Members of the E2D (UbcH5) Family Mediate the Ubiquitination of the Conserved Cysteine of Pex5p, the Peroxisomal Import Receptor

According to current models of peroxisomal biogenesis, newly synthesized peroxisomal matrix proteins are transported into the organelle by Pex5p. Pex5p recognizes these proteins in the cytosol, mediates their membrane translocation, and is exported back into the cytosol in an ATP-dependent manner. We have previously shown that export of Pex5p is preceded by (and requires) monoubiquitination of a conserved cysteine residue present at its N terminus. In yeasts, and probably also in plants, ubiquitination of Pex5p is mediated by a specialized ubiquitin-conjugating enzyme, Pex4p. In mammals, the identity of this enzyme has remained unknown for many years. Here, we provide evidence suggesting that E2D1/2/3 (UbcH5a/b/c) are the mammalian functional counterparts of yeast/plant Pex4p. The mechanistic implications of these findings are discussed.

. In mammals and plants but not in yeasts/fungi, Pex5p is also in charge of targeting PTS2-containing proteins to the organelle matrix (4). In this case, however, the Pex5p-PTS2 interaction is not direct but rather bridged by Pex7p, the socalled PTS2 receptor (5).
According to current models, after binding cargo proteins in the cytosol, Pex5p interacts with the peroxisomal docking/ translocation machinery, a multisubunit protein complex comprising mostly transmembrane proteins (1,2). The core subunits of this machinery are Pex13p, Pex14p, and the RING peroxins Pex2p, Pex10p, and Pex12p (6 -8). The interaction of the Pex5p⅐cargo protein complex with this machinery ultimately results in the partial insertion of Pex5p into the peroxisomal membrane with the concomitant translocation of cargo proteins into the matrix of the organelle (9,10). Interestingly, this step is ATP-independent, suggesting that the driving force for protein translocation derives from the strong protein-protein interactions that Pex5p establishes with some components of the docking/translocation apparatus (11)(12)(13).
Pex5p is not consumed in this pathway. It is instead exported in an ATP-dependent way back into the cytosol to promote further rounds of protein transportation (11)(12)(13). For some time, it was assumed that this energy-requiring process comprised only one step: the dislocation of Pex5p from the docking/ translocation complex by the mechanoenzymes Pex1p and Pex6p, two members of the ATPases associated with various cellular activities (AAA) family of proteins (11,13). However, recent data uncovered another ATP-dependent step, this one involving the ubiquitin-conjugating cascade (UCC). Indeed, data suggesting that both yeast and mammalian Pex5p have to be ubiquitinated to be recognized by the export machinery were provided (14,15). Interestingly, ubiquitin is conjugated to Pex5p not through the usual isopeptide bond but rather via a thiol ester bond involving a conserved cysteine present at the N terminus of the peroxin (14,16). Although thiol ester bonds are at the heart of the chemistry used by the enzymes of UCC, they are, at least apparently, quite uncommon in ubiquitin-substrate conjugates (17). In fact, besides Pex5p, only one other protein, a major histocompatibility complex class I molecule, was reported to be ubiquitinated in this way (18).
The UCC acting on yeast Pex5p during its normal and transient passage through the peroxisomal membrane is now fairly * This work was supported by grants from Fundação para a Ciê ncia e Tecnologia (PTDC program) and Fundo Europeu de Desenvolvimento Regional, Portugal, by the European Union VI Framework program Grant LSHG-CT-2004-512018, Peroxisomes in Health and Disease, and by a grant from the Deutsche Forschungsgemeinschaft within the SFB 642 and the Land North Rhine-Westphalia, Germany. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. □ S The on-line version of this article (available at http://www.jbc.org) contains a supplemental table and a supplemental figure. 1 Both authors contributed equally to this work. 2 Supported by Fundação para a Ciê ncia e Tecnologia. 3 To whom correspondence should be addressed. characterized (13, 15, 16, 19 -22). After the ATP-dependent activation of ubiquitin catalyzed by UBA1, the only ubiquitinactivating enzyme (E1) present in yeast (23), ubiquitin is transferred to Pex4p (UBC10) (24,25). This ubiquitin-conjugating enzyme (E2) is a peripheral component of the peroxisomal membrane in yeasts and probably also in plants. In the absence of Pex22p, a peroxisomal intrinsic membrane protein, Pex4p is unstable, suggesting that the two proteins form a structural/functional unit (26 -29). The identity of the ubiquitin ligase(s) (E3) acting at the terminal step of this cascade is still unknown. It seems likely, however, that this role is played by the RING peroxin(s) of the docking/translocation machinery, as supported by the observation that yeast Pex4p interacts with Pex10p (30). Much less is known regarding the UCC acting on mammalian Pex5p. Actually, the only known facts about this issue are that Pex5p is ubiquitinated at the peroxisomal membrane after the ATP-independent membrane insertion step, that this ubiquitination is a prerequisite for its export, and that ubiquitinated Pex5p is exported into the cytosol in an ATP␥S-sensitive process (14). Several factors may explain this lack of knowledge. Perhaps the most important is related to the fact that the main approach that has been used to identify mammalian peroxins (i.e. homology searches using yeasts/fungi sequences as queries) has failed in this case. Indeed, no in silico evidence for a mammalian PEX22 gene could be found in several recent studies (31,32). For Pex4p, a different problem applies. Mammals have more than 30 ubiquitin-conjugating enzymes (reviewed in Ref. 33). However, none of these proteins seem to display a particularly close sequence relationship to the yeast/fungi/ plants Pex4p (27). Proteomic studies aiming at defining the protein repertoire of mammalian peroxisomes have been of no help in this subject either (34 -36). No evidence for a peroxisome-associated E2 was found, and none of the novel peroxisomal proteins identified in those studies displays even a remote relationship to the yeast/plant Pex22p. Clearly, other approaches are needed to address this problem.
During the last years, we have been using an in vitro system to characterize the Pex5p-mediated import pathway in mammals. Two different versions of this in vitro system have been employed. In one version, a postnuclear supernatant (PNS) from rat liver is used as the source of peroxisomes and other required components. In this version, all the steps of the Pex5pmediated import pathway can be monitored. These include 1) the cargo-dependent but ATP-independent insertion of Pex5p into the peroxisomal docking/translocation machinery (12,37), 2) the stage 2/stage 3 ATP␥S-insensitive transition, which we now know reflects ubiquitination of the conserved cysteine at the N terminus of Pex5p (14,38,39), and 3) the subsequent ATP-dependent (and ATP␥S-sensitive) export step of ubiquitinated Pex5p back into the cytosol (12,39,40). In the other version, highly purified peroxisomes from rat liver are used in the in vitro reactions. Here, cargo-dependent but ATP-independent insertion of Pex5p into the docking/translocation machinery is still possible. However, because the system lacks the required amounts of ubiquitin and/or components of the UCC, transition of stage 2 Pex5p into stage 3 Pex5p is no longer possible, and the peroxin becomes arrested at the docking/ translocation machinery (12). In this work, we have explored the properties of this incomplete version to identify the rat liver ubiquitin-conjugating enzyme(s) mediating the peroxisomal ubiquitination of the conserved cysteine residue of Pex5p.
Cytosol Preparation, Fractionation, and Purification of E2s-Livers from male rats fasted overnight were homogenized in 5 volumes of ice-cold 0.25 M sucrose, 20 mM Tris-HCl, pH 7.5, 1 mM EDTA-NaOH, pH 7.5, supplemented with 2 g/ml E-64. A cytosolic fraction was obtained from this homogenate after the following sequential centrifugations: 10 min at 600 ϫ g, 20 min at 12,300 ϫ g, and 1 h at 100,000 ϫ g. Cytosolic Fraction I and Fraction II were obtained according to Ciechanover et al. (42). Covalent affinity chromatography was performed according to Ref. 42 with modifications. One hundred and fifty mg of cytosolic proteins (at 30 mg/ml) were subjected to precipitation with 16% (w/v) polyethylene glycol 4000 (added from a 50% (w/v) stock solution in 20 mM Tris-HCl, pH 8.0, 150 mM NaCl) for 20 min on ice. The pellet obtained after centrifugation at 10,000 ϫ g for 20 min was resuspended in 5 ml of 50 mM Tris-HCl, pH 7.5, 5 mM ATP, 10 mM MgCl 2 , and 0.2 mM DTT. The clarified sample was applied at a 0.1 ml/min flow rate to a 0.8-ml ubiquitin-agarose column (10 mg/ml; BIOMOL), previously equilibrated with 5 volumes of buffer A (50 mM Tris-HCl, pH 7.5, 2 mM ATP, 5 mM MgCl 2 , 0.2 mM DTT). The column was washed at 0.5 ml/min as follows: 3 volumes of buffer A, 6 volumes of 50 mM Tris-HCl, pH 7.5, and 3 volumes of 50 mM Tris-HCl, pH 7.5, 1 M KCl. Elution was performed at 0.1 ml/min flow rate with 3 volumes of a high pH/DTT buffer (50 mM Tris-HCl, pH 9.0, 20 mM DTT). The eluate was immediately neutralized with a 100 mM Tris-HCl, pH 7.5, 33 g/ml bovine serum albumin solution. The wash and elution fractions were subjected to three cycles of concentration/dilution with 20 mM Tris-HCl, pH 7.5, supplemented with 0.2 mM DTT, using Vivaspin 500 columns (3,000 molecular weight cut-off polyethersulfone; Vivascience), pretreated for 30 min at room temperature with 600 l of 0.1% (w/v) bovine serum albumin to minimize protein loss by adsorption. In the subsequent steps, only Eppendorf protein LoBind microcentrifuge tubes were used. Concentrated eluates (100 -150 l) were incubated with 30 l (bed volume) of DEAE-cellulose (DE52, Whatman) for 30 min at 18°C. The beads were washed two times for 10 s with 100 l of 20 mM Tris-HCl, pH 7.5, and the bound proteins were eluted twice with 50 l of 20 mM Tris-HCl, pH 7.5, 0.5 M KCl, for 5 min. Throughout the entire purification procedure, aliquots of all fractions were withdrawn for SDS-PAGE analysis and in vitro import experiments to test for E2 activity. With the exception of the cytosolic fraction, which was kept frozen at Ϫ70°C for 5-10 days, the entire procedure, activity assays included, was always performed in a single day.
In Vitro Import Experiments-Import reactions using 400 g of PNS from rat liver or human skin fibroblasts were performed as described before (39). In vitro import experiments containing 50 g of purified peroxisomes were performed in import buffer comprising 0.25 M sucrose, 50 mM KCl, 5 mM MOPS-KOH, pH 7.2, 3 mM MgCl 2 , 20 M methionine, 10 M ZnCl 2, and 2 g/ml E-64. ATP or ATP␥S was used at 5 mM final concentration. Where indicated, 25 g of GST⅐ubiquitin (GST⅐Ub (45)) or 4 g of bovine ubiquitin (Sigma) were used per import reaction. TPRs-Pex5p (46), a PTS1-containing peptide (CRYHLKPLQSKL), and a control peptide (CRYHLKPLQLKS) were used exactly as described before (37). Where specified, the reactions were supplemented with 250 ng of recombinant human E1 and 10 ng of the relevant E2. The reactions were incubated for 20 min at 37°C, and the organelles were isolated by centrifugation and analyzed by SDS-PAGE and autoradiography.
Protein Identification by Mass Spectrometry-Following staining with colloidal Coomassie Brilliant Blue G-250, gel bands of interest were cut out and destained by alternately incubating them with 20 l of 10 mM NH 4 HCO 3 and 20 l of 5 mM NH 4 HCO 3 /50% acetonitrile (v/v) for 10 min each. In-gel digestion was performed overnight at 37°C using trypsin dissolved in 10 mM NH 4 HCO 3 buffer (pH 7.8). The resulting peptides were extracted twice with 10 l of acetonitrile/5% formic acid (50:50, (v/v)). Subsequently, acetonitrile was removed in vacuo, and samples were acidified by the addition of 5% formic acid to a final volume of 20 l. Online reversed-phase nano-HPLC separations were performed using the Dionex LC Packings HPLC systems (Dionex LC Packings, Idstein, Germany) as described (47), with slight modifications. Peptides were separated on a C18 nano column (C18 PepMap; Dionex) using solvent A (0.1% formic acid) and solvent B (0.1% formic acid in 84% acetonitrile, (v/v)) to perform a binary linear gradient of 5-30% solvent B in 34 min. Electrospray ionization tandem mass spectrometry was performed on a Bruker Daltonics HCT ultra ion trap instrument (Bremen, Germany) equipped with a nanoelectrospray ion source (Bruker Daltonics) as described (36). Peak lists of tandem mass spectrometry spectra acquired were generated using DataAnalysis 3.4.179 with default parameters. For peptide identifications via MASCOT (release version 2.2) (48) and SEQUEST (Version TurboSEQUEST v. 27) (49), peak lists were correlated with a composite data base consisting of the Rat International Protein Index (Rat IPI V3.30) data base (at the European Bioinformatics Institute) containing 41,269 protein entries and a duplicate of the same data base, in which the amino acid sequences were randomly shuffled. Data base searches were performed with tryptic specificity, allowing two missed cleavages and oxidation of methionine. Proteins were assembled on the basis of peptide identifications using the Pro-teinExtractor tool (version 1.0) in ProteinScape (version 1.3 S.R. 2, Bruker Daltonics) as described (36). The presence of each protein isoform was confirmed by the identification of at least one unique peptide. Only protein identifications up to an accumulated false discovery rate of 0% are reported.
Cell Culture-Human primary skin fibroblast were grown in a standard Dulbecco's modified Eagle's medium supplemented with 20% fetal bovine serum. The cells were grown at 37°C in the presence of 5% CO 2 . Cells at 90% confluence were harvested with a cell scraper, washed three times with phosphate-buffered saline, and centrifuged at 2,000 ϫ g for 10 min at 4°C. The resulting pellet was resuspended in SEM buffer supplemented with E-64. The sample was homogenized using a tight-fit glass-Teflon homogenizer, and the homogenate was centrifuged at 600 ϫ g for 10 min at 4°C. Aliquots of the PNS fraction were frozen in liquid N 2 and stored at Ϫ70°C.  35 S-labeled ⌬C1-Pex5p was incubated in import buffer containing cytosol, peroxisomes, GST⅐Ub and ATP. The organelles were isolated by centrifugation, solubilized with detergent, and subjected to a pull-down assay using glutathione-Sepharose beads. The total (T), unbound (Unb), and bound fractions (in this case, treated (Bϩ) or not (BϪ) with DTT) were subjected to SDS-PAGE. Three equivalents of the bound fraction in relation to the total and unbound samples were loaded onto the gel.

E2D-mediated Ubiquitination of Mammalian Pex5p
Miscellaneous-The recombinant human E1 and E2s used in the in vitro import experiments were purchased from BIOMOL and were diluted in SEM buffer with 0.1% bovine serum albumin immediately before use. The anti-PMP70 antibody (50) was kindly provided by Dr. Wilhelm W. Just (University of Heidelberg, Heidelberg, Germany). The antibody directed to human Pex5p was described before (51). NEM and DTT treatment of protein samples, GST pull-down assays, SDS-PAGE, and Western blotting were performed as described recently (14).

RESULTS
To identify the mammalian counterpart of yeast Pex4p, we first asked whether or not we could detect its activity in purified rat liver peroxisomes by subjecting these organelles to in vitro import reactions supplemented with GST⅐Ub and recombinant E1. 35 S-labeled ⌬C1-Pex5p synthesized in vitro in an E. colibased transcription/translation system was used in most of these experiments as the substrate for ubiquitination. As shown before, this truncated version of Pex5p (which comprises amino acid residues 1-324 of human Pex5p) is completely competent in all the steps occurring at the peroxisomal protein translocation machinery (i.e. insertion, ubiquitination, and export) but, unlike the full-length peroxin, contains only one cysteine, the conserved Cys-11 (14,37,40). Thus, the existence of thiol-sensitive ubiquitinated ⌬C1-Pex5p in these assays reflects ubiq-uitination at this cysteine (thiol ester bonds, unlike isopeptidic bonds, are thiol-sensitive; (42)). It should also be noted that GST⅐Ub (and not ubiquitin) was used in these experiments because it is used efficiently by the UCC acting on Cys-11 of Pex5p or ⌬C1-Pex5p when the complete version of the in vitro import system is employed. However, in contrast to ubiquitin, conjugation of GST⅐Ub to Pex5p or ⌬C1-Pex5p gives rise to peroxisomal species that are not substrates for the peroxisomal export machinery (14). This choice simplifies the ubiquitination assays because only organelle pellets need to be analyzed.
As shown in Fig. 1A, purified rat liver peroxisomes are not competent in promoting ubiquitination of ⌬C1-Pex5p when GST⅐Ub and recombinant E1 are included in the import buffer (lane Per ϩ E1). A similar result is obtained when ⌬C1-Pex5p is incubated with a cytosolic fraction lacking organelles (lane Cyt). In contrast, thiol-sensitive GST⅐Ub⅐⌬C1-Pex5p (Fig. 1A, arrow a; see also Fig. 1B) is easily observed when 35 S-labeled ⌬C1-Pex5p is subjected to import reactions containing both purified peroxisomes and the cytosolic fraction (lane Per ϩ Cyt). Apparently, the cytosolic fraction contains at least one component (besides the E1 enzyme) required for the UCC acting on Cys-11 of Pex5p, a conclusion further corroborated by experiments performed with PNS fractions from rat liver and human skin fibroblasts (see below). Many different hypotheses could be forwarded to explain these observations. A plausible one is to assume that the required cytosolic component is an E2 enzyme. This possibility was addressed in the experiments described below.
To purify E2 enzymes from rat liver cytosol, we tried to use the two-step procedure developed by Ciechanover et al. (42). In the first step, cytosolic proteins are separated into two fractions by anion-exchange chromatography using DE52. The so-called Fraction I contains proteins (ubiquitin included) that do not bind to DE52. Fraction II is obtained by eluting the column with a high salt buffer and contains E1, several E2s, and many other proteins. In the second step, the ubiquitin-depleted Fraction II is subjected to covalent affinity chromatography using an ubiquitin-derivatized matrix. Unexpectedly, no ubiquitinating activity could be detected when using the DE52 Fraction II in our assays. Fraction I alone also produced a negative result. However, the addition of recombinant E1 to this fraction (but not to Fraction II) resulted in a full recovery of the ubiquitinating activity (see supplemental Fig. S1), an observation that sup- ported the idea that the cytosolic fraction contains the required E2.
The absence of E1 in Fraction I together with the more complicated problem that this fraction contains high amounts of ubiquitin precludes its utilization in the covalent affinity chromatography step. We found that by subjecting a cytosolic fraction to polyethylene glycol precipitation, we could recover the majority of the ubiquitinating activity in the pellet, whereas most of rat liver ubiquitin remained in the supernatant ( Fig. 2A  and data not shown). This ubiquitin-depleted fraction was then subjected to the covalent affinity chromatography step. As shown in Fig. 2A, virtually all the ubiquitinating activity present in the polyethylene glycol-precipitated fraction was recovered in the high pH/DTT eluate of the ubiquitin-agarose column (lane E, lower panel). As expected, the complexity of this protein fraction was still relatively high ( Fig. 2A, upper panel, lane  E). To further purify the component of interest, we explored the fact that it does not bind to DE52 (see supplemental Fig. S1). As shown in Fig. 2B, lower panel, almost no ubiquitinating activity could be detected when using the DE52-bound protein fraction in our assay. In contrast, a significant amount of GST⅐Ub⅐⌬C1-Pex5p was generated in the presence of the unbound protein fraction. SDS-PAGE analysis of this protein fraction revealed the presence of two protein bands (Fig. 2B, upper panel, lane Unb). Mass spectrometry analysis identified E2L3 (UbcH7) as the main protein present in the slower migrating band 1. This same E2 (or a fragment of it) together with E2N (UbcH13) and the three closely related E2 enzymes E2D3 (UbcH5c), E2D2 (UbcH5b), and E2D1 (UbcH5a) were identified in the faster migrating protein band 2 (see supplemental Table I).
To determine which of the five E2s identified (if any) is capable of mediating ubiquitination of Pex5p, a set of 13 human recombinant E2s comprising the homologues of the 5 rat E2s described above was tested in our ubiquitination assay. Particular care was given to the concentration of these enzymes in the assay. In the experiment presented in Fig.  2B, the amount of the DE52-unbound protein fraction used in the ubiquitination reaction (lower panel) was 2% of the amount loaded in the gel shown in the upper panel.
Assuming that the strongest protein band visible in the Coomassie Bluestained gel (containing E2L3/ UbcH7) contains ϳ0.5 g of protein, then 10 ng of the correct E2 per import reaction should be sufficient to observe an ubiquitinating activity. As shown in Fig. 3A, from all the E2s tested at this concentration, only UbcH5a (E2D1), UbcH5b (E2D2), and UbcH5c (E2D3) are capable of ubiquitinating ⌬C1-Pex5p. The ubiquitinated Pex5p species are thiol-sensitive (Fig. 3B), and their production requires catalytically active versions of these E2s ( Fig.  3A; lower panel, last three lanes). Finally, no ubiquitination of Pex5p could be detected in these assays in the absence of peroxisomes (data not shown).
The results presented above indicate that any of the three UbcH5 family members are capable of ubiquitinating ⌬C1-Pex5p at the conserved cysteine residue. However, it remained unclear whether or not this event is linked to the normal transit of Pex5p through the peroxisomal protein translocation machinery. To clarify this issue, we explored the fact that insertion of full-length Pex5p into the docking/translocation machinery is cargo protein-dependent (37). As explained in detail elsewhere (12), the soluble phase of these import reactions contains high amounts of peroxisomal proteins that have leaked from the organelles during their preparation and manipulation. Under standard in vitro import conditions, these proteins are sufficient to allow efficient insertion of the peroxin into the peroxisomal translocation machinery. If, however, these proteins are sequestered by adding to the import reactions a vast excess of a recombinant protein comprising only the cargo protein-binding domain of Pex5p (TPRs-Pex5p), then insertion of Pex5p is no longer possible (46). As shown in Fig.  4A, almost no ubiquitination of Pex5p was observed when cargo proteins were sequestered from the import medium (compare lanes 2 and 3). As expected, the addition of a syn-thetic PTS1-containing peptide reverted the inhibitory action of TPRs-Pex5p (lane 4). Thus, the UbcH5-mediated ubiquitination of Pex5p occurs only after the cargo-dependent membrane insertion step, suggesting that only Pex5p molecules engaged in the protein transport process are substrates for this ubiquitination event. One additional observation supporting this conclusion derives from the fact that UbcH5-mediated conjugation of (authentic) ubiquitin to Pex5p or ⌬C1-Pex5p results in export-competent species. Indeed, as shown in Fig.  4B, significant amounts of ubiquitinated Pex5p or ⌬C1-Pex5p (arrows b) were found in the soluble phase of import reactions performed in the presence of ATP. In contrast, GST⅐Ub conjugates (arrows a) remained in the organelle fraction in agreement with our previous observation that these conjugates are not substrates for the peroxisomal export machinery (14).
The capacity of UbcH5 enzymes in mediating the ubiquitination of peroxisomal Pex5p giving rise to export-competent species is also evident in experiments in which the behavior of the endogenous rat liver peroxisomal Pex5p was monitored. As shown in Fig. 4C, about half of the total peroxisome-associated rat Pex5p was already in its ubiquitinated form after just 2.5 min of incubation in import buffer containing E1, ubiquitin, ATP, and UbcH5c (lane 2P). After 20 min of incubation under these conditions, the vast majority of rat Pex5p was ubiquitinated and, importantly, a significant fraction was now found in the soluble phase of the reaction (lanes 3P and 3S). In agreement with previous results (14), the addition of ATP␥S (instead of ATP) to these reactions did not affect ubiquitination of rat Pex5p but completely blocked its export (lanes 4P and 4S).
To assess whether our findings may also be extended to humans, total organelles isolated from a PNS from human skin fibroblasts were incubated with 35 S-labeled ⌬C1-Pex5p in import medium supplemented with GST⅐Ub, E1, and UbcH5c or UbcH7. As shown in Fig. 5, thiol-sensitive GST⅐Ub⅐⌬C1-Pex5p is readily obtained when UbcH5c (or UbcH5a or UbcH5b; data not shown) is included in the import reactions. In contrast, inclusion of UbcH7 in the import medium does not promote ubiquitination of ⌬C1-Pex5p. As expected, exactly the same results were obtained with a rat liver PNS.

DISCUSSION
The aim of this work was to identify the mammalian ubiquitin-conjugating enzyme(s) mediating the ubiquitination of the conserved cysteine of Pex5p at the peroxisomal membrane. For this purpose, we first developed a peroxisome-dependent Pex5p-ubiquitination assay, which was then used to monitor the distribution of the relevant E2 during the fractionation procedure of rat liver cytosol. Using the method described by Ciechanover et al. (42) (with some modifications), it was possible to obtain a low complexity protein fraction in which five E2s were identified. Subsequent ubiquitination assays using recombinant human versions of these E2s revealed that three of them, all members of the E2D family, were indeed capable of mediating ubiquitination of Pex5p. Importantly, the E2D-mediated Pex5p ubiquitination was shown to occur at the right residue of the receptor (i.e. Cys-11) and at the right moment/location (i.e. after the docking/translocation step). Finally, E2D-mediated ubiquitination of Pex5p resulted in export-competent species.  35 S-labeled Pex5p (0.2 l of a rabbit reticulocyte lysate per reaction) was mixed with 50 g of purified peroxisomes in import buffer supplemented with ATP, recombinant E1, and GST⅐Ub. The sample in lane 1 received no further components, whereas those in lanes 2-4 received UbcH5c. In addition, TPRs-Pex5p plus a control peptide or a PTS1-containing peptide were added to samples in lanes 3 and 4, respectively. After incubation at 37°C for 20 min, the samples received NEM, and the organelles were sedimented and processed for SDS-PAGE under non-reducing conditions. B, 35 S-labeled ⌬C1-Pex5p (upper panels) or 35 S-labeled Pex5p (0.2 l of a rabbit reticulocyte lysate per reaction; lower panels) were incubated with purified peroxisomes in import buffer containing E1, ATP, and UbcH5c in the presence of Ub or GST⅐Ub. The reactions were supplemented with NEM and separated into an organelle fraction (P) and a soluble phase (S) by centrifugation. The samples were then analyzed under non-reducing conditions by SDS-PAGE and autoradiography and probed with an antibody against PMP70 to assess the separation into an organelle pellet and supernatant. Exactly the same results were obtained when UbcH5a/b isoforms were used in these experiments (data not shown). a, GST-Ub-⌬C1-Pex5p/Pex5p species; b, Ub-⌬C1-Pex5p/Pex5p species; lane I, 10% of the 35 S-labeled proteins used in each lane. C, purified rat liver peroxisomes were incubated in import buffer supplemented with E1 and ubiquitin in the presence (lanes 2-4) or absence of UbcH5c (lanes 1). ATP was used in reactions in lanes 1-3, whereas the reaction in lanes 4 received ATP␥S. Incubation at 37°C was for 2.5 min (lanes 2) or 20 min (lanes 1, 3, and 4). After the addition of NEM, the reactions were subjected to centrifugation to separate organelles (lanes P) from soluble proteins (lanes S). Samples were then subjected to SDS-PAGE under non-reducing conditions and analyzed by Western blotting using an antibody directed to human Pex5p. Rat Pex5p and its ubiquitinated form (arrow b) are indicated. The asterisk indicates a nonspecific protein band occasionally detected by the antibody. I, untreated rat liver peroxisomes (50 g of protein). Numbers to the left indicate the molecular masses of protein standards in kDa.
Taken together, these findings strongly suggest that we have identified the rat liver E2s that participate in the UCC required for the Pex5p-mediated import pathway. Considering that members of the E2D family are ubiquitously expressed in human tissues (52,53) and the results obtained with human organelles (Fig. 5), it is likely that this conclusion is also valid for humans.
The data presented here provide explanations for many of the observations reported before and reveal interesting properties of the mammalian Pex5p-mediated import pathway. For instance, it is now easier to understand why a peroxisome-associated E2 was never found in the several studies aiming at defining the mammalian peroxisomal protein repertoire. As shown in Fig. 5, the simple procedure of subjecting a PNS to a short centrifugation results in the complete depletion of the Pex4p-like activity from the sedimented organelles. Thus, besides lacking a specialized E2, as yeast/plant Pex4p may be regarded, mammalian peroxisomes also appear to lack their own pool of the E2 component required for ubiquitination of Pex5p. Furthermore, this observation also provides a biochemical ground to the in silico data, suggesting that mammals, as well as many other organisms, lack functional counterparts of yeast/plant Pex22p and is in agreement with the fact that E2-E3 interactions are generally very weak and transient with large dissociation constants (54,55).
Finally, our data may also have implications to the field of human peroxisomal biogenesis disorders. Indeed, considering that any of the three enzymes identified here, each encoded by a different gene, is capable of supporting the UCC acting on Pex5p and the fact that several important biological roles have been ascribed to these E2s (e.g. in controlling the cellular levels of p53 or IB␣ (56, 57)), it now seems unlikely that defects in an E2 enzyme in humans will ever be linked to an isolated defect in peroxisomal biogenesis, as observed in yeasts and plants.
The finding that members of the E2D family participate in a novel, previously uncharacterized, ubiquitin-conjugating cascade is not surprising. Indeed, mammalian E2D enzymes are known to be highly active with a variety of E3s (58) and, at least in rat liver, they probably mediate a large fraction of the total ubiquitinating activity (59). However, by showing that ubiquitination of a cysteine residue in a particular protein does not require a specialized E2, our results do raise the possibility that this unusual type of ubiquitination is more common than presently realized.