U box proteins as a new family of ubiquitin-protein ligases.

The U box is a domain of approximately 70 amino acids that is present in proteins from yeast to humans. The prototype U box protein, yeast Ufd2, was identified as a ubiquitin chain assembly factor that cooperates with a ubiquitin-activating enzyme (E1), a ubiquitin-conjugating enzyme (E2), and a ubiquitin-protein ligase (E3) to catalyze ubiquitin chain formation on artificial substrates. E3 enzymes are thought to determine the substrate specificity of ubiquitination and have been classified into two families, the HECT and RING finger families. Six mammalian U box proteins have now been shown to mediate polyubiquitination in the presence of E1 and E2 and in the absence of E3. These U box proteins exhibited different specificities for E2 enzymes in this reaction. Deletion of the U box or mutation of conserved amino acids within it abolished ubiquitination activity. Some U box proteins catalyzed polyubiquitination by targeting lysine residues of ubiquitin other than lysine 48, which is utilized by HECT and RING finger E3 enzymes for polyubiquitination that serves as a signal for proteolysis by the 26 S proteasome. These data suggest that U box proteins constitute a third family of E3 enzymes and that E4 activity may reflect a specialized type of E3 activity.

The abundance of cellular proteins is regulated by a balance between their synthesis and degradation, and such regulation is central to many cellular functions. Two predominant pathways for the degradation of cellular proteins have been identified: the vacuolar pathway (mediated by lysosomes, endosomes, and the endoplasmic reticulum, for example) and the cytoplasmic ubiquitin-mediated pathway. The ubiquitin proteolytic pathway plays an important role in the degradation of short lived regulatory proteins (1), including those that participate in the cell cycle, cellular signaling in response to stress and to extracellular signals, morphogenesis, the secretory pathway, DNA repair, and organelle biogenesis (2,3). This pathway includes two distinct steps as follows: the covalent attachment of multiple ubiquitin molecules to the protein substrate, and the degradation of the ubiquitinated protein by the 26 S proteasome complex. The system responsible for ubiquitin attachment consists of several components that act in concert (4,5). A ubiquitin-activating enzyme (E1), 1 with ATP as a substrate, catalyzes the formation of a thioester bond between itself and ubiquitin and then transfers the activated ubiquitin to a ubiquitin-conjugating enzyme (E2). Whereas some E2s transfer ubiquitin directly to a substrate, others require the participation of a third component, termed a ubiquitin-protein ligase (E3). E3 is thought to be the component of the ubiquitin conjugation system that is most directly responsible for substrate recognition (5,6).
A new class of ubiquitination enzyme (termed E4), the prototype of which is yeast Ufd2, was recently identified (35). Together with E1 (Uba1), E2 (Ubc4), and a HECT-type E3 (Ufd4), E4 (Ufd2) is required for the assembly of a polyubiquitin chain on artificial substrates that consist of proteins fused to ubiquitin at their NH 2 termini and that are preferentially targeted for degradation. In yeast, Ufd2 is implicated in cell survival under stress conditions and is associated with Cdc48, which belongs to the large family of AAA-type ATPases that are thought to possess protein folding activity. Ufd2 and its homologs in other eukaryotes share a conserved domain desig-nated the U box (35). The U box of Ufd2 mediates the interaction of this protein with ubiquitin-conjugated targets and therefore appears to be an essential functional unit of E4 enzymes. The predicted three-dimensional structure of the U box is similar to that of the RING finger, despite the lack in the former of the hallmark metal-chelating residues of the latter (36). This observation prompted us to investigate the possibility that U box proteins in general possess the ability to function as E3s in E2-dependent ubiquitination and that E4 activity might reflect a specialized type of E3 activity that targets the ubiquitinated ubiquitin moiety fused to other proteins for further ubiquitination.
We now show that all six mammalian U box proteins tested mediate ubiquitination in conjunction with E1 and E2 and in the absence of other E3 components. Deletion of the U box domain or mutation of its most conserved amino acid resulted in the loss of E2-dependent ubiquitination activity, suggesting that this activity depends on an intact U box. The U box proteins appeared to catalyze their own ubiquitination as well as that of heterologous substrates. One of these proteins, UFD2a, attached ubiquitin to lysine residues of ubiquitin other than those at positions 29, 48, or 63, a property that is not usually exhibited by HECT or RING finger-containing E3s. These observations suggest that U box proteins are indeed E3s, some of which may function as E4s to mediate the assembly of polyubiquitin chains on proteins ubiquitinated by another E3 enzyme.

EXPERIMENTAL PROCEDURES
Cloning of Mammalian U Box Protein cDNAs-The sequences of mouse UFD2a (accession number AI593754), UFD2b (accession number AI390103), CHIP (accession number NM019719), and KIAA0860 (accession numbers AB020667 and AW476623) cDNAs as well as those of human CYC4 (accession number HSU37219) and PRP19 (accession number XM006045) cDNAs were obtained from the GenBank TM data base. The cDNAs for UFD2a, UFD2b, CHIP, and KIAA0860 were amplified by the polymerase chain reaction (PCR) with Taq polymerase (Takara, Tokyo, Japan) from mouse testis cDNA (CLONTECH, Palo Alto, CA), whereas those for CYC4 and PRP19 were amplified from human thymus cDNA or human brain cDNA, respectively. The PCR primers were as follows: 5Ј-CAG TCA CTA CAG CAT TAC CTG GAG-3Ј  and 5Ј-AGC GAC GAC AAC TAG GCA GAA TAG-3Ј for UFD2a; 5Ј-CCT  TTT GGT CGC TGT AGC TGT GA-3Ј and 5Ј-TTC CAA GTA CAC ACA  ACG GCT GGG-3Ј for UFD2b; 5Ј-GGC TGC GAG ATC TAG GTG-3Ј and  5Ј-CTT CCG TCT CCA GAT CCT-3Ј for CHIP; 5Ј-TGC TTG GAA GCT  ATT GTT CCA GTT-3Ј and 5Ј-TGG CCT GGC TTG GAA GCC CCT  GTG-3Ј for KIAA0860; 5Ј-GGC GCC ATG TCC CTA ATC TGC TCC-3Ј and 5Ј-GGC CTA CAG GCT GTA GAA CTT GAG-3Ј for CYC4; and 5Ј-GGC GCC ATG TCC CTA ATC TGC TCC-3Ј and 5Ј-GGC CTA CAG GCT GTA GAA CTT GAG-3Ј for PRP19. The amplified fragments were phosphorylated with T4 polynucleotide kinase (New England Biolabs, Beverly, MA), subcloned into pBluescriptII SK ϩ (Stratagene, La Jolla, CA), and sequenced. In the case of UFD2a, UFD2b, and KIAA0860, the cDNA fragments were used as probes to isolate full-length clones from a mouse T cell cDNA library in ZAP (Stratagene). The positive clones obtained from of 1 ϫ 10 6 plaques for each screening were isolated, and the inserts were subcloned into pBluescriptII SK ϩ .
To generate U box deletion mutants of UFD2a or CHIP, we performed PCR with pBluescriptII SK ϩ containing UFD2a or CHIP cDNA as a template and with a sense primer beginning at the start codon and an antisense primer beginning proximal to the U box. To generate a U box deletion mutant (⌬U) of KIAA0860, we performed PCR with pBlue-scriptII SK ϩ containing KIAA0860 cDNA as a template and either with a sense primer beginning at the start codon and an antisense primer beginning proximal to the U box, or with a sense primer beginning distal to the U box and an antisense primer beginning at the stop codon. Both amplified products were mixed and subjected to a second PCR with a sense primer beginning at the start codon and an antisense primer beginning at the stop codon. To generate RING finger deletion mutants of KIAA0860, we performed PCR with wild-type or ⌬U mutant KIAA0860 cDNA as a template and with a sense primer beginning at the start codon and an antisense primer beginning proximal to the RING finger domain.
Complementary DNAs encoding mouse UFD2a, UFD2b, CHIP, and KIAA0860 as well as human CYC4 and PRP19 tagged at their NH 2 termini with the FLAG epitope were generated by PCR, sequenced, and subcloned into pcDNA3 (Invitrogen). The plasmids pT7-7-Ubc2A, -Ubc2B, -Ubc3, -Ubc4, -UbcH5C, -UbcH6, -UbcH7, and -UbcH8 (Novagen, Madison, WI) were kindly provided by H. Yasuda (Tokyo Pharmaceutical University), and pCGN-HA-Ub was kindly provided by M. Nakao (Kumamoto University). A cDNA encoding hemagglutinin epitope (HA)-tagged ubiquitin was generated by PCR with pCGN-HA-Ub as template, with a sense primer that includes an EcoRI site, the HA tag sequence, and the beginning of the ubiquitin coding region and with an antisense primer that includes the ubiquitin stop codon and a XhoI site. The resulting PCR product was subcloned into pGEX-6P-1. To generate the various lysine residue mutants of ubiquitin, we performed site-directed mutagenesis with a Quick Change kit (Stratagene) and with mutated oligonucleotide primers corresponding to each lysine site.
Production of Recombinant Proteins in Bacteria-Glutathione S-transferase (GST) fusion proteins were expressed in Escherichia coli strain DH5␣ cultured in the presence of 0.1 mM isopropyl-␤-D-thiogalactopyranoside. Bacterial cells were resuspended in phosphate-buffered saline (PBS) and lysed by sonication, and cellular debris was removed by centrifugation for 20 min at 13,000 ϫ g. Glutathione-Sepharose 4B beads (Amersham Pharmacia Biotech) were added to the resulting supernatant, and the mixture was rotated at 4°C overnight. The beads were washed in PBS, and GST fusion proteins were eluted with 50 mM Tris-HCl (pH 8.0) containing 10 mM reduced glutathione. To generate the ubiquitin mutants, we subjected the purified GST fusion proteins attached to the glutathione-Sepharose 4B beads to digestion for 4 h at 4°C with PreScission protease (Amersham Pharmacia Biotech) in a buffer containing 50 mM Tris-HCl (pH 8.0), 150 mM NaCl, 1 mM EDTA, and 1 mM dithiothreitol. Only the ubiquitin derivatives, not the GST moiety nor the GST-ubiquitin fusion proteins, were detected in the eluate by electrophoresis and Coomassie Blue staining (data not shown).
His 6 -tagged proteins were expressed in E. coli strain BL21(DE3)pLysS (Novagen) incubated in the presence of 0.1 mM isopropyl-␤-D-thiogalactopyranoside. The recombinant proteins were purified with the use of ProBond resin (Invitrogen).
Baculovirus Expression System-Baculoviruses were generated in Sf9 cells with the use of BacPAK6 virus DNA and pBacPAK9-GST containing the relevant cDNA (23). The recombinant proteins were purified from Sf9 cell lysates according to the protocol described for bacterially expressed GST fusion proteins. The plasmid pFASTBAC HTa containing the relevant cDNA was subjected to recombination with the baculoviral genome in HB10BAC, and the resulting recombinant viral genome was introduced into Sf9 cells by transfection in order to generate recombinant baculovirus. The infected Sf9 cells were lysed, and the recombinant proteins were purified by the protocol described for bacterially expressed His 6 -tagged proteins. Only the purified recombinant proteins were detected by electrophoresis and Coomassie Blue staining.
Northern Blot Analysis-Mouse and human multiple-tissue Northern blots (CLONTECH) were subjected to hybridization with U box protein or ␤-actin (mouse or human) cDNA probes that had been labeled with [␣-32 P]dCTP. The blots were washed at 68°C in 0.2ϫ standard saline citrate containing 0.1% SDS and were then exposed and analyzed with a BAS-2000 instrument (Fuji Film, Kanagawa, Japan).
Production of Antibodies-Polyclonal antibodies to UFD2a were generated in rabbits by standard procedures with the use of a synthetic peptide corresponding to residues 129 -142 of mouse UFD2a (MEV-DENDRREKRSL) coupled to keyhole limpet hemocyanin.
Immunofluorescence Staining-HeLa cells grown on glass coverslips were transfected by calcium phosphate precipitation and prepared for immunostaining as described (37). In brief, the cells were fixed for 20 min at room temperature with 4% formaldehyde in PBS and then incubated for 1 h at room temperature with a mouse monoclonal antibody (M5) to the FLAG epitope (1 g/ml) (Sigma) in PBS containing 0.1% bovine serum albumin and 0.1% saponin. They were then incubated for 1 h at room temperature with Alexa 546-labeled goat polyclonal antibodies to mouse immunoglobulin (Molecular Probes, Eugene, OR) at a dilution of 1:500 and stained with Hoechst 33258 (Wako, Osaka, Japan). Cells were covered with a drop of GEL/MOUNT (Biomeda, Foster City, CA) and then viewed and photographed with a Nikon Eclipse E800M microscope equipped with a color chilled 3CCD camera (model C5810; Hamamatsu Photonics, Hamamatsu, Japan).
In some experiments, the ubiquitination reaction was terminated by the addition of 400 l of a solution containing 10 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1% Nonidet P-40, 0.1% sodium deoxycholate, 0.1% SDS, 1 mM EDTA, 10 mM NaF, and 10 mM sodium pyrophosphate. After incubation for 1 h at 4°C with 50 l of protein G-Sepharose beads (Amersham Pharmacia Biotech) and removal of the beads, the mixture was incubated first for 4 h at 4°C with 5 g of mouse monoclonal antibodies to GST (kindly provided by S. Tanaka, Hokkaido University) or antibodies to the FLAG epitope (M5) and then for 4 h at 4°C with protein G-Sepharose beads. The immunoprecipitates were washed four times with an ice-cold solution containing 50 mM Tris-HCl (pH 7.6), 300 mM NaCl, 0.1% Triton X-100, aprotinin (10 g/ml), leupeptin (10 g/ml), 10 mM iodoacetamide, 1 mM phenylmethylsulfonyl fluoride, 0.4 mM Na 3 VO 4 , 0.4 mM EDTA, 10 mM NaF, and 10 mM sodium pyrophosphate and were then subjected to immunoblot analysis with antibodies to ubiquitin, horseradish peroxidase-conjugated antibodies to mouse immunoglobulin, and ECL reagents.
For chelation experiments, glutathione-bound proteins were incubated for 18 h at 4°C with three changes of PBS containing 5 mM TPEN (Sigma). The TPEN-chelated GST fusion proteins were then washed with PBS and eluted from the glutathione-Sepharose 4B beads with 50 mM Tris-HCl (pH 8.0) containing 10 mM reduced glutathione (8).
In Vitro Transcription and Translation-35 S-Labeled FLAG-Nedd4, -UFD2a, and -KIAA0860 were synthesized with the use of a TNT Quick-coupled Transcription/Translation System (Promega) and [ 35 S]methionine (ICN, Costa Mesa, CA). Translation was performed in a 100-l reaction mixture containing 98 l of reticulocyte extract and 2 g of circular plasmid containing the relevant cDNA. To remove ubiquitin and potential fraction I-containing E2 enzymes, we applied the translation mixture containing the labeled protein to DEAE-cellulose (Amersham Pharmacia Biotech) in a solution containing 10 mM Tris-HCl (pH 7.5) and 0.1 mM dithiothreitol. After extensive washing of the resin with 100 mM NaCl, the labeled protein was eluted with 400 mM NaCl and subjected to in vitro assay of self-ubiquitination.

Identification of Mammalian U Box
Proteins-A search of the GenBank TM data base for mammalian proteins that contain a U box domain identified six molecules as follows: mouse UFD2a, UFD2b, CHIP, and KIAA0860 and human CYC4 and PRP19 (Fig. 1A). UFD2a and UFD2b appear to be mammalian homologs of Saccharomyces cerevisiae Ufd2; they contain the U box domain at their COOH termini and, unlike the other four proteins identified, do not possess other known motifs. In addition to their U box domains, CHIP contains three tandem tetratricopeptide repeats (TPRs), KIAA0860 contains a RING finger, CYC4 contains a cyclophilin-like motif, and PRP19 contains six WD40 repeats. Alignment of their amino acid sequences revealed a marked similarity among the U box domains of the yeast and mammalian proteins (Fig. 1B). The U box sequences of UFD2a, UFD2b, CHIP, and KIAA0860 are more closely related to each other than to those of CYC4 and PRP19. The proline residue corresponding to proline 924 of S. cerevisiae Ufd2 is perfectly conserved not only in the six mammalian U box proteins but also in all U box proteins from other organisms that are present in the data base (data not shown).
Expression Analysis of Mammalian U Box Proteins-The tissue specificity of expression of the genes for these mammalian U box proteins was examined by Northern blot analysis of polyadenylated RNA from a variety of mouse or human tissues ( Fig. 2A). The genes for UFD2a, UFD2b, and KIAA0860 are expressed in liver, heart, brain, kidney, and testis. CHIP mRNA was most abundant in skeletal muscle but was also present in smaller amounts in heart, liver, kidney, and pancreas. CYC4 gene transcripts were detected in heart, brain, placenta, skeletal muscle, kidney, and pancreas, and PRP19 mRNA was present in all tissues examined.
The subcellular localization of the mammalian U box proteins was examined by immunofluorescence analysis of HeLa cells transfected with vectors encoding FLAG epitope-tagged versions of these proteins. Cells were also transfected with plasmids encoding FLAG-tagged versions of Skp2 and IKK2 (IB kinase 2) as controls for nuclear and cytoplasmic expression, respectively. Whereas KIAA0860 and CYC4 appeared to be localized to the nucleus of transfected HeLa cells, UFD2b and CHIP were localized to the cytoplasm (Fig. 2B). UFD2a and PRP19 exhibited a nuclear localization in some cells and a cytoplasmic localization in others, suggesting that the localization of these proteins might be regulated in a cell cycle-dependent manner.
Ubiquitin-Protein Ligase Activities of Mammalian U Box Proteins-To determine whether the mammalian U box proteins mediate ubiquitination, we subjected them to an in vitro assay with recombinant proteins produced in E. coli, which does not express components of the ubiquitin-conjugating system, thereby avoiding contamination with or copurification of eukaryotic proteins such as E3s (8). The U box proteins were expressed as GST or histidine (His 6 )-tagged fusion constructs and were isolated by affinity chromatography. Recombinant ubiquitin, recombinant E1, and E. coli lysate containing recombinant E2 (UbcH5C) were added to the assay. Potential substrates included the added fusion proteins and bacterial proteins. UFD2a and Nedd4 (an E3 of the HECT family) each mediated polyubiquitination in the presence of both E1 and E2 (Fig. 3A). To examine the requirements for E1, E2, UFD2a, and ubiquitin in the ubiquitination reaction, we performed in vitro ubiquitination assays with the various combinations thereof in which one component (E1, E2, UFD2a, or ubiquitin) was missing from the reaction mixture (Fig. 3B). All components were required for polyubiquitination. Some of the ubiquitinated species detected were smaller than UFD2a (GST-FLAG-UFD2a, 162 kDa), suggesting that they were derived from bacterial proteins, although it remains possible that the recombinant protein might be degraded during the reaction. Given that immunoblot analysis with antibodies to the FLAG epitope revealed a shift in the mobility of the band corresponding to GST-FLAG-UFD2a, UFD2a was itself likely a target of polyubiquitination ( Fig. 3B; see also Figs. 6B and 7). To confirm that bacterial proteins were ubiquitinated in the reaction, we subjected the reaction mixture after completion of the incubation to immunodepletion with antibodies to GST (for GST-FLAG-UFD2a), and the resulting supernatant was then subjected to immunoblot analysis with antibodies to ubiquitin (Fig. 3C). Although the GST-FLAG-UFD2a protein was almost completely removed by the immunodepletion procedure, the remaining supernatant still contained a substantial amount of ubiquitinated protein. We thus conclude that the ubiquitinated species detected in the assay are derived from both bacterial proteins and UFD2a itself. On the other hand, the autoubiquitination of E2 (UbcH5C) was not apparent (Fig. 3B; see also Fig. 4B). Like UFD2a, UFD2b (see Fig. 4A), CHIP (Fig. 3, D-F), KIAA0860 (Fig. 3, E and F), as well as CYC4 and PRP19 (see Fig. 4A) also mediated polyubiquitination in a manner dependent on the presence of both E1 and E2. Given the absence of other E3 factors from the assay mixture, these U box proteins, like Nedd4 and Mdm2 (RING finger-containing E3) (Fig. 3F), by definition function as E3 enzymes.
To exclude the possibility that the apparent E3 activity of the U box proteins was nonspecific and attributable to their presence at physiologically irrelevant concentrations, we compared the ubiquitination activities of UFD2a, CHIP, and KIAA0860 with those of equimolar amounts of Nedd4 (Fig. 3E). The E3 activities of these U box proteins were similar to (CHIP and KIAA0860) or greater than (UFD2a) that of Nedd4.
The ubiquitination activity of the RING finger-containing Mdm2 requires divalent cations, as revealed by the loss of activity induced by chelation with TPEN, a potent chelator for Zn 2ϩ (Fig. 3F). Such chelation did not inhibit the activities of Nedd4 (a HECT family E3) or E3s containing a U box domain (Fig. 3F), even though the predicted structure of this domain resembles that of the RING finger (36). The observation that KIAA0860, which contains both a U box and a RING finger, was resistant to TPEN treatment suggested that the RING finger domain of this protein is dispensable for ubiquitination activity.
E2 Preference of Mammalian U Box Proteins-The variety of the possible combinations of E2 and E3 may be relevant to the diversity of substrate specificity for ubiquitination. However, most E3s exhibit specificity for a relatively narrow range of E2 enzymes. We thus examined the preference of the mammalian U box proteins for eight different E2 enzymes in the presence of E1 (Fig. 4A). The ubiquination activity of UFD2a, UFD2b, CHIP, and KIAA0860 was greatest with Ubc4 or UbcH5C, whereas only Ubc3 cooperated with PRP19. CYC4 exhibited greatest activity with Ubc2B or Ubc3 but was also active with UbcH7.
To rule out the possibility of ubiquitination of E2 in the in vitro ubiquination reaction, we performed immunoblot analysis with antibodies to the His 6 tag. Only a low level of E2 ubiquitination was apparent, with Ubc4 and UbcH5C showing virtually no susceptibility to ubiquitination; in contrast, as shown above (Fig. 3, C and D), immunoblot analysis with antibodies to the FLAG epitope revealed a marked mobility shift for UFD2a (Fig. 4B), resulting from the self-ubiquitination of this protein.
Requirement of the U Box Domain for E3 Activity-To determine whether the U box is essential for ubiquitination by mammalian U box proteins, we introduced deletions or point mutations into UFD2a (Fig. 5A) and CHIP (Fig. 5C). Deletion were incubated for 2 h at 30°C in the presence of rabbit E1 and bacterial lysate containing E2 (UbcH5C). As a negative control, an E3 component was omitted from the reaction mixture (mock). The reaction mixture was resolved under reducing conditions by SDS-PAGE, and the separated proteins were subjected to immunoblot analysis (IB) with antibodies to ubiquitin (anti-Ub). B, E1, E2, and ubiquitin dependence of the polyubiquitination activity of UFD2a. In vitro ubiquitination assays were performed with the indicated combinations of E1, E2 (His 6 -UbcH5C), ubiquitin, and GST-FLAG-UFD2a (top panel). The reaction mixture was also subjected to immunoblot analysis with antibodies to the His 6 tag (middle panel) or to the FLAG epitope (bottom panel) in order to detect UbcH5C and UFD2a proteins, respectively. C, polyubiquitination of bacterial proteins by UFD2a. In vitro ubiquitination assays were performed as in B, after which the reaction mixture was subjected to immunoprecipitation (IP) with antibodies to GST. The resulting supernatant (upper right panel) or the intact reaction mixture (upper left panel) was then subjected to immunoblot analysis with antibodies to ubiquitin (Ub). Immunodepletion of GST-FLAG-UFD2a was confirmed by immunoblot analysis of the corresponding samples with antibodies to the FLAG epitope (lower panels). D, E1-and E2-dependent polyubiquitination by CHIP. Equal amounts of GST-Nedd4, GST-UFD2a, or GST-CHIP were incubated for 2 h at 30°C in the absence or presence of rabbit E1 or bacterial lysate containing E2 (UbcH5C), as indicated. As a negative control, an E3 component was not added to the reaction mixture (mock). The reaction mixture was resolved under reducing conditions by SDS-PAGE, and the separated proteins were subjected to immunoblot analysis with antibodies to ubiquitin. E, comparison of the efficiencies of polyubiquitination mediated by Nedd4, UFD2a, CHIP, and KIAA0860. The activity of each GST fusion protein (10, 3, and 1 pmol) was assayed in the presence of E1 and E2 as described in A and B. Coomassie Blue (CB) staining of the gel is shown below the immunoblot (IB). Asterisks indicate the positions of the E3 proteins. F, effect of divalent cation chelation on the ubiquitination activities of Mdm2, Nedd4, CHIP, and KIAA0860. Bead-bound GST fusion proteins were incubated in the absence (Ϫ) or presence (ϩ) of 5 mM TPEN before assay of ubiquitination activity in the presence of E1 and E2. The positions of the GST-E3 fusion proteins are indicated on the right of the Coomassie Blue-stained gel.
of the U box domain (⌬U) or point mutation (P1140A) of the conserved proline at position 1140 abolished the E3 activity of UFD2a (Fig. 5B). Similarly, CHIP mutants lacking the U box domain or with a P270A point mutation failed to exhibit polyubiquitination activity (Fig. 5D). Furthermore, substitution (H261A) of the histidine residue at position 261 of CHIP, which corresponds to the residue (tryptophan 408) of Cbl, a RING finger-type E3, implicated in E2 binding (17), also resulted in the loss of E3 activity.
To determine whether the U box in its middle portion or the RING finger at its COOH terminus is responsible for the E3 activity of KIAA0860, we generated a series of mutants of this protein (Fig. 5E). Whereas deletion of the U box domain (⌬U) abolished E3 activity, deletion of the RING finger (⌬R) did not affect ubiquitination, indicating that the U box, not the RING finger, is responsible for E3 activity of KIAA0860 (Fig. 5F). This observation is consistent with the resistance of KIAA0860 to TPEN treatment (Fig. 3F). Mutagenesis (P306A) of the conserved proline at position 306 of KIAA0860 also abolished E3 activity, whereas activity was unaffected by substitution (C292A) of cysteine 292. Collectively, these data indicate that the U box domain is essential for the E3 activity of mammalian U box proteins and that the conserved proline is indispensable for the function of the U box domain.
Polyubiquitination by U Box Proteins of Themselves, as Well as of Heterologous Targets-Proteins that catalyze ubiquitination often undergo automodification. With the use of immunoprecipitation and immunoblot analysis, we first examined the ability of the mammalian U box proteins to facilitate E2-dependent ubiquitination of heterologous substrates (Fig. 6A). After in vitro ubiquitination reactions, the U box proteins were immunoprecipitated and then subjected, together with the intact reaction mixture, to immunoblot analysis with antibodies to ubiquitin. The distinct patterns and extents of ubiquitination apparent between the intact reaction mixture and the U box protein immunoprecipitates suggested that the U box proteins are indeed able to catalyze the ubiquitination of heterologous substrates. To assess further whether the U box proteins themselves also undergo ubiquitination, we prepared 35 S-labeled versions of these proteins by in vitro translation in a reticulocyte lysate and then tested them in the ubiquitination assay (Fig. 6B). As with Nedd4, a substantial proportion of UFD2a and KIAA0860 exhibited a shift in electrophoretic mobility toward the top of the gel, indicative of automodification by polyubiquitination. These results thus indicated that U box proteins serve as E3s to mediate ubiquitination of themselves as well as of heterologous substrates.
Unusual Lysine Residue Specificity for Polyubiquitination by Mammalian U Box Proteins-Given that yeast Ufd4 and Ufd2 require lysine residues 29 and 48 of the ubiquitin moiety of the substrate to mediate its polyubiquitination (35), we examined which lysine residues of ubiquitin are required for polyubiquitination by mammalian U box proteins. Neither Nedd4 nor Mdm2 was able to mediate ubiquitination with a ubiquitin mutant (K48R) in which lysine 48 was replaced with arginine ( Fig. 7A), consistent with the notion that these E3s conjugated the COOH-terminal glycine of one ubiquitin molecule to the internal lysine 48 of the adjacent ubiquitin molecule. The dependence on lysine 48 of ubiquitin appeared to differ among the mammalian U box proteins. Whereas UFD2b, CYC4, and PRP19 were not able to function with the K48R mutant of ubiquitin (Fig. 7A), the extent of ubiquitination mediated by UFD2a appeared unaffected or even slightly increased with the K48R mutant. CHIP and KIAA0860 were each able to mediate ubiquitination with the K48R mutant ubiquitin, although with a markedly reduced efficiency (Fig. 7A).
The lysine residue specificity of UFD2a was further characterized. The ubiquitination of 35 S-labeled UFD2a was thus monitored with ubiquitin mutants in which the lysine residues at positions 6, 11, 27, 29, 33, 48, or 63 were individually substituted with arginine (Fig. 7B). All of the ubiquitin mutants supported polyubiquitination of UFD2a, although the extent of polyubiquitination with K6R and K27R mutants was slightly FIG. 4. E2 preference of mammalian U box proteins. A, in vitro ubiquitination assays were performed with mammalian U box proteins in the presence of E1 and in the absence (mock) or presence of bacterial lysates containing the indicated His 6 -tagged E2 proteins (1 l). The reaction mixture was resolved under reducing conditions by SDS-PAGE, and the separated proteins were subjected to immunoblot analysis with antibodies to ubiquitin. Anti-Ub, anti-ubiquitin. B, in vitro ubiquitination reactions were performed with UFD2a, after which the reaction mixture was resolved under reducing conditions by SDS-PAGE, and the separated proteins were subjected to immunoblot analysis either with antibodies to the His 6 tag for E2s (upper panel) or with antibodies to the FLAG epitope for UFD2a (lower panel).
reduced compared with that apparent with the wild-type protein (Fig. 7C). Given that the roles of lysines at positions 29, 48, and 63 of ubiquitin have been extensively analyzed, we next generated ubiquitin mutants in which two or all three of these residues were replaced with arginine. UFD2a mediated polyu-biquitination even with these double and triple mutants of ubiquitin (data not shown). The lysine residues at positions 6 and 11 of ubiquitin have been implicated previously (38) in polyubiquitination. We therefore generated mutant ubiquitins in which four or five lysine residues, including lysine 6 or lysine 11, were replaced with arginine. Again, a substantial extent of polyubiquitination by UFD2a remained apparent with these ubiquitin mutants, including the K6R/K11R/K29R/K48R/K63R mutant (data not shown). To exclude the possibility that several ubiquitin monomers are attached at more than one site of UFD2a (referred to here as multiubiquitination), we prepared a recombinant ubiquitin (K0) in which all lysines are substituted with arginines (Fig. 7D), and we included this mutant in the ubiquitination assay. Only the monoubiquitinated form of UFD2a was generated with the K0 mutant of ubiquitin (Fig.  7E), suggesting that the ladder pattern of ubiquitination observed with wild-type ubiquitin is indeed due to polyubiquitination (not multiubiquitination) of UFD2a.
To determine which lysine residues of ubiquitin support polyubiquitination by UFD2a, we prepared a series (K1) of ubiquitin mutants in which all but one lysine residue are replaced with arginine (Fig. 7D). The extent of UFD2a-mediated polyubiquitination with the K1 ubiquitin mutant with an intact lysine 33 (K1(33)) was similar to that apparent with wild-type ubiquitin (Fig. 7E). In addition, a reduced extent of polyubiquitination was also observed with the mutant K1 (27). These data do not necessarily rule out the possibility that other lysines at positions 6,11,29,48, and 63 support polyubiquitination because of possible structural changes induced by the replacement of multiple lysine residues with arginine. We finally examined whether the ubiquitin mutant K27R/K33R is functional with UFD2a (Fig. 7F). This mutant supported polyubiquitination of UFD2a to almost the same extent as that observed with K1(27) (Fig. 7G), suggesting that lysines at positions other than 27 and 33 can be used by UFD2a. Collectively, these experiments with various ubiquitin mutants indicate that UFD2a is able to utilize multiple lysine residues of ubiquitin in addition to lysine 48 to mediate the formation of polyubiquitin chains. DISCUSSION Our results indicate that the U box domain, in the appropriate molecular context, facilitates E2-dependent ubiquitination. The U box proteins thus define a third family of E3 enzymes, in addition to the HECT and RING finger families of E3s. This observation is somewhat unexpected, given that yeast Ufd2, the prototype U box protein, was originally described as a factor (designated E4) that promotes polyubiquitination of artificial ubiquitin fusion proteins in conjunction with E1, E2, and E3 (35,39). In this scenario, the E3 (Ufd4, a HECT-type E3) catalyzes oligo-ubiquitination, but not polyubiquitination; E4 is required for elongation of the oligo-ubiquitin chain and consequent recognition of the substrate by the 26 S proteasome. We have now shown, however, that six unrelated mammalian U box proteins, including the yeast Ufd2 homologs UFD2a and UFD2b, mediate polyubiquitination in the presence of E1 and E2 in a manner that is independent of other E3s. Thus, we propose that E4 activity is a specialized type of E3 activity that recognizes oligo-ubiquitinated artificial fusion proteins as substrates. Whether the mammalian U box proteins also possess E4 activity and whether E4 activity is physiologically relevant remain to be determined.
With regard to the mechanism by which U box proteins function together with E2s to catalyze ubiquitination, it is possible that, analogous to HECT family E3s, U box proteins both bind to E2 and form thiol ester intermediates with ubiquitin. However, no cysteine residue within the U box is conserved among members of this family, making this possibility unlikely. Indeed, some U box proteins, such as UFD2a, do not contain any cysteine residues within the U box domain. Alternatively, the U box may constitute a site of interaction with E2 that allows the direct transfer of ubiquitin from E2 to target lysines. This notion is consistent with the observations that the amino acid sequences of the U box domains of UFD2a, UFD2b, CHIP, and KIAA0860 (all of which prefer Ubc4 or UbcH5C as an E2) are more similar to each other than they are to those of the U box domains of CYC4 and PRP19 (neither of which functions with Ubc4 or UbcH5C) and that the U box is structurally similar to the RING finger (which recruits E2 in RING finger family E3s). However, we failed to detect a physical interaction of U box proteins with the relevant E2 enzymes either by yeast two-hybrid analysis or with biochemical assays (data not shown), suggesting that if these proteins do interact they do so with low affinity. In general, physical E2-E3 interactions are not necessarily detected in other E2-E3 systems. The formation of a polyubiquitin chain by linkage of the COOH terminus of one ubiquitin moiety to lysine 48 of the adjacent ubiquitin had been thought to mark a protein for proteolysis by the 26 S proteasome. However, recent observations (40) indicate that polyubiquitin chains can also be assembled through conjugation to lysine residues of ubiquitin other than lysine 48, and the resulting chains appear to function in distinct biological processes. A short lysine 29-linked chain might thus constitute a signal to recruit E4 before proteasomal proteolysis (35). Polyubiquitination through lysine 63 of ubiquitin appears to mark proteins that contribute to DNA repair (41), to the cellular response to stress (42), to inheritance of mitochondrial DNA (43), to endocytosis of certain plasma membrane proteins (44), or to ribosomal function (45). In addition, TRAF6, a RING finger-type E3, in conjunction with the E2 Ubc13 and and the Ubc-like protein Uev1A, targets lysine 63 of ubiquitin and plays an important role in IB phosphorylation in the NF-B signaling pathway (46). Our data now show that some U box E3s catalyze polyubiquitination by targeting lysine residues other than lysine 48, whereas HECT and RING finger E3s (with the exception of TRAF6) usually target lysine 48 for polyubiquitination. Moreover, UFD2a exhibits broad specificity for ubiquitin lysine residues in polyubiquitination. Hence, it is possible that proteins that have undergone polyubiquitination by UFD2a display heterogeneous or multiply branched struc-tures that might possess distinct biological functions. However, it is unclear whether it is possible for two ubiquitin moieties to be simultaneously conjugated to different lysine residues of a single ubiquitin molecule or whether conjugation of the first ubiquitin may physically interfere with the conjugation of the second ubiquitin on a different lysine. The formation of multiply branched ubiquitin chains by UFD2a might be expected to give rise to irregularly spaced ubiquitinated species on SDS-PAGE; however, we did not observe such electrophoretic patterns. The structure and function of the unusual polyubiquitin chains formed by UFD2a thus require further characterization.
In yeast, Ufd2 is implicated in cell survival under stress conditions, a role that is consistent with the association of Ufd2 with the AAA-type ATPase Cdc48, which possesses chaperone activity. We have also shown 2 that UFD2a, a mammalian homolog of yeast Ufd2, interacts with VCP, which appears to be the mammalian ortholog of yeast Cdc48, suggesting that the association of AAA-type ATPases with U box proteins has been conserved through evolution and thus may be functionally important. Furthermore, the TPR-containing U box protein CHIP binds to the molecular chaperones HSP70 and HSP90 (47), and CYC4 contains a cyclophilin-like peptidylprolyl isomerase do- FIG. 7. Ubiquitin lysine residue specificity of polyubiquitination by mammalian U box proteins. A, the ubiquitination activities of Nedd4, Mdm2, and mammalian U box proteins were assayed in vitro in the absence (mock) or presence of wild-type (WT) ubiquitin (Ub) or its K48R mutant. Bacterial lysate containing UbcH5C was used as the E2 for Nedd4, Mdm2, UFD2a, UFD2b, CHIP, and KIAA0860, whereas bacterial lysate containing Ubc3 was used for CYC4 and PRP19. The reaction mixtures were subjected to immunoblot analysis with antibodies to ubiquitin. B, schematic representation of ubiquitin mutants in which the lysine residues at positions 6, 11, 27, 29, 33, 48, or 63 were individually substituted with arginine. The substituted arginines are indicated in bold. C, in vitro ubiquitination assays were performed with 35 S-labeled UFD2a and bacterial lysate containing E2 (UbcH5C) in the absence (mock) or presence of wild-type ubiquitin or the ubiquitin mutants shown in B. The reaction mixtures were subjected to SDS-PAGE and autoradiography. D, schematic representations both of a series (K1) of ubiquitin mutants in which all but the indicated single lysine residues (shown in bold) are replaced with arginine and of the K0 mutant of ubiquitin, in which all lysines are substituted with arginine. E, 35 S-labeled UFD2a was subjected to the ubiquitination assay in the presence of bacterial lysate containing E2 (UbcH5C) and in the absence (mock) or presence of wild-type ubiquitin or the ubiquitin mutants shown in D. F, schematic representation of the K27R/K33R mutant, in which lysines 27 and 33 are substituted with arginine. G, 35 S-labeled UFD2a was subjected to the ubiquitination assay in the presence of bacterial lysate containing E2 (UbcH5C) and in the absence (mock) or presence of wild-type ubiquitin or the ubiquitin mutants shown in F. main (48). These observations suggest that U box proteins may serve as E3s that physically and functionally associate with chaperones. The U box-type E3s may therefore play an important role in the cellular response to stress or to misfolded or otherwise damaged proteins. The combination of CHIP with HSP90 induces ubiquitination of the glucocorticoid receptor (49), and CHIP together with HSP70 targets immature cystic fibrosis transmembrane conductance regulator for proteasomal degradation (50). The folding of both the glucocorticoid receptor and cystic fibrosis transmembrane conductance regulator is controlled by molecular chaperones. In these two instances, CHIP was proposed to act as a co-chaperone that regulates the balance between protein folding and degradation by ubiquitination. However, our present results demonstrate that CHIP itself possesses ubiquitin ligase activity. Other U box proteins likely play important roles in the ubiquitination of specific substrates, especially under conditions of cellular stress, when they may serve to link the processes of protein folding by molecular chaperones and protein degradation by the ubiquitin-proteasome pathway.