Yeast Pex14p possesses two functionally distinct Pex5p and one Pex7p binding sites.

Current evidence favors a cycling receptor model for the import of peroxisomal matrix proteins. The yeast Pex14 protein together with Pex13p and Pex17p form the docking subcomplex at the peroxisomal membrane and interact in this cycle with both soluble import receptors Pex5p and Pex7p. In a first step of a structure-function analysis of Saccharomyces cerevisiae Pex14p, we mapped its binding sites with both receptors. Using the yeast two-hybrid system and pull-down assays, we showed that Pex5p directly interacts with two separate regions of ScPex14p, amino acid residues 1-58 and 235-308. The latter binding site at the C terminus of ScPex14p overlaps with a binding site of Pex7p at amino acid residues 235-325. The functional assessment of these two binding sites of ScPex14p with the peroxisomal targeting signal receptors indicates that they have distinct roles. Deletion of the N-terminal 58 amino acids caused a partial defect of matrix protein import in pex14delta cells expressing the Pex14-(59-341)-p fragment; however, it did not lead to a pex phenotype. In contrast, truncation of the C-terminal 106 amino acids of ScPex14p completely blocked this process. On the basis of these and other published data, we propose that the C terminus of Pex14p contains the actual docking site and discuss the possibility that the N terminus could be involved in a Pex5p-Pex14p association inside the peroxisomal membrane.

A crucial step of the receptor cycle, which could provide directionality to the process, is the docking event at the peroxisomal membrane (1,2). Studies from several laboratories provide evidence that Pex14p is the central component of a docking complex, which also contains Pex17p and Pex13p. As Pex14p interacts with several other membrane-bound proteins in addition to both PTS receptors, it has been proposed to be the point of convergence of the two peroxisomal import pathways (6).
There are conflicting data concerning the nature of the association of Pex14p with the peroxisomal membrane, and the topology of the protein is still not entirely solved. Although we originally report that Saccharomyces cerevisiae Pex14p is extractable with carbonate (6), others found that Pex14p from S. cerevisiae (7) and other organisms (8 -13) behaves as an integral membrane protein. However, there is agreement that the C terminus of the protein is exposed to the cytosol (9,10,12,13). Rattus norvegicus Pex14p was found to be tightly associated with a fraction of Pex5p that behaves as an integral membrane protein (13). It was shown that the first 130 amino acids of Pex14p are highly protected from exogenously added protease by the peroxisomal membrane and that this domain is responsible for the strong interaction of Pex14p with the organelle membrane (12,13).
To explore the function of Pex14p in greater detail, we mapped its binding sites for the two PTS receptors. Unexpectedly, two separate and functionally different binding sites were found for the PTS receptor Pex5p, one in each of the two termini. The C-terminal binding site overlapped with a single binding site determined for Pex7p and was essential for matrix protein import. In contrast, deletion of the N-terminal binding site reduced the efficiency but did not abolish matrix protein import. We proposed a model in which the cytosolically exposed C terminus of Pex14p served as the actual docking site and in which the N terminus could be involved in a Pex14p-Pex5p association inside the peroxisomal membrane.

EXPERIMENTAL PROCEDURES
Yeast Strains and Culture Conditions-The S. cerevisiae wild-type strain used in this study was UTL-7A (14). Yeast complete (yeast extract/peptone/dextrose) and minimal media (SD) have been described previously (15). The yeast knock-out strain pex14⌬ was a derivate of UTL-7A. To delete PEX14, the kanMX4 gene was used as a selective marker for insertion into the genomic locus (16). Deletion cassettes containing the kanMX4 gene and the 5Ј and 3Ј flanking regions of the Pex14p-ORF (open reading frame) were constructed by PCR using pFA6a-kanMX4 (16) as a template and Ku 289 (5Ј-gaaacctcaagtaaaacagagaagttgtaaggtgaataaggacgtacgctgcaggtcgac-3Ј) and Ku 290 (5Јaattacaatttccgttaaaaaactaattacttacatagaattgcgatcgatgaattcgagctcg-3Ј) as primers. For yeast two-hybrid experiments the yeast strain PCY2 was used (17).
Plasmids-For the expression of Myc-Pex7p, Pex14p and Pex14p fragments in a pex14 deletion strain of S. cerevisiae plasmids were cloned using the vectors pRS316, pRS416, and YEp351. For Pex14p and its derivatives, endogenous promoters and terminations were used. The resulting constructs are listed in TABLE ONE. Plasmids used for expressing His 6 -Pex14-(1-58)-p, GST-Pex14-(234 -341)-p, GST, and Pex5p in Escherichia coli are based on pET21d and pGEX-4T-2 vectors and are also listed in TABLE ONE. Detailed cloning strategies are available on request.
Anti-rabbit IgG-coupled horseradish peroxidase (Amersham Biosciences) was used as the secondary antibody. Immunoreactive complexes were visualized using anti-rabbit IgG-coupled horseradish peroxidase in combination with the ECL TM system from Amersham Biosciences. Proteins in polyacrylamide gels were visualized by Coomassie staining according to Ref. 24.
In Vitro Binding Assays-Transformed E. coli BL21(DE3) cells were cultured and induced with 1 mM isopropyl 1-thio-␤-D-galactopyranoside. Myc-Pex7p overexpression in pex14⌬ cells was induced by the addition of 100 mg of CuSO 4 ⅐5H 2 O/liter of culture containing 2% glucose. For the analysis of the interaction between Pex5p and Pex14-(1-58)-p-His 6 as well as the binding of GST-Pex14-(235-341)-p to Myc-Pex7p, 0.2 g of E. coli BL21(DE3) cells expressing Pex14-(1-58)-p-His 6 , GST-Pex14-(235-341)-p, GST, 0.3 g of E. coli BL21(DE3) cells expressing Pex5p, and 1 g of pex14⌬ yeast cells overexpressing Myc-Pex7p were resuspended in a 3ϫ volume of lysis buffer (for Ni 2ϩ -NTA columns, 200 mM Tris/HCl, pH 8.0, 150 mM NaCl, 5 mM imidazole; for GSH-Sepharose columns, 50 mM Tris/HCl, pH 7.5, 100 mM NaCl, 0.2% (w/v) Triton X-100). Cells were lysed by the addition of a 4ϫ volume of glass beads (ဧ 0.5 mm) and vortexing 10 times for 1 min each. Glass beads, cell debris, and unlysed cells were sedimented by a centrifugation step at 1500 ϫ g for 5 min. Thereafter, E. coli lysates were centrifuged for 10 min at 88,000 ϫ g and yeast lysates for 30 min at 200,000 ϫ g to pellet insoluble material. Supernatants were mixed and incubated with 100 l of Ni 2ϩ -NTA-agarose (Qiagen) or 100 l of GSH-Sepharose (GE Healthcare) as  Proteins were eluted with lysis buffer enriched with 10 mM reduced glutathione. Every two fractions were pooled before analysis. As protease inhibitors 1 mM phenylmethylsulfonyl fluoride, 5 mM NaF, 2 g/ml leupeptin, and 2 g/ml pepstatin were used. For the analysis of the binding between GST-Pex14-(235-341)-p and Pex5p, 2 g of cells expressing one or the other fusion protein or GST were used. Lysis buffer did not contain Triton X-100 but 1 mM dithiothreitol. As protease inhibitors, 2 M leupeptin, 2 M pepstatin, 200 M Pefabloc SC, 5 mM benzamidine, and 5 mM NaF were used. Cells were lysed using a French pressure cell press. Soluble fractions were obtained by centrifugation as described above, mixed, and incubated at 4°C for 1 h before they were subjected to columns with 0.8 ml of GSH-Sepharose. Washing was performed using 32 ml of lysis buffer. Recovered proteins were eluted with 16 ml of lysis buffer enriched with 10 mM reduced glutathione in 0.5-ml fractions.
Cell Fractionation-Spheroplasting of yeast cells, homogenization, and differential centrifugation at 25,000 ϫ g of homogenates were performed as described previously (15).
For separation of cell organelles by density gradient centrifugation, post-nuclear supernatants of wild-type and mutant strains were loaded onto continuous 20 -53% (w/v) sucrose density gradients (25 ml). The gradient buffer contained 5 mM MES, 1 mM EDTA, 1 mM KCl, 0.1% (v/v) ethanol/KOH, pH 6.0, at 4°C. After a centrifugation step of 1.5 h (Sorvall-SV288, 19,500 rpm), ϳ30 fractions of 1 ml each were collected. 500 l of each fraction of any gradient were processed for Western blotting using trichloroacetic acid, whereas the other part was used for enzyme measurements.
The suborganellar localization of proteins was determined by extraction of 25,000 ϫ g organelle pellets by either high salt treatment with buffer containing 10 mM Tris/HCl, pH 8.0, 500 mM KCl, and 1 mM phenylmethylsulfonyl fluoride or using carbonate buffer containing 100 FIGURE 1. Two-hybrid analysis of the interaction between Pex14p fragments and Pex5p or Pex7p, respectively. PCY2 was co-transformed with plasmids encoding proteins as indicated and tested for ␤-galactosidase activity using a filter assay and X-gal as a substrate. Three independent double transformants are shown. mM Na 2 CO 3 , pH 11.5, and 1 mM phenylmethylsulfonyl fluoride. After incubation for 30 min at 4°C, the samples were spun for 1 h at 200,000 ϫ g through a cushion of 10 mM Tris/HCl, pH 8.0, and 250 mM sucrose.
Two-Hybrid Assay-The two-hybrid assay was based on the method of Fields and Song (29). The open reading frames or coding regions of specific fragments of PEX genes were fused to the DNA binding domain or transcription-activating domain of GAL4 into the vectors pPC86 and pPC97 (17) (see TABLE ONE). Co-transformation of two-hybrid vectors into PCY2 yeast cells was performed according to Gietz and Sugino (30). Double transformants were selected on SD synthetic medium without tryptophan and leucine. ␤-galactosidase activity of transformed cells was determined by a filter assay described by Rehling et al. (31), using X-gal as substrate.
Electron Microscopy-Potassium permanganate fixation and preparation of intact yeast cells were performed according to Ref. 15.
Indirect Immunofluorescence Microscopy-Immunolabeling of yeast cells and fluorescence microscopy were performed as described by Rehling et al. (31).

Pex5p Binds the N and C Termini of ScPex14p-To understand how
ScPex14p facilitates the import of peroxisomal matrix proteins, we first mapped the binding sites of the two PTS receptors within this protein.
We used the yeast two-hybrid system and first tested N-and C-terminal deletion mutants of Pex14p for the binding of Pex5p. The short N-terminal fragment Pex14-(1-58)-p was chosen because it comprises the most conserved region of Pex14 proteins (data not shown), and Schliebs et al. reports that the analogous region of human Pex14p directly binds HsPex5p (32). As expected, Pex14-(1-58)-p did interact with the PTS1 receptor (Fig. 1A). Surprisingly, the complementary fragment Pex14-(59 -341)-p, which was actually meant as a control, also interacted with the PTS1 receptor. To test whether ScPex14p contains two separate binding sites of Pex5p, we used a series of N-terminal deletions and found that the C-terminal fragment Pex14-(235-341)-p interacts with Pex5p, whereas the core fragment Pex14-(59 -234)-p missing both termini does not (Fig. 1A). In addition, we verified the interaction of both termini of Pex14p with the PTS1 receptor using an in vitro assay. For this purpose, Pex14-(1-58)-p and Pex14-(235-341)-p (fused to a His 6 or GST tag, respectively) and Pex5p (without any tag) were expressed in E. coli. Both recombinant Pex14p fragments clearly bound Pex5p, whereas this was not the case in the controls (Fig. 2, A and B). These data strongly indicated the existence of two separate direct Pex5p binding sites on ScPex14p, one in the N-terminal 58 amino acid residues and one in the last 106 amino acid residues.
The C Terminus of Pex14p Also Interacts with Pex7p-To determine the binding region of Pex7p, we assayed the same truncated versions of ScPex14p as described above using again the two-hybrid system. Fig. 1B shows that the C-terminal fragment Pex14-(235-341)-p (which contains a binding site of the PTS1 receptor) also interacts with the PTS2 receptor. All other ScPex14p fragments did not bind to Pex7p (Fig. 1B; data not shown). An analogous pull-down experiment as described above for the binding of Pex5p (Fig. 2, A and B) was carried out with Myc-Pex7p and GST-Pex14-(235-341)-p. As Myc-Pex7p could not be stably overexpressed in E. coli, a yeast lysate containing Myc-Pex7p but no wild-type Pex14p was used. The results shown in Fig. 2C demonstrate the specific interaction of the PTS2 receptor with the C terminus of Pex14p. These data suggest that, although Pex5p directly interacts with ScPex14p through two distinct regions, Pex7p possesses only one binding site in the C terminus of Pex14p.
To determine whether Pex5p and Pex7p bind to the same amino acid sequence within the C terminus of ScPex14p, we shortened the Pex14-(235-341)-p fragment further on both ends and tested these constructs in the yeast two-hybrid assay. The results (Fig. 1C) can be summarized in two points. First, the smallest fragment that interacted with Pex7p was Pex14-(235-325)-p. Second, Pex5p binds to the even smaller regions Pex14-(235-308)-p and Pex14-(250 -341)-p. In conclusion, the amino acid residues in the C terminus of ScPex14p that interact with the two PTS receptors overlap, but they are not identical.
Deletion of the N-terminal 58 Amino Acid Residues of ScPex14p Leads to a Partial Import Defect-The overall import activity of ScPex14p can be assayed by the ability of pex14⌬ cells expressing wild-type Pex14p to grow on oleate as the sole energy and carbon source or by determining the presence and/or properties of peroxisomes using electron microscopy, immunofluorescence, and cell fractionation. All of these techniques were used to assess the functional significance of the identified PTS receptor binding sites within ScPex14p.

. Expression of Pex14p fragments and phenotypical analysis of cells expressing Pex14-(59 -341)-p or Pex14-(1-234)-p by oleate growth test. A and B, equal fractions of yeast wild-type cells, pex14⌬ cells, as well as mutant cells expressing Pex14p full-length and (A)
Pex14-(59 -341)-p (A) or Pex14-(1-234)-p (B) were analyzed by SDS-PAGE and Western blotting. For immunodecoration, anti-Pex14p antibodies or anti-Pex14-(1-58)-p-His 6 antibodies were used, respectively. C and D, cells of indicated yeast strains were grown overnight on glucose minimal media. Subsequently, dilutions were prepared, and 2 l of each dilution was spotted onto oleate plates. The plates were then incubated for 3-5 days at 30°C and scored for the appearance of colonies. In comparison to wild-type cells, the cells expressing Pex14-(59 -341)-p showed reduced growth on oleate as the sole carbon source, and cells expressing Pex14-(1-234)-p were not able to utilize oleate as did pex14⌬ cells.
We first tested pex14⌬ cells expressing Pex14p without the N-terminal 58 amino acid residues under the endogenous promoter of Pex14p. The steady state concentration of this fragment in these cells was similar to that of Pex14p in wild-type cells (Fig. 3A). We observed that pex14⌬ cells expressing the N-terminal deletion mutant Pex14-(59 -341)-p grew significantly less efficiently on oleate than wild-type cells (Fig. 3C) but that they do contain peroxisomes, which according to their morphology and size, are indistinguishable from those of cells expressing wild-type Pex14p (Fig. 4). As the reduced ability to grow on oleate suggested a partial impairment of the protein import, we explored this possibility in more detail. Immunofluorescence microscopy and cell fractionation studies clearly confirmed this assumption for both the PTS1 and surprisingly the PTS2 import pathway. Antibodies against the PTS1 protein Pcs60p and the PTS2 protein thiolase revealed, besides a punctate immunofluorescence pattern, also a distribution of both proteins over the entire range of cells (Fig. 5). Moreover, differential centrifugation experiments showed that catalase and thiolase could only partially be pelleted with the peroxisomes at 25,000 ϫ g, and significant portions were found in the supernatant fractions (Fig. 6). Interestingly, the same result was obtained with Fox1p (Acyl-CoA oxidase). This protein possesses neither a PTS1 nor a PTS2, but as recently shown, its import is nevertheless dependent on Pex5p (20,33).
To establish whether Pex14-(59 -341)-p localizes to peroxisomal membranes and to compare peroxisomes of pex14⌬ cells expressing this N-terminal deletion mutant to those of wild-type cells, we analyzed cell lysates by density gradient centrifugation. Fig. 7, A and B, shows that, in sucrose gradients, cells with the N-terminally truncated Pex14p have a peak of catalase activity at a density of ϳ1.21 g/cm 3 , just as wild-type cells. It reflects the presence of a population of mature peroxisomes. However, the gradient from cells expressing Pex14-(59 -341)-p contained additional catalase activity at the top region of the gradient. Importantly, Pex14-(59 -341)-p co-migrated, just as did wildtype Pex14p, with Pex13p, another exclusively membrane-bound peroxin.
All of these findings together led us to conclude that the N-terminal 58 amino acid residues of Pex14p are necessary for the efficiency of peroxisomal protein import but are not essential.

Deletion of the C-terminal 106 Amino Acid Residues of Pex14p
Results in a pex Phenotype-A similar series of experiments were conducted with pex14⌬ cells expressing the deletion mutant Pex14-(1-234)-p, which lacks the two overlapping Pex5p and Pex7p binding sites of the cytosolically exposed C terminus of Pex14p, in amounts comparable with Pex14p in wild-type cells (Fig. 3B). These cells did not grow on oleate (Fig. 3D) and did not possess morphologically recognizable peroxisomes (Fig. 4), and differential centrif-   ugation demonstrated that the matrix proteins catalase, thiolase, and Fox1p were soluble. As in pex14⌬ cells, these proteins were found in the 25,000 ϫ g supernatant fraction (Fig. 6). Immunofluorescence microscopy corroborated the conclusion of a complete defect of matrix protein import. Antibodies against Pcs60p and thiolase, bona fide PTS1 and PTS2 proteins, respectively, yielded a diffuse immunofluorescence pattern (Fig. 5). Moreover, density gradient centrifugation demonstrated the complete absence of mature peroxisomes at a density of 1.21 g/cm 3 . The Pex14-(1-234)-p fragment co-migrated together with the integral peroxisomal membrane protein Pex13p near the top of the gradient, indicating the presence of per-oxisomal ghosts (Fig. 7D). Pex13p was found in the same gradient fractions as peroxisomal membrane proteins of pex14⌬ cells (Fig. 7,  C and D).
As all of these different approaches revealed the same properties of cells expressing Pex14-(1-234)-p as those of pex14⌬ cells, we conclude that the C-terminal binding sites of Pex5p and Pex7p are essential for matrix protein import into peroxisomes and that their deletion leads to a null mutant phenotype of the corresponding cells.  fide membrane-bound peroxins in sucrose gradients, we wanted to investigate whether the N-and C-terminal deletions have an effect on the topology of Pex14p in the peroxisomal membrane. For this, we subjected 25,000 ϫ g pellets, obtained from wild-type or pex14 deletion cells expressing either one of the two Pex14p fragments, to high salt and carbonate extraction. Fig. 8 shows that all three Pex14p forms were not extracted by high salt treatment. In contrast, although both wild-type Pex14p and Pex14-(1-234)-p are partially resistant to carbonate treatment, Pex14-(59 -341)-p could be completely released. In each case, the integral peroxisomal membrane protein Pex3p was exclusively found in the extracted pellet fractions.
We conclude from these data that, although all Pex14p variants tested are associated with the peroxisomal membrane, only those Pex14p forms with an intact N terminus, i.e. wild-type Pex14p and Pex14-(1-234)-p, partially behave as integral membrane proteins. This suggests that the N-terminal binding site of Pex5p in Pex14p has no effect on targeting per se but significantly contributes to the insertion of Pex14p into the peroxisomal membrane.

DISCUSSION
Accumulated evidence suggests that Pex14p plays a key role in the docking event of the two soluble import receptors at the peroxisomal membrane (1,2). Central to this notion are the observations that Pex14p binds both Pex5p and Pex7p. To gain further insight into this important step, we mapped the binding sites of Pex5p and Pex7p on ScPex14p using the yeast two-hybrid system and pull-down experiments. Surprisingly, we found that Pex5p interacts directly and independently with two different regions of ScPex14p, the N-terminal 58 amino acid residues and ϳ60 amino acids in the C terminus of the protein. The latter region overlaps with the binding site for Pex7p.
The functional assessment of these two different sites of interactions of ScPex14p with the PTS receptors indicates that they have distinct roles. Deletion of the N-terminal 58 amino acids partially reduces matrix protein import; however, it does not lead to a pex phenotype. In contrast, truncation of the C-terminal 106 amino acids of ScPex14p completely blocks matrix protein import. Cells expressing Pex14-(1-234)-p exhibit a phenotype comparable with that of pex14⌬ cells. It is important to note that the steady state concentration of both the N-and the C-terminal mutant constructs of ScPex14p found in these cells is comparable with that of ScPex14p in wild-type cells, and both mutants localize to peroxisomal membranes. On the basis of these data, we propose that the cytosolically exposed C terminus of ScPex14p is the actual docking site for Pex5p and Pex7p.
What is the role of the Pex5p binding site within the N terminus of ScPex14p? Previous work using rat liver peroxisomes has established that the N terminus of Pex14p is protected from exogenously added proteases by the peroxisomal membrane (8,12) and that Pex14p forms a tight complex with a membrane-bound fraction of Pex5p that exhibits the properties of an integral membrane protein (13). More recently, two different Pex14p-associated protease-resistant populations of Pex5p could be resolved (34). These data led Azevedo and co-workers to conclude that Pex14p could be more than just a docking protein for Pex5p, perhaps even forming part of a translocation machinery. In accordance with this view, we would like to propose that the interaction between ScPex14p and Pex5p inside the peroxisomal membrane occurs via the N-terminal binding site in Pex14p. This is supported by the presented data regarding the effect the N terminus has on the topology of Pex14p.
Presently, there are no conclusive experimental data to explain why the deletion of the N-terminal binding site of Pex5p in Pex14p also affects the PTS2 pathway. However, it is conceivable that a slightly changed topology of Pex-(59 -341)-p has an effect on its interaction with other membrane-bound peroxins of the translocation machinery.
It is interesting to speculate that this interaction between the N terminus of Pex14p and Pex5p could help to form or stabilize the observed Pex14p-Pex5p complex inside the peroxisomal membrane. If such a scenario is true, our present finding that the deletion of the N-terminal 58 amino acid residues of ScPex14p only partially impairs (but not totally blocks) matrix protein import would mean that the formation of the Pex14p-Pex5p complex is important for the efficiency of import under physiological conditions but is not absolutely required for the process to occur. In turn, this conclusion then argues that not Pex14p but Pex5p, perhaps with the help of Pex13p, is the essential component of such a putative import channel. This view would be in agreement with the intriguing finding that, in Hansenula polymorpha, the pex phenotype resulting from the absence of Pex14p can be rescued by an overexpression of Pex5p (35). The question whether Pex14-(1-234)-p in contrast to Pex14-(59 -341)-p forms a complex with Pex5p in the peroxisomal membrane of S. cerevisiae will be the subject of future experiments.