotein B100, with gp78 and Hrd1 remains to be determined. Also, elucidation of which substrates that might be specific for Rfp2 will be instrumental in further mapping its biological function.
Two E2 conjugating enzymes have been specifically coupled to the ER, namely the mammalian homologues of yeast Ubc6p and Ubc7p. One fundamental difference between them is that although mmUbc7 (official gene name Ube2g2) needs to be recruited to the ER membrane by a docking protein (Biederer et al., 1997 ; Fang et al., 2001 ), mmUbc6 (official gene name Ube2j2) is permanently anchored to the ER membrane by way of its C-terminus (Lenk et al., 2002 ). Interestingly, we have recently found that endogenous Rfp2 binds ectopically expressed mmUbc6 but not mmUbc7 (data not shown). Whether mmUbc6 represents the bona fide E2 of Rfp2 remains to be formally proven, but these data would support preferential usage of mmUbc6 over mmUbc7 by Rfp2.
The role for Rfp2 in ERAD was further supported by the identification of several ER-resident proteins in an MS screen for Rfp2-interacting proteins (Figure 10A). One of the proteins identified in the screen was VCP, and this interaction was also confirmed by in vivo coimmunoprecipitation of VCP with Rfp2 (Figure 10, B and C). VCP is a multifunctional type II AAA-ATPase that has been intimately linked to proteasomal degradation in general and ERAD in particular (Ye, 2006 ). VCP, in complex with Ufd1 and Npl4, is required to extract ubiquitinated proteins to the cytosol from the ER for proteolysis during ERAD (Ye et al., 2001 ). For example, gp78 cannot mediate degradation of CD3-δ without interaction with VCP (Zhong et al., 2004 ). One possibility is that interaction with VCP dictates substrate specificity for Rfp2. In a recent investigation, a VCP-containing complex was shown to be essential for the function of a yeast homologue to MARCH-VI, in its function to degrade membrane bound, but not soluble substrates in the ER (Ravid et al., 2006 ).
VCP has been reported to bind to ubiquitinated proteins during the process of degradation. Therefore, one might argue that it binds Rfp2 simply because Rfp2 is a short-lived proteasomal substrate. However, the binding of VCP to Rfp2 does not seem to be correlated with the ubiquitination of Rfp2 as the Rfp2-ΔRING mutant, which is deficient in polyubiquitination, was shown to bind VCP as well (data not shown). Furthermore, binding of VCP to Rfp2 does not seem require overexpression of either protein, as immunoprecipitation of endogenous Rfp2 also yielded several VCP peptides in the MS analysis (see table in Figure 10A). From the limited interaction mapping done, it seems that the C-terminal part anchoring Rfp2 to the ER membrane is essential for VCP binding. More detailed mapping is however required to establish how these proteins interact and function together in vivo.
The RFP2 gene has been implicated in the pathogenesis of CLL, multiple myeloma and other malignancies, due to its chromosomal location at 13q14.3, a region frequently deleted in these tumors (Kapanadze et al., 1998 ; Elnenaei et al., 2003 ). Preliminary studies have indicated that overexpression of the dominant negative mutant Rfp2[C13A] increases cell proliferation in HCT116 and HEK293 cells (data not shown). These results are in line with a putative tumor suppressor function of Rfp2 although further experiments are needed to substantiate these findings and to delineate the underlying mechanism.
Recent evidence shows that altered protein degradation through defective ubiquitin-mediated proteolysis underlie many malignant diseases (Pagano and Benmaamar, 2003 ). Proteasome inhibitors have consequently been considered attractive candidates for cancer therapy because of their ability to induce cell cycle arrest and apoptosis (Ludwig et al., 2005 ). In relation to the present study, both CLL and multiple myeloma cells have been demonstrated to be sensitive to proteasomal inhibition with the drug bortezomib (PS341 or Velcade; Kelley et al., 2004 ; Mitsiades et al., 2005 ). The fact that multiple myeloma cells have a high load of misfolded ER proteins as a consequence of the production of large amounts of immunoglobulins has been suggested to be one of the reasons why these cells are highly sensitive to proteasomal inhibitors (Obeng et al., 2006 ). Furthermore, the UPR has been demonstrated to be essential for differentiation of B-cells, underscoring that UPR and ERAD have important roles in physiological processes (Wu and Kaufman, 2006 ). For example, B-cells lacking XBP-1, a critical signaling intermediate in UPR and the induction of ER stress, are unable to differentiate into plasma cells (Reimold et al., 2001 ). The finding that a common genetic aberration in CLL and multiple myeloma results in deletion of an ERAD E3 ubiquitin ligase, namely RFP2, is in this respect intriguing. One speculative possibility is that loss of Rfp2 leads to defects in induction of the UPR and consequently impaired B-cell differentiation. How Rfp2 might be involved in UPR signaling awaits investigation, and identification of target substrates for Rfp2 ubiquitin ligase activity will be crucial to dissect the mechanism behind the putative tumor suppressor function of Rfp2 in CLL and other malignancies.
The authors acknowledge the technical support provided by A. C. Björklund. This work was supported by grants from the Swedish Cancer Society, the Swedish Cancer Foundation, Gustav V Jubileumsfond, The Swedish Research Council, Karolinska Institute Foundations, the Knut and Alice Wallenberg Foundation, and the Leukaemia Research Fund. This work was partly funded by a grant from Pfizer.
This article was published online ahead of print in MBC in Press (http://www.molbiolcell.org/cgi/doi/10.1091/mbc.E06-03-0248) on February 21, 2007.