Peroxisomal matrix protein import is facilitated by cycling receptors that recognize their cargo proteins in the cytosol by a peroxisomal targeting sequence (PTS) and ferry them to the peroxisomal membrane. Subsequently, the cargo is translocated into the peroxisomal lumen, whereas the receptor is released to the cytosol for further rounds of protein import. This cycle is controlled by the ubiquitination status of the receptor, which is best understood for the PTS1-receptor. While polyubiquitination of PTS-receptors results in their proteasomal degradation, the monoubiquitinated PTS-receptors are exported to the cytosol and recycled for further rounds of protein import. Here, we describe the identification of two ubiquitination cascades acting on the PTS2 co-receptor Pex18p. Using in vivo and in vitro approaches, we demonstrate that the polyubiquitination of Pex18p requires the ubiquitin-conjugating enzyme (E2) Ubc4p, which cooperates with the RING (really interesting new gene)-type ubiquitin-protein ligases (E3) Pex2p as well as Pex10p. Monoubiquitination of Pex18p depends on the E2 enzyme Pex4p (Ubc10p), which functions in concert with the E3 enzymes Pex12p and Pex10p. Our findings for the PTS2-pathway complement the data on PTS1-receptor ubiquitination and add up to a unified concept of the ubiquitin-based regulation of peroxisomal import.
Peroxisomes form a dynamic subcellular compartment in almost all eukaryotic cells . They can contain up to 50 different enzymes within their lumen, the peroxisomal matrix. These sets of enzymes link this organelle to distinct biochemical reaction pathways and physiological functions, which may vary depending on species, cell type or growth condition. The breakdown of fatty acids via beta-oxidation is considered to be a conserved function of peroxisomes, while other tasks linked to this organelle include the biosynthesis of plasmalogens and bile acid in mammals as well as the involvement in the photorespiration in plants or the biosynthesis of penicillin in certain fungi [2-4]. The vital role of peroxisomes is highlighted by the fact that dysfunction of human peroxisomes is associated with a spectrum of severe peroxisomal disorders like, e.g. Zellweger Syndome [5-7].
The functionality of this organelle is governed by dynamically operating import machineries for peroxisomal membrane and matrix proteins [8-10]. Matrix proteins usually harbor either a carboxy-terminal peroxisomal targeting sequence 1 (PTS1) or an amino-terminal PTS2 amino acid sequence, which is recognized by the soluble PTS1-receptor Pex5p or the PTS2-recognition factor Pex7p, respectively. The majority of peroxisomal matrix proteins is imported by the PTS1-receptor Pex5p. This cycling receptor binds its cargo in the cytosol and ferries it to a docking complex (Pex13p, Pex14p and Pex17p in yeast) at the peroxisome [8, 9]. The cargo is supposed to translocate over the membrane via a transiently opened import pore  and is released into the peroxisomal matrix. Finally, the PTS1-receptor is exported back to the cytosol in order to facilitate further rounds of matrix protein import. This final dislocation step is accomplished by the concerted action of certain peroxisomal membrane subcomplexes, collectively referred to as the exportomer . This molecular machinery comprises of enzymes required for the ubiquitination as well as the ATP-dependent extraction of the receptor from the membrane. A prerequisite for the dislocation step is the monoubiquitination of Pex5p on a conserved cysteine [13-16]. In yeast, this modification requires the presence of the peroxisomal RING (really interesting new gene)-finger complex, which consists of the E3 enzymes Pex2p, Pex10p and Pex12p, as well as the peroxisome-specific E2 enzyme Pex4p (Ubc10p) and its membrane anchor Pex22p [15-19]. The AAA (ATPases associated with various cellular activities)-type ATPases Pex1p and Pex6p dislocate the ubiquitinated Pex5p from the peroxisomal membrane back to the cytosol [13-15, 20, 21], where the ubiquitin is removed by deubiquitinating enzymes [22-24]. In case the export reaction is not working properly, Pex5p enters a quality control pathway, where it is polyubiquitinated on lysine residues and then degraded by the 26S proteasome. This alternative extraction pathway from the membrane depends on the formation of the polyubiquitin chain, which itself requires the presence of the RING-complex as well as the E2 enzyme Ubc4p or the partially redundant Ubc5p and Ubc1p [15, 16, 18, 25-27].
The concept of the PTS2-pathway is in many aspects comparable to the PTS1-dependent matrix protein import [28, 29]. The PTS2-recognition factor Pex7p cycles between the cytosol and the peroxisomal compartment [30, 31]. However, unlike the PTS1-receptor, Pex7p is necessary but not sufficient to carry out all steps of the import cycle as it requires species-specific auxiliary proteins. These PTS2-co-receptors are the redundant Pex18p and Pex21p in Saccharomyces cerevisiae, the orthologous Pex20p in Yarrowia lipolytica, Neurospora crassa, Hansenula polymorpha and Pichia pastoris or Pex5L, the longer of two splice isoforms of Pex5p, in mammals and plants . This bipartite assembly of the PTS2-receptor module is also evident in the molecular mechanism of the PTS2-protein import. In S. cerevisiae, Pex7p first binds the PTS2-cargo in the cytosol and then interacts with Pex18p before it can efficiently reach the peroxisomal membrane . Interestingly, the cargo-bound Pex7p can only translocate across the membrane when Pex18p is monoubiquitinated on the conserved cysteine . These results are in line with the ‘Export-driven import model’ , which suggests that matrix proteins can only be translocated and imported when the receptor is ubiquitinated and exported. However, the enzymes that catalyze this reaction so far have not been identified.
Here we describe the discovery of two ubiquitination cascades acting on the PTS2-co-receptor Pex18p. We demonstrate that the polyubiquitin chains of Pex18p, which mediate the turn-over of Pex18p, are generated by the E2 enzyme Ubc4p in concert with the RING-ligases Pex2p and Pex10p, while the monoubiquitin signal, which is essential for matrix protein import and receptor recycling, is attached to the co-receptor by Pex4p and the RING-ligases Pex12p and Pex10p.
Monoubiquitination of myc-Pex18p is catalyzed by the E2-enzyme Pex4p, while poly-ubiquitination is carried out by Ubc4p
Ubiquitination is a posttranslational modification that requires the activity of a three-step enzyme cascade . The ubiquitin-activating enzyme (E1) activates the ubiquitin via an AMP-bound intermediate and transfers it to the ubiquitin-conjugation enzyme (E2). In a final step, a protein-ubiquitin ligase (E3) binds both E2 and substrate and thereby facilitates the transfer of the ubiquitin moiety onto the substrate protein. S. cerevisiae has genes coding for 1 E1 enzyme, 11 E2 enzymes and approximately 80 E3 enzymes [36, 37].
We wanted to identify the requirements for the ubiquitination of the PTS2-co-receptor Pex18p. In the past, analysis of Pex18p ubiquitination, especially mono-ubiquitination was hampered by its low amount. In a previous study, we have isolated and concentrated Pex18p via co-purification with the Pex14p-complex . In order to study the modification of Pex18p in a direct manner, we optimized the preparation of Pex18p controlled by biochemical and mass spectrometry-based approaches. For this purpose, we constructed an amino-terminal myc-tagged version of Pex18p, which is expressed under the control of a copper-inducible promoter in yeast cells (Figure S1A). We verified the functionality of the fusion-protein by assaying its capability to complement the growth defect of co-receptor-deficient peroxisomal mutants on medium with oleate as sole carbon-source (Figure 1A). Because the beta-oxidation pathway is restricted to the peroxisomal compartment in yeast, the functionality of peroxisomes can be determined by the consumption of oleate and cell growth on oleate plates . The strain deleted for the redundant PTS2-co-receptors Pex18p and Pex21p does not grow on oleate , while the same strain transformed with a plasmid encoding for myc-Pex18p is capable to grow under these peroxisome-selective conditions (Figure 1A). In a previous study, we have analyzed the complementation properties of certain point mutated species of Pex18p that carried no fusion-tag and that were expressed under the control of the PEX18-promotor . The conserved cysteine at position 6 was mutated to a serine in order to inhibit monoubiquitination, while the lysine residues 13 and 20 were exchanged for arginine residues to block polyubiquitination . Because the non-mutated and the K13R/K20R mutant were able to complement the double deletion strain, while the C6S failed to do so, we can conclude that the myc-tag does not interfere with the functionality of Pex18p because it behaves similar to the non-tagged version.
In order to identify the ubiquitin-conjugating enzymes that are required for mono- or polyubiquitination of myc-Pex18p, we performed a series of immunoprecipitations by using antibodies against the myc-tag. We discovered that several high-molecular weight species of myc-Pex18p accumulate when the cells were lyzed in the presence of the proteasome inhibitor MG132 (Figure 1B). These slower migrating species were also detectable with ubiquitin antibody and were absent when MG132 was omitted or in case the K13/20R mutant of myc-Pex18p was used (data not shown). These polyubiquitinated Pex18p forms are linked to the rapid turn-over of Pex18p in wild-type cells [33, 40]. While it is known that the polyubiquitination of the PTS1-receptor Pex5p depends primarily on Ubc4p [17, 25, 26] and the partially redundant Ubc5p [25, 26] and Ubc1p , the E2 enzyme required for the polyubiquitination of the PTS2-co-receptor Pex18p has not been described before. We tested if the formation of polyubiquitinated myc-Pex18p species depends on the presence of one of the E2 enzymes of the Ubc4p-family (Figure 1B). While polyubiquitinated myc-Pex18p could still be isolated from strains deleted for UBC5 and UBC1, no modification of myc-Pex18p was detected in the ubc4Δ strain. The finding that the formation of ubiquitin chains requires the presence of Ubc4p strongly indicates that this Ubc-enzyme acts as E2 for the formation of polyubiquitinated myc-Pex18p species.
The export of the PTS1-receptor Pex5p, as well as the PTS2-co-receptors Pex18p in S. cerevisiae and Pex20p in P. pastoris depends on the monoubiquitination of a conserved cysteine [13-16, 33, 41, 42]. While it is known for Pex5p from S. cerevisiae that this modification is catalyzed by the peroxisomal Ubc-enzyme Pex4p [15, 16], the E2 protein responsible for the monoubiquitination of Pex18p remained unknown. Because the position of the cysteine is conserved between Pex5p and Pex18p, we investigated whether the PTS2-co-receptor may also be a potential target of Pex4p. For this purpose, we isolated a myc-Pex18p-variant that carries the K13R and K20R double point mutation that blocks the polyubiquitination of PTS2-co-receptor. This is of importance because the deletion of PEX4 elicits the polyubiquitinaton of Pex18p, which would prevent the clear view on potential monoubiquitinated species in this experiment . Furthermore, Pex18p is only transiently monoubiquitinated and therefore this modification can only be visualized by inhibition of deubiquitinating enzymes with NEM . Hence, we isolated myc-Pex18p(K13R;K20R) from wild-type and pex4Δ cells in the presence of NEM (Figure 1C). The result clearly demonstrated that while myc-Pex18p was monoubiquitinated in wild-type cells, the modification was completely absent in the pex4Δ cells. A similar result was obtained, when the cysteine-mutant myc-Pex18p(C6S;K13R;K20R) was used, which lacks the target amino acid for monoubiquitination. These findings strongly indicate that Pex18p represents the second known target of Pex4p in S. cerevisiae.
The E3 activity of all three RING-peroxins is essential for the PTS2-depedent matrix protein import
We were interested to identify the E3 enzymes that regulate the import of PTS2-proteins and catalyze the different ubiquitination modifications of Pex18p. An important family of E3 enzymes contains a RING-domain, which is essential for the ubiquitin-protein ligase activity of these proteins . The canonical RING domain coordinates two zinc ions in a cross-brace manner via one histidine and seven cysteine residues. In the case of the RING-peroxins, Pex2p and Pex10p harbor a well conserved typical RING motif that coordinates two zinc ions, whereas the RING of Pex12p can only bind one zinc ion . The RING-peroxins form a distinct subcomplex at the peroxisomal membrane [45, 46]. Recent work demonstrates that Pex2p [18, 47], Pex10p [18, 27, 47] and Pex12p [18, 47] display ubiquitin-protein ligase activity in S. cerevisiae and Arabidopsis thaliana. The only known physiological target of components of the RING-complex both in vivo and in vitro is the PTS1-receptor Pex5p [18, 27]. Previous studies demonstrated that the overall function of peroxisomes in general as well as the PTS1-protein import in particular, are abrogated in cells harboring RING-peroxin variants that lack an intact RING-domain [18, 48, 49].
In this study, we generated novel point mutated plasmid-encoded versions of the RING-peroxins Pex2p(C238S), Pex10p(C301S) and Pex12p(C354S), which are expressed under the control of their own promoters (Figure 2A). We verified that each of the corresponding recombinant RING-domains has lost its E3 activity by in vitro autoubiquitination assays (data not shown). Because these mutations affect one of the zinc-coordination sites, we first tested whether these RING-mutants could still perform their functional role in the biogenesis of peroxisomes (Figure 2B). For this purpose, we performed growth tests on oleic acid as sole carbon source. Because the functionality of peroxisomes is essential for growth of S. cerevisiae on this carbon source, a growth defect is indicative for peroxisomal dysfunction . The intact as well as the mutated RING-peroxin genes were expressed in the corresponding deletion strains (Figure 2B). The wild-type strain served as positive control, which did grow and displayed halos around the dilution spots, indicating consumption of the oleic acid. In contrast, the untransformed deletion strains stopped to grow early. The deletion strains that were transformed with the corresponding intact RING-peroxins displayed a growth behavior similar to the wild-type strain, while all point mutated variants with substitution of conserved cysteine residues failed to complement the deletion strains (Figure 2B). These data demonstrate that the RING-peroxin mutants Pex2p(C238S), Pex10p(C301S) and Pex12p(C354S) are not capable to function as a substitute for the wild-type proteins in supporting the general functionality of peroxisomes.
In order to address the question, whether the intact RING-domains of Pex2p, Pex10p and Pex12p are not only required for PTS1-dependent matrix protein import [18, 50-52] but also for PTS2-import, we monitored oleic acid-induced cells by fluorescence microscopy (Figure 2C). The synthetic marker protein PTS2-DsRed  revealed a punctate staining pattern in wild-type cells, which is typical for labeling of peroxisomes with intact PTS2-import pathway (Figure 2C). In contrast, PTS2-DsRed exhibited an overall labeling of the cells, indicating that the marker is mislocalized to the cytosol of the untransformed deletion strains pex2Δ, pex10Δ and pex12Δ. This import defect was corrected by expression of the corresponding non-mutated RING-peroxin gene. However, when the mutated variants of the RING-peroxins were expressed, a cytosolic distribution of the PTS2-DsRed was observed, which was comparable with the staining revealed from the untransformed deletion strains. The observed biogenesis and import defects are caused by the functional loss of the corresponding RING-peroxin because the cellular amount of the point mutant is similar to the non-mutated form (Figure S1B). The results demonstrate that the mutations of the RING-domain result in a loss of function of the proteins and that an intact RING-domain and therefore E3 activity of each RING-peroxin is essential for PTS2-dependent matrix protein import.
Polyubiquitination of Pex18p is catalyzed by the E3 enzymes Pex2p and Pex10p in vitro and in vivo
Depending on the ubiquitin-conjugating enzyme, Pex18p is modified in two different ways (Figure 1B,C). We were able to demonstrate that polyubiquitination of myc-Pex18p depends on the E2 enzyme Ubc4p. As next step, we aimed at the identification of the E3 enzyme that cooperates with Ubc4p in this process. For this purpose, we performed in vitro ubiquitination assays with recombinant and purified proteins (Figure 3A). This approach should reveal which E3 proteins are capable to perform the transfer of ubiquitin to the substrate in a direct reaction.
The peroxisomal RING-type ligases Pex2p, Pex10p and Pex12p were considered as possible candidates because they are involved in the different ubiquitin-modifications of the PTS1-receptor Pex5p [18, 27]. We used RING-domain-containing truncations of the RING-peroxins, which comprised the coding sequences of Pex2p aa 215–271 (Pex2p-RING), Pex10p aa 238–337 (Pex10p-RING) and Pex12p aa 293–399 (Pex12p-RING) fused to GST [18, 44, 54]. The recombinant fusion proteins of the RING-peroxin truncations as well as the full-length His-Ubc4p and the target protein Pex18p-His were isolated by affinity chromatography and subjected to the in vitro ubiquitination reactions in different combinations. Modified forms of Pex18p-His were absent when ubiquitin, E1 or Pex18p-His were omitted from the reaction (Figure 3A). Interestingly, Pex18p-His remained unmodified in the presence of GST-Pex12p(RING), while polyubiquitination of the His-tagged PTS2-co-receptor was observed when the RING-domain of either Pex2p or Pex10p was present in the assay (Figure 3A). The modified Pex18p-His species were both detectable with Pex18p-antibody (Figure 3A) as well as ubiquitin-antibody (data not shown). In summary, the in vitro data demonstrate that Pex18p-His can be polyubiquitinated by Ubc4p in cooperation with Pex2p(RING) or, alternatively, Pex10p(RING).
In order to investigate, if the observed E3 requirement is also relevant in the living cell, we purified myc-Pex18p and its polyubiquitinated forms from strains expressing different RING-peroxin activity mutants (Figure 3B). The immunoprecipitation experiments were carried out in a similar manner as described for the analysis of the E2 requirement of the myc-Pex18p polyubiquitination. However, in the present experiment, we applied pex2Δ cells transformed with constructs coding for Pex2p or Pex2p(C238S), pex10Δ cells harboring Pex10p or Pex10p(C301S) as well as the pex12Δ strain expressing the plasmid-coded genetic information for Pex12p or Pex12p(C354S). Interestingly, it turned out that the polyubiquitination of myc-Pex18p remained unaffected in all cases, thus even when the corresponding deletion strains were transformed with the corresponding E3 activity mutants (Figure 3B). The result suggested an unexpected redundancy of certain RING-peroxins in myc-Pex18p polyubiquitination. In order to find out if all three RING-peroxins or just two of them act redundantly on myc-Pex18p, we analyzed the combinations of one non-mutated RING-peroxin with two mutated ones (Figure 3C). We observed a block of polyubiquitination in the double deletion strain pex2Δpex10Δ, while the modification was comparable to wild-type when Pex10p(C301S) was co-expressed with Pex2p. At the end, it turned out that polyubiquitination of myc-Pex18p was only abolished, when both activity mutants, Pex10p(C301S) and Pex2p(C238S), were combined. (Figure 3C). These data indicate that the endogenous Pex12p alone is not capable to catalyze the ubiquitination-reaction and that Pex2p and Pex10p act in a redundant manner in myc-Pex18p-ubiquitination. This finding is in agreement with the results of the in vitro assay (Figure 3A). Therefore, we conclude that the Ubc4p-dependent polyubiquitination of myc-Pex18p is catalyzed by the E3 enzymes Pex2p and Pex10p.
Monoubiquitination of Pex18p is catalyzed by the E3 enzymes Pex12p and Pex10p in vitro and in vivo
We wanted to identify the E3 enzyme that is responsible for Pex18p-monoubiquitination. In a similar approach as described for the analysis of the E3 requirement for the generation of polyubiquitin chains, we performed in vitro ubiquitination assays with heterologously expressed and purified proteins (Figure 4A). This defined experimental setup should be capable to investigate which E2 and E3 components are directly involved in the transfer of ubiquitin to the substrate. The peroxisomal RING-type ligases Pex2p, Pex10p and Pex12p were analyzed as possible candidates. We made use of the truncated RING-constructs, which comprise the RING-domain region fused to GST [18, 44, 54]. The GST-RING-domain constructs as well as the full-length Pex4p and Pex18p-His were isolated by affinity chromatography and subjected to the in vitro ubiquitination reactions in different combinations. Modified species of Pex18p-His were not detectable when the E1 enzyme, ubiquitin or Pex18p-His were omitted from the reaction (Figure 4A). Interestingly, the combination of Pex4p with GST-Pex12p(RING) as well as the one with GST-Pex10p(RING) resulted in the generation of monoubiquitinated Pex18p-His species. However, Pex18p-His remained unmodified when GST-Pex2p(RING) was used as E3 factor. Therefore, the in vitro data confirm that Pex18p is indeed a substrate of Pex4p and lead to the conclusion that Pex18p-His can be monoubiquitinated by the E3 enzyme Pex12p(RING) or, alternatively, Pex10p(RING).
In order to analyze, if the E3 requirement found in vitro is also of physiological relevance in vivo, we purified myc-Pex18p and its monoubiquitinated forms from strains expressing different RING-peroxin activity mutants (Figure 4B). The immuno-precipitation experiments were performed in a comparable way as described for the analysis of the E3 enzymes involved in the polyubiquitination of myc-Pex18p. This meant that we used pex2Δ cells harboring Pex2p or Pex2p(C238S), pex10Δ cells transformed with either Pex10p or Pex10p(C301S) constructs as well as the pex12Δ strain with Pex12p or Pex12p(C354S). Notably, the only difference to the experiment described for the analysis of polyubiquitinated myc-Pex18p (Figure 3B) was that we supplemented the cell lysate during preparation with NEM and not with MG132 (Figure 4B). Again it turned out that also the monoubiquitination of myc-Pex18p remained visible even in the cells where the corresponding deletion strain was transformed with the Ub-ligase activity mutants (Figure 4B). Again we observed a certain redundancy of RING-peroxins, but now in for myc-Pex18p monoubiquitination. As a consequence, we analyzed the combinations of one non-mutated RING-peroxin with two mutated proteins in vivo (Figure 4C). The double deletion strain pex12Δpex10Δ served as negative control. Monoubiquitination of myc-Pex18p was clearly detectable when Pex10p(C301S) was co-expressed with Pex12p. The modification was absent when both activity mutants, Pex10p(C301S) and Pex12p(C354S), were combined. These results strongly suggest that the endogenous Pex2p alone is not sufficient to catalyze the monoubiqutination and that Pex12p and Pex10p act in a redundant manner on myc-Pex18p. This finding corroborates the results of the in vitro assays (Figure 4A). Therefore we conclude that both Pex12p and Pex10p can act as E3 component of the modification cascade resulting in the Pex4p-dependent monoubiquitination of the PTS2-co-receptor Pex18p.
The PTS2-co-receptor Pex18p of S. cerevisiae was the very first peroxin that has been found to be ubiquitinated, when modified forms of Pex18p were described in 2001 . More recently, we were able to characterize this posttranslationally modified PTS2-co-receptor in more detail and discovered that Pex18p is polyubiquitinated on two lysine residues and monoubiquitinated on a cysteine residue . In this study, we describe two ubiquitination cascades that are responsible for these two different ubiquitin-modifications (Figure 5).
The only other peroxisomal protein known to be posttranslationally modified in a similar manner is the PTS1-receptor Pex5p of S. cerevisiae. Interestingly, our recent data indicate that both receptor proteins share similarities concerning their ubiquitination requirements, but also display some differences. It has been noted before that the PTS2-co-receptor Pex18p resembles the amino-terminal half of the PTS1-receptor Pex5p. This was based on similarities in the amino acid sequence and shared binding partners at the peroxisomal docking complex  as well as the fact that a chimeric protein, consisting of the cargo-binding C-terminus of Pex5p and Pex18p lacking its Pex7p-binding site, can import PTS1-proteins . As shown recently, the stability of both Pex5p and Pex18p depends on polyubiquitination on two conserved lysine residues, while membrane topology and export are governed in both cases by the monoubiquitination on a conserved cysteine [15, 16, 33]. In this study, we shed more light on the similar and therefore mechanistic related modules of the PTS1- and PTS2-receptor ubiquitination machineries, by the discovery that both receptor proteins utilize the same ubiquitin-conjugating enzymes. We demonstrate that Ubc4p catalyzes the poly-ubiquitination of Pex18p both in vivo and in vitro. This is also the case for S. cerevisiae Pex5p [17, 18, 25, 26]. Even more significant is the finding that Pex4p is required for the monoubiquitination of Pex18p. After Pex5p [15, 16], Pex18p represents only the second known physiological target of this peroxisomal localized E2 enzyme that has been found to be a substrate both in vivo and in vitro.
However, there are also variations to this theme. In general, the presence of all three constituents of the peroxisomal RING-complex Pex2p, Pex10p and Pex12p is essential for the ubiquitination of Pex5p [17, 18, 25-27] as well as of Pex18p, as demonstrated here. On the basis of work with truncated versions that lacked the catalytic RING-domain in vivo as well as in vitro ubiquitination assays, Pex12p has been described to represent the E3 enzyme responsible for the monoubiquitination of Pex5p, while Pex2p was primarily found to be responsible for the polyubiquitination . Another in vivo study noticed that a Pex10p-mutant reduces the polyubiquitination of Pex5p , while a recent in vitro study suggested that Pex10p synergistically enhances the E3 activity of the Pex12p/Pex4p and Pex2p/Ubc4p enzyme pairs . To our surprise, we found that Pex18p can be directly monoubiquitinated by Pex4p in concert with Pex10p as well as Pex12p in vitro and that both E3 enzymes act redundantly also in vivo. Moreover, we could show that the Ubc4p-dependent polyubiquitination of Pex18p is carried out by Pex10p and Pex2p in vivo and in vitro.
The direct involvement of two E3 enzymes for each modification in the case of Pex18p is in contrast to the situation found for the PTS1-receptor and might be explained by one of the main regulatory differences between these two proteins. Pex5p is relatively stable under wild-type conditions and is only significantly polyubiquitinated when certain components of the exportomer are missing or non-functional . In contrast, Pex18p displays a high turn-over rate already under wild-type conditions [33, 40]. Thus, while the monoubiquitination of Pex5p is the dominant modification under normal conditions, the modifications of the Pex18p population is expected to be divided between recycling-related mono- and degradation-linked polyubiquitination. This circumstance might deliver an explanation and working model why two E3 enzymes are linked to each ubiquitination form of Pex18p.
The decision whether Pex18p enters the mono-ubiquitination-dependent recycling pathway or the polyubiquitination-dependent degradation pathway may be hypothetically related to the binding dynamics of Pex18p. In case Pex18p is functionally impaired, it may stay longer at the membrane complexes because it is not efficiently recognized by the recycling machinery and therefore may render it better accessible to the polyubiquitination machinery that normally should have a lower affinity to Pex18p. However, this model requires further elucidation.
The finding that the Pex10p(C301S) mutant does not complement the PEX10-deletion strain on oleate plates, even though its functional loss can be compensated by Pex12p for the essential monoubiquitination, indicates that Pex10p might have other important functions as well. One possibility could be that Pex10p ubiquitinates also other peroxisomal biogenesis factors. One hypothesis could even be that this important additional substrate is Pex10p itself, because it has been demonstrated to undergo autoubiquitination in vitro [18, 27, 47]. It is known from several RING-type ligases that their autoubiquitination is an important regulatory device in vivo [57, 58]. However, if this is also the case for Pex10p remains to be elucidated.
A recent publication described the ubiquitination requirements of Pex20p in P. pastoris . While S. cerevisiae utilizes the partially redundant Pex18p and Pex21p as PTS2-co-receptors for Pex7p , P. pastoris contains Pex20p as sole PTS2-co-receptor protein . In contrast to the situation described for S. cerevisiae Pex18p, the P. pastoris study indicated that all three RING-peroxins are required for both the monoubiquitination as well as the polyubiquitination of Pex20p in vivo . One possible explanation for the observed different E3 requirement of these two PTS2-co-receptors may be species-specific differences in the regulation of the PTS2-pathway. One significant functional difference is the finding that Pex20p seems to cycle between peroxisome and cytosol even without Pex7p [41, 59], while Pex18p needs the presence of cargo-bound Pex7p to reach efficiently the peroxisomal membrane . Furthermore, Pex20p of H. polymorpha and Y. lipolytica has been reported to interact directly with the cargo [60, 61], which is not the case for Pex18p, because the association is always bridged by Pex7p [32, 39, 53]. A methodical difference between our study and the Pex20p study concerns the RING-mutants. While the Pex20p study uses RING-peroxins that carry two mutated cysteines in the first zinc-coordinating motif of each RING-domain, our study utilizes RING-variants that carry a single cysteine mutation in the second zinc-coordination sites of Pex2p and Pex10p or a single cysteine mutation in the single zinc-coordination motif of Pex12p (Figure 2A). We decided to introduce a single point mutation in the mentioned positions because this is supposed to inhibit the E3 activity without causing major structural damage to the protein . In fact, we observed before that the mutated RING-finger of Pex10p(C301S) has no E3 activity but still interacts with the RING-domains of Pex2p and Pex12p .
Another difference concerns the data that a pex4Δ strain of P. pastoris not only inhibits the monoubiquitination, but also leads to shorter polyubiquitin chains and slowed down degradation of the Pex20p cysteine-mutant. This led the authors to suggest that Pex4p might have a minor role in the polyubiquitination of Pex20p . However, we do not observe a direct role of S. cerevisiae Pex4p in the generation of polyubiquitin chains on Pex18p in vivo and in vitro.
Our findings indicate that the heterotrimeric RING-peroxin complex displays in part a differentiated responsibility of its three ligases toward the two known substrates, Pex5p and Pex18p. It is interesting to note that similar observations have been reported on the only other described heterotrimeric RING-protein complex with E3 activity, the Polycomb complex consisting of Ring1a, Ring1b and Bmi1. In this case, Ring1a, but not Ring1b, uses topoisomerase Top2alpha as target . Still, Ring1a is also capable to act on the histone H2A in a redundant manner to Ring1b . However, it is not known how the access of both Ring1a and Ring1b toward H2A is coordinated or how ubiquitination of Top2alpha by Ring1b is prevented.
Another interesting point concerns the finding that Pex10p is in principle capable to catalyze mono- as well as polyubiquitination of Pex18p. While RING-type ligases, like MDM2, facilitate the (multiple) monoubiquitination of their target at a low concentration, the same target gets polyubiquitinated at a high concentration of the ligase . Another example of a mode of action used by RING-ligases that catalyze different ubiquitin-modifications is the involvement of selective partners during catalysis, like different E2 enzymes . This latter scenario is comparable to the situation found for Pex10p in the Pex4p-dependent mono- and Ubc4p-dependent polyubiquitination of Pex18p. However, the molecular switch between the interplay with either Pex4p or Ubc4p remains to be defined.
Our data on the two ubiquitination cascades acting on Pex18p complement the findings regarding the ubiquitination requirements of Pex5p in the PTS1 pathway and therefore add up to a unified concept of the ubiquitination machinery of the peroxisomal import receptors. Concerning E2/E3 enzyme pairs, the Ubc4p/Pex2p axis is linked to the polyubiquitination and degradation of the receptor proteins, while the Pex4p/Pex12p axis plays a central role in monoubiquitination, receptor recycling and matrix protein translocation. It is interesting to note that Pex10p displays an additional function in the PTS2-pathway by directly ubiquitinating Pex18p. In the future, it will be important to elucidate how the additional Pex10p-requirement is mechanistically integrated in the interplay with the other E3 enzymes and how this process contributes to the translocation of folded matrix proteins over the peroxisomal membrane.
Materials and Methods
Yeast strains and culture conditions
The S. cerevisiae strain UTL-7A (MATa, ura3-52, trp1, leu2-3/112) was used as wild-type strain for the generation of several isogenic deletion strains by the ‘short flanking homology’ method as described . The following deletion strains were generated by this method: pex2Δ , pex4Δ , pex10Δ , pex12Δ , pex18Δ , ubc1Δ, ubc4Δ, ubc5Δ , pex2Δ/pex10Δ (this study), pex2Δ/pex12Δ (this study), pex10Δ/pex12Δ (this study) and pex18Δ/pex21Δ . Yeast media have been described previously .
Plasmids and cloning strategies
The myc fused PEX18 versions were constructed by cloning of the open reading frame of PEX18 into pRSmyc5 or pRSmyc6  using the primers RE3268 (TTTAAATGATCATG AATAGTAACCG ATGCCAAAC) and RE2983 (GA GCTCGAGTTAA GCAATTCTGTCTTCAACATC). The point mutation C6S was introduced by using the primers RE3269 (TTTAAATGATCATGAATA GTA ACCGATCC CAAAC) and RE2983 (GAGCT CGAGT TAAGCAA TTCTGTCTTCAACATC). The resulting DNA-fragment was cloned into the host vector (BclI, XhoI). The K13R/K20R mutations were introduced by using the primers RE3268 (TTTAAATGATCA TGAATAGTAACCGATGCCAAAC) and RE2983 (G ACTCGAGTTAAGCAATTCTGTCTTCAACATC). Site-directed mutagenesis was applied to construct the mutated PEX2, PEX10 and PEX12. Using the primers RE2700 (CATACCAGATCGCTTGT AGTC GTGCGAACTACTGT) and RE2701 (ACAGTAGTT CGCACGACTACAAGCGATCTGGTATG) and pRS316-PEX2  as template resulted in the point mutation C238S. The C301S mutation was constructed by using the primer RE2702 (TCCTAGCTGCGCGCCAAGTGGACATCTATTTTG T) and the primer RE2703 (ACAAAATA GATGTCCACTTGGCGCGCAGCTAGGA) and pRS316-PEX10 as a template. The pRS316-PEX12 was mutated by using the primers RE3065 (TGGAAACAGGATACGTGGCAAGCTACCCGTGT) and RE 3066 (ACACGGGTAGCTTGCCACGTA TCCTGTTTCCA), resulting in Pex12p(C354S).
The plasmids encoding for GST-Pex2p (aa 215–271), GST-Pex4p, GST-Pex10p (aa 238–337), GST-Pex12p(aa 293–399), His-Ubc4p , Pex18p-His  and DsRed-PTS2  have been described previously.
Genes coding for recombinant proteins were expressed in Escherichia coli BL21(DE3). Cells were harvested, diluted in phosphate-buffered saline (PBS, 137 mm NaCl, 2.7 mm KCl, 4.3 mm Na2HPO4, 1.4 mm KH2PO4, pH 7.3), containing protease inhibitors (8 µm antipain, 0.3 µm aprotinin, 1 µm bestatin, 10 µm chymostatin, 5 µm leupeptin, 1.5 µm pepstatin, 1 mm benzamidine and 1 mm PMSF (Boehringer) and broken using a French press. The 100 000 × g supernatant containing the soluble fusion proteins was loaded on affinity columns. The affinity purification of proteins from cells expressing GST-Pex12p (RING) and GST-Pex4p has been described previously . Purification of His-fusion proteins like His-Ubc4p and Pex18p-His was performed as described earlier [18, 32]. In this case, the Pex18p-His was obtained by renaturation of the former insoluble Pex18p-His . Recombinant yeast E1 and ubiquitin were purchased from Sigma.
In vitro ubiquitination assays
In vitro ubiquitination of recombinant Pex18p-His contained 0.85 µg of Pex18p-His, 5 µg Ubiquitin, 0.65 µg of GST-RING fusion proteins, 0.1 µg of yeast E1, 0.8 µg E2-enzyme as indicated and 10 μL buffer containing 2 mm ATP, 50 mm Tris-HCl pH 7.5, 2 mm MgCl2, 1.5 µm ZnCl2 and 0.1 mm dithiothreitol. After incubation at 30°C for 90 min with gentle shaking, the reaction was stopped by addition of SDS-sample buffer and heating at 95°C for 5 min.
In vivo ubiquitination assays
Oleate-induced yeast cells were harvested, washed twice and resuspended in lysis-buffer (0.2 m HEPES, 1 m potassium acetate and 50 mm magnesium acetate, pH 7.5) and protease inhibitors 8 µm antipain, 0.3 µm aprotinin, 1 µm bestatin, 10 µm chymostatin, 5 µm leupeptin, 1.5 µm pepstatin, 1 mm benzamidine and 1 mm PMSF; 5 mm NaF—Boehringer). For the preparation of polyubiquitinated mycPex18p from wild-type cells, lysis buffer was supplied with 100 µm MG132 (Sigma). In order to stabilize the monoubiquitinated form of mycPex18p, a final concentration of 20 mm NEM (Sigma) was added to the buffer. The breakage of the cells was achieved by adding 3 g of glass beads (0.5 mm) followed by vortex-mixing (ten 60 seconds bursts with breaks of at least 60 seconds on ice). Samples were transferred to Corex tubes and centrifuged at 1500 × g (Rotor) for 10 min. Supernatants were normalized for protein concentration and volume, and membranes were sedimented by centrifugation at 100 000 × g, followed by trichloroacetic acid precipitation and sample preparation .
Immunoprecipitation of denatured proteins was carried out using acetone-washed and dried sediments of trichloroacetic acid precipitated samples. The sediment was resuspended in 100 μL of urea cracking buffer (50 mm Tris–Cl, pH 7.5, 6 m urea, 1% SDS) and incubated for 10 min at 65°C. Subsequently, 1 mL of Tween 20-IP buffer [50 mm Tris–Cl, pH 7.5, 150 mm NaCl, 0.5% Tween 20, 0.1 mm ethylenediaminetetraacetic acid (EDTA)] and 10 μL of 100 mg/mL bovine serum albumin were added. After sedimentation of non-dissolved material, myc-antiserum was added, and the mixture was incubated under continuous swirling for 4 h at 4°C. Subsequently, 80 μL of myc- antibody preloaded Dynabeads-Anti-Mouse IgG (Invitrogen) were added, and the mixture was further incubated for 1 h at 4°C. The immunoprecipitated material was subsequently subjected to sedimentation, washed twice with Tween 20-IP buffer, once with Tween 20-urea buffer (100 mm Tris–Cl, pH 7.5, 200 mm NaCl, 2 m urea, 0.5% Tween 20), and once with Tris-buffered saline buffer (50 mm Tris–Cl, pH 7.5, 150 mm NaCl). Finally, the beads were boiled in 50 μL of IP-sample buffer (125 mm Tris–Cl, pH 6.8, 6% SDS, 10% ß-mercaptoethanol, 20% glycerol, 0.1% bromphenol blue) and prepared for immunoblot analysis .
Analysis of live cells for dsRed fluorescence microscopy was performed with a Zeiss Axioplan microscope and axiovision 4.1 software (Zeiss). Before inspection, cells were grown for 2 days on solid minimal medium containing oleic acid as a sole carbon source .
TCA precipitated cell lysate were separated by SDS-PAGE 4–12% NuPAGE™ Bis-Tris Gel (Invitrogen) and bands were cut out the local area of the gel. In the next steps, the in-gel digestion with trypsin (Promega), the peptide extraction and the measurement on the electrospray tandem mass spectrometry (ESI-MS/MS) was performed as previously described .
Immuno-reactive complexes were visualized using anti-rabbit- IgG IRDye800CW conjugated or anti-mouse IRDye800CW conjugated in combination with the Odyssey® infrared imaging system from LI-COR, Biosciences. Polyclonal rabbit antibodies were raised against Pex18p  and ubiquitin (Sigma). Monoclonal mouse antibodies were raised against myc (clone 9E10).
This work was supported by grants of the Deutsche Forschungsgemeinschaft to H.E.M. (SFB 642), R. E. (FOR 1905, SFB 642) and H. W. P. (FOR 1905, SFB 642).