A Mobile PTS2 Receptor for Peroxisomal Protein Import in Pichia pastoris

P. pastoris strains used in this study were as follows: PPY12 (arg4, his4; parental wild-type strain); pex7.1 (his4, pex7, ARG4::pTW84 (PTS2-GFP (S65T)/GAPDH; PTS2-BLE/GAPDH)); pex7Δ (his4, pex7::PpARG4); and fox3Δ (his4, fox3::PpARG4). S. cerevisiae strains used in this study were as follows: BJ1991 (MAT α, leu2, trp1, ura3-251, prb1-1122, pep4-3) (wild-type strain); pex7 (pex7, ura3-251, leu2, his3); and L40 (MATa, his3Δ200, trp1-901, leu2-3,112, ade2, LYS2::(lexAop)₄-HIS3, URA3::(lexAop)₈-lacZ (for two-hybrid assays). Escherichia coli strains used in this study were as follows: DH5α (recA, hsdR, supE, endA, gyrA96, thi-1, relA1, lacZ; for cloning procedures), and SG13009 (nal^s, str^s, rif^s, lac⁻, ara⁻, gal⁻, mtl⁻, F⁻, recA⁻; for protein expression).

Minimal methanol contained the following: 0.25% (NH₄)₂SO₄, 0.02% MgSO₄, 0.39% NaH₂PO₄, 0.05% yeast extract, 0.1% (vol/vol) Vishniac minerals mix, 0.1% vitamins, 20 mg/ml amino acids as needed, and 0.5% methanol as carbon source. Minimal oleate medium contained 0.2% oleate and 0.02% Tween 40 as carbon source. Minimal oleate plates were made according to Elgersma et al. (1993). Other media used in this study were described by Wiemer et al. (1996).

A culture of P. pastoris PPY12+pTW84 (PTS2-GFP(S65T)/GAPDH, PTS2-BLE/GAPDH)) cells was grown for 15 h on rich glucose medium (YPD), diluted 12-fold in 600 ml YPD, and grown for an additional 8 h. Cells were pelleted and washed twice with water and once with 100 mM sodium citrate, pH 5.5. The cells were then resuspended in 150 ml 100 mM Na-citrate, pH 5.5, and 25-ml batches were treated with increasing concentrations (0–2%) N-methyl-N-nitro-nitrosoguanidine (NTG) for 30 min. Mutagenized cells were washed three times with 100 mM sodium citrate, pH 5.5, resuspended in 20 ml YPD, and recovered for 2 h. Small aliquots were used to determine the survival rate, and the remainder was frozen in glycerol at −80°C. Cells with an NTG survival of 4% were used to inoculate 50 ml YPD and were grown for 24 h. These cells were pelleted by centrifugation at 2,500 g, washed with water, and transferred to 300 ml rich oleate medium (YPO) and induced for 15 h. These cells were collected, resuspended in 250 ml YPO, aliquoted in five portions of 50 ml, and induced in oleate medium for an additional 2 h. Increasing amounts of phleomycin were added to these cultures (to concentrations of 0–100 μg/ml), and the cells were incubated for 3 h at 30°C. The phleomycin-treated cells were washed twice with water and resuspended in YPD to recover for 2 h at 30°C. Cells were diluted and plated on YPD to determine the survival rate. Colonies obtained from the batch showing a 5% survival rate were tested for growth on minimal oleate and methanol medium, and they were analyzed for PTS2–green fluorescent protein (GFP) import by fluorescence microscopy.

A genomic library was transformed into pex7.1 cells. Three different plasmids (pM1, pM2, and pM3), which restored growth on oleate, were isolated. Subclone analysis revealed that pM1 contained ∼1 kb of the 5′ end of a gene (PpPAT1) that is able to suppress the pex7.1 phenotype. pM2 and pM3 had an overlapping fragment of 2.8 kb. This 2.8-kb insert contained the PpPEX7 gene, which was sequenced on both strands. The PpPEX7 sequence data have been submitted to the GenBank database under accession number AF021797.

A 2.6-kb genomic fragment containing the PEX7 gene was cloned in a three-fragment ligation, as a SalI-EcoRI and EcoRI-BamHI fragment in a SalI/BamHI–digested pUC18, resulting in pM19. The PEX7 gene knockout was made by cloning the PpARG4 gene as a blunt fragment in the blunted NsiI sites (86 bp upstream of the ATG and 28 bp downstream of the stop codon of PEX7) of pM19. The polylinker of pUC18 was modified by ligating adaptors P43 and P44 (see Table I) in the EcoRI and BamHI sites so that these sites were destroyed and the BglII, KpnI, EcoRV, NsiI, and XbaI sites were introduced (pM24). The PEX7 gene was amplified by PCR using primers P33 and P34, thereby introducing a KpnI site immediately upstream of the ATG, and was cloned as a KpnI-NsiI fragment in pM24. To eliminate PCR errors, the BamHI-Nsi fragment was replaced by the BamHI-NsiI fragment of the genomic clone, and the remainder of the PEX7 gene was sequenced. The PEX7 gene on this plasmid (pM27) was used for further manipulations.

For NH tagging, a linker encoding the NH tag (primers P26/P28) was ligated in the EcoRI site of pUC18, resulting in pEL210. PEX7 was cloned as a KpnI-NsiI fragment from pM27 in the KpnI/PstI sites of pEL210, resulting in pM22. The NH-tagged PEX7 was cloned downstream of the ACO1 promoter in the BglII/EcoRI-digested pTW72 (a pHILD2-based plasmid in which the alcohol oxidase promoter was replaced by the ACO1 promoter), using the BglII/HindIII fragment from pM22 and a EcoRI/ HindIII adaptor (P49/P50), resulting in plasmid pM39. The nontagged gene was cloned downstream of the ACO1 promoter by replacing the BglII/AvrII fragment from pM39 (encoding NH-PEX7) by the BglII/XbaI fragment from pM27 (encoding PEX7), resulting in pM70. The ACO1 promoter was replaced by the endogenous PEX7 promoter by replacing the NdeI/BamHI fragment from pM39, containing the ACO1 promoter, with the genomic XhoI/BamHI fragment and the NdeI/XhoI adapter P122/P123, resulting in pM78.

To introduce the C347ter mutation, we ligated linker P124/P125 in an EcoRI/SalI-digested pM27, resulting in pM75, and sequenced the inserted linker. To introduce the G249V and A248R point mutations in PEX7, we designed primers that introduced the desired mutation and a unique restriction site at the location of the mutation. The 5′ part of the gene was amplified by PCR with primers P32 (internal primer just upstream of the BamHI site) and P128 (G249V) or P126 (A248R). The 3′ part of the gene was amplified by PCR with primers P35 (a primer just downstream of PEX7) and P129 (G249V) or P127 (A248R). The PCR fragments corresponding to the 5′ segments of PEX7 were digested with BamHI and PstI (G249V) or BamHI and XbaI (A248R), and the PCR fragments corresponding to the 3′ segments of PEX7 were digested with PstI and EcoRI (G249V) or XbaI and EcoRI (A248R). These fragments were ligated in a three-fragment ligation into the BamHI/EcoRI-digested pM27, resulting in pM76 (G249V) and pM77 (A248R). The BamHI/EcoRI inserts were sequenced. The mutated PEX7 genes were cloned as BamHI/SalI fragments, downstream of the ACO1 promoter in a BamHI/XhoI–digested pM70, resulting in pM85 (C347ter), pM86 (G249V), and pM87 (A248R). The mutated PEX7 genes were cloned as BamHI/SalI fragments downstream of the PEX7 promoter in a BamHI/XhoI-digested pM78, resulting in pM82 (C347ter), pM83 (A248R), and pM84 (G249V).

PpFox3-SKL was amplified by PCR using primers based on the sequence of the PpFOX3 gene so that a BamHI site was introduced directly upstream of the ATG, and an SKL sequence and an XbaI site were added at the 3′ end of the gene. The PCR product was digested with BamHI and XbaI and was cloned downstream of the ACO1 promoter between the BglII/SpeI sites of pM39.

For the yeast two-hybrid analysis, plasmids were generated that expressed appropriate protein fusions to the DNA binding (DB) domain of LexA, and the activation domain (AD) of VP16. The LexA(DB)-ScFox3p (DB-ScFox3p) fusion protein was made by cloning ScFOX3 as a BamHI/ XbaI fragment in pBTM116B, resulting in pM35. pBTM116 (Bartel et al., 1993) was modified by introducing a polylinker in all three open reading frames, resulting in pBTM116A, B, and C. The DB-ScPTS2 fusion protein was made by digesting pM35 with NcoI/SalI, followed by a filling-in reaction using Klenow polymerase and religation, thereby deleting ScFOX3 downstream of the PTS2 (pM52). The DB-ScFox3pΔPTS2 fusion protein was made by digesting pM35 with NotI/NcoI, followed by a filling-in reaction using Klenow polymerase and religation, thereby deleting the PTS2 signal of ScFOX3 (pM51). The AD-PpPex7p fusion protein was made by cloning PpPEX7 as a BglII/SalI fragment from pM27 in the BamHI/SalI-digested pVP16C, resulting in pM34. pVP16 (Hollenberg et al., 1995) was modified by introducing a polylinker in all three open reading frames, resulting in pVP16A, B, and C. The RCDP mutations were introduced in the two-hybrid vector as BamHI/SalI fragments into a BamHI/SalI-digested pM34, resulting in pM79 (Pex7p(G249V)), pM80 (Pex7p(A248R)), and pM81 (Pex7p(C347ter)).

To raise antibodies against Pex7p, we cloned the DNA encoding amino acids 1–143 of Pex7p as a KpnI/HpaI fragment from pM27 into pQE30 (QIAGEN Inc., Chatsworth, CA), which was first digested with HindIII, blunted with Klenow polymerase, and subsequently digested with KpnI. The (His)₆-tagged protein was expressed in E. coli SG13009 and purified under denaturing conditions on Ni²⁺-NTA beads according to the manufacturer's manual (QIAGEN Inc.). The protein was further purified by SDS-PAGE, visualized with 0.25 M KCl, 1 mM DTT, and subsequently excised and eluted from the gel in elution buffer (50 mM Tris, pH 8.0, 0.1% SDS, 0.1 mM EDTA, 5 mM DTT, 0.15 M NaCl). This purified protein was used to immunize rabbits.

The antibody was purified using a pex7Δ depletion column and, subsequently, a Pex7p affinity column. The depletion column was made by coupling overnight ∼20 mg of proteins from a cell-free pex7Δ lysate to 0.6 g of cyanogen bromide–activated Sepharose 4B (Sigma Immunochemicals, St. Louis, MO) in coupling buffer (0.1 M NaHCO₃, 0.5 M NaCl). After blocking the column for 2 h with blocking buffer (1 M ethanolamine, 0.1 M NaHCO₃, 0.5 M NaCl, pH 8.0), the column was washed extensively, alternating between coupling buffer and low pH wash buffer (20 mM sodium acetate, 0.5% [vol/vol] acetic acid, 0.5 M NaCl, pH 3.5) and subsequently with PBS, followed by 0.2 M glycine, 1 mM EGTA, pH 2.4, followed by PBS. After overnight incubation with 10 ml of serum, the serum was drained off and collected together with the first 3 ml of PBS wash. This purified serum was used for further purification on an affinity column. This was done essentially in the same manner as the depletion column, except that 2 mg of purified Pex7p was bound to the column, and after binding the serum to the column, it was washed extensively with PBS, low pH wash buffer (0.3% [vol/vol] acetic acid, 50 mM sodium acetate, 1 M NaCl, pH 4.5), high pH wash (0.1 M NaHCO₃, 1 M NaCl, pH 9.0), and again with PBS. The antibodies were eluted with 15 ml of 0.2 M glycine, 1 mM EGTA, pH 2.4, and immediately neutralized to pH 7.0 with 1 M Tris base. Finally, the antibodies were concentrated to ∼1 ml using a Centriprep 30 (Amicon, Beverly, MA). The antibody was used at a 1:10,000 dilution for Western blotting.

Subcellular fractionations of oleate-grown cells were essentially performed as described by Van der Leij et al. (1992), in the presence of protease inhibitors (1 mM PMSF, 0.2 mg/ml NaF, and 2 μg/ml of chymostatin, leupeptin, antipain, and pepstatin). 6 ml of a 1,000 g postnuclear supernatant was layered on a continuous 16–35% Nycodenz gradient (30 ml), with a 4-ml cushion of 42% Nycodenz dissolved in 5 mM MES, pH 6.0, 1 mM EDTA, 1 mM KCl, and 8.5% sucrose. The sealed tubes were centrifuged for 2.5 h in a vertical rotor at 29,000 g at 4°C. Afterwards, fractions were collected and analyzed by SDS-PAGE and Western blotting.

Fluorescence microscopy for GFP analysis was done as described by Monosov et al. (1996). Fluorescence microscopy of semithin sections and immuno-EM analysis of ultrathin sections was done as follows: P. pastoris cells were grown to log phase and fixed in either periodate-lysine-paraformaldehyde (75 mM phosphate buffer containing 2% formaldehyde, 70 mM lysine, and 10 mM sodium periodate, pH 6.2) for 6 h, or in 4% paraformaldehyde contained in 100 mM phosphate buffer, pH 7.4, for 18–24 h at 4°C. The cells were washed, pelleted, and embedded in 1% ultralow temperature gelling agarose, and were trimmed into 1-mm³ blocks. The cells were cryoprotected in 2.3 M sucrose containing 20% polyvinyl pyrollidone for 1 h, mounted on aluminum cryopins, and frozen in liquid nitrogen. Ultrathin cryosections were then cut on a Reichert Ultracut E microtome (Leica Inc., Heerbrugg, Switzerland) equipped with an FC4 cryostage, and sections were collected onto 300 mesh, Formvar/carbon-coated nickel grids and immunolabeled as follows. Grids were washed with several drops of PBS containing 5% FCS and 0.01 M glycine, pH 7.4. The grids were blocked in 10% FCS and incubated for 16 h with αNH antibodies in PBS containing ∼10 μg/ml specific antibody. After washing the grids in PBS, they were incubated for 2 h with a 50× diluted goat anti–rabbit 5-nm gold conjugate (Amersham Life Sciences, Arlington Heights, IL). The grids were washed several times with PBS, subsequently with water, and then the grids were embedded in a mixture containing 2.7% polyvinyl alcohol (mol wt = 10,000), 0.2% methyl cellulose (400 cP), and 0.2% uranyl acetate. The sections were observed and photographed on a 1200EX II TEM microscope (Jeol Ltd., Tokyo, Japan).

Digitonin titrations were essentially performed as described by Zhang and Lazarow (1995), with modifications as described in Elgersma et al. (1996a).

For TCA lysates, cells (50 ml of culture OD₆₀₀ = 0.7) were pelleted by centrifugation, washed, and collected in a 2-ml Eppendorf tube. After the addition of 500 μl of 10% TCA and ∼200 μl of glass beads, the cells were broken by vigorous vortexing for 30 min at 4°C. Cell debris was pelleted by centrifugation for 30 s at 6,000 rpm, and the supernatant was further centrifuged for 20 min at 12,000 rpm at 4°C in a microcentrifuge. The pellet was washed once with acetone, dried, and resuspended in Laemmli sample buffer for SDS-PAGE.

The negative screening procedure for isolating P. pastoris pex mutants that has been used thus far (Gould et al., 1992; Liu et al., 1992) has not resulted in the isolation of mutants that are specifically disturbed in the PTS2 import pathway. Therefore, we developed a positive selection procedure that would enrich for such mutants. In S. cerevisiae, such a scheme has been developed, which is based on the ability of peroxisomes to import the bleomycin resistance protein (BLE) fused to the PTS1 (Elgersma et al., 1993). In wild-type cells, BLE-PTS1 is imported into peroxisomes, thereby sequestering the BLE protein from its toxic ligand, phleomycin. Peroxisomal import mutants (pex) are unable to import BLE-PTS1 into their peroxisomes, which allows the protein to bind phleomycin, thereby preventing its toxicity. We modified this procedure by fusing the BLE protein to the PTS2. Since a PTS2 sequence of an endogenous P. pastoris protein had not been identified, we tested the PTS2 sequence of S. cerevisiae thiolase (Fox3p) in P. pastoris. We therefore made a fusion of the first 17 amino acids of thiolase (containing the PTS2 signal) to GFP (PTS2-GFP). When expressed in oleate-grown P. pastoris, we observed a punctate GFP expression pattern, suggesting that PTS2-GFP was imported into the peroxisomes and that the ScPTS2 is functional in P. pastoris (Fig. 1,E). We also observed some cytosolic labeling, however, suggesting that PTS2-GFP import was not as efficient as GFP-SKL import (Fig. 1, E versus C).

A yeast strain (PPY12 + pTW84) that expressed PTS2-GFP was made, and the BLE fused to the PTS2 sequence (PTS2-BLE) under the control of the glyceraldehyde 3-phosphate dehydrogenase promoter (GAPDH) from a plasmid integrated in the genome. These cells were mutagenized, induced on oleate, and treated with phleomycin. We expected that PTS2 import mutants would be able to grow on methanol because none of the known enzymes involved in methanol metabolism contain a PTS2 import signal. Conversely, because thiolase is required for growth on oleate, we expected that a PTS2 import–deficient mutant would be unable to grow on oleate. Therefore, colonies that had survived the phleomycin selection were screened for their ability to grow on methanol but not on oleate. Two mutants (fox3.1 and pex7.1) clearly fulfilled this criterion (shown for pex7Δ in a growth curve in Fig. 2, A and B). Inspection of PTS2-GFP import revealed, however, that one of these mutants (fox3.1) was unaffected in PTS2 protein import. Cloning of the complementing gene revealed that this mutant was deficient in the thiolase gene itself (Koller, A., and S. Subramani, unpublished results).

The second mutant (pex7.1) was not only specifically disturbed in its growth on oleate (Fig. 2, A and B), but was also unable to import PTS2-GFP (Fig. 1,F), whereas import of GFP-SKL was normal (Fig. 1,D). This suggested a general deficiency of pex7.1 cells in the import of PTS2-containing proteins. The pex7.1 mutant was therefore further characterized by biochemical subcellular fractionation. In wild-type cells, we observed an equal distribution of thiolase between the organellar pellet fraction and the cytosolic supernatant fraction. As expected, thiolase was exclusively present (27,000 g) in the cytosolic supernatant fraction of pex7.1, providing additional evidence that the import of PTS2-containing proteins was disturbed (Fig. 2 C). Targeting of the membrane protein Pex3p was normal, as was the import of acyl CoA oxidase (Aco1p), because these proteins were predominantly present (27,000 g) in the organellar pellet fraction. Furthermore, the import of presumed PTS1-containing proteins such as catalase (Cta1p) and trifunctional enzyme (Tfe1p) was unaffected, suggesting that the import deficiency of pex7.1 is specific for PTS2-containing proteins only.

The pex7.1 mutant was complemented for growth on oleate medium using a genomic library. Two plasmids (pM1 and pM2) with nonoverlapping inserts restored growth on oleate, although growth was better when using plasmid pM2 (Fig. 2 D). Further analysis of plasmid pM1 revealed that complementation of the oleate-minus phenotype was caused by a truncated protein (amino acids 1–342) with highest homology to ScPat1p, a peroxisomal ABC transporter (Hettema et al., 1996). Interestingly, it has been reported that the peroxisomal ABC transporter Pmp70 can suppress the peroxisomal protein import deficiency of a mammalian pex2-deficient cell line (Gartner et al., 1994). Therefore, we tested whether plasmid pM1 was able to restore the PTS2-GFP import in the pex7.1 mutant. Microscopy analysis did not reveal any restoration of import of PTS2-GFP in these cells, suggesting that this protein did not act as a suppressor of the PTS2 import deficiency (data not shown). It has been recently demonstrated that ScPat1p and ScPat2p are involved in the transport of (activated) long-chain fatty acids across the peroxisomal membrane (Hettema et al., 1996). We therefore speculate that the truncated PpPat1p somehow disturbed the normal functioning of the Pat1p/Pat2p complex, thereby allowing leakage of the partially completed β oxidation product (3-oxoacyl-CoA) into the cytosol, allowing cytosolic thiolase to complete the β oxidation in this pex7.1 mutant.

The second clone (pM2) was not only able to restore growth on oleate of the pex7.1 mutant, but it also corrected the import deficiency of PTS2-GFP (data not shown). Further analysis revealed that the complementing gene encoded a protein of 376 amino acids with a calculated molecular mass of 42.4 kD. The protein contains six WD-40 repeats and has 43% sequence identity with ScPex7p (Fig. 3). Evidence that we cloned a true orthologue of ScPex7p was obtained by expressing the gene in S. cerevisiae. It was able to complement the growth defect of S. cerevisiae pex7 (Fig. 4 A); hence, the gene is designated PpPEX7.

Pex7p from S. cerevisiae has been shown to interact with the PTS2 sequence in multiple ways (Rehling et al., 1996; Zhang and Lazarow, 1996). Because PpPex7p complements the Scpex7 mutant, we expected PpPex7p to interact with the PTS2 sequence. This was tested using the yeast two-hybrid system (Fields and Song, 1989). Therefore, we fused Pex7p to the LexA activation domain (AD-PpPex7p) and Sc-thiolase to the LexA DNA–binding domain (DB-ScFox3p). These plasmids were coexpressed in S. cerevisiae L40 cells with either an empty control plasmid or with each other. High induction of the HIS3 gene and β-galactosidase was obtained when AD-PpPex7p and DB-ScFox3p were coexpressed, whereas no induction was observed when AD-PpPex7p or DB-ScFox3p was coexpressed with a control plasmid (shown for HIS3 expression in Fig. 4,B). This indicates that indeed Pex7p interacts with Fox3p in vivo. To test whether this interaction was dependent upon the presence of the PTS2, we made a construct from which the 17 NH₂-terminal amino acids of thiolase were removed (DB-ScFox3pΔPTS2). No induction of the HIS3 gene and β-galactosidase was observed when DB-ScFox3pΔPTS2 was coexpressed with AD-PpPex7p, indicating that the interaction of PpPex7p with thiolase was PTS2 dependent. To test whether the PTS2 alone would be sufficient to bind to PpPex7p, we fused the first 17 amino acids of ScFox3p to the LexA DNA–binding domain (DB-ScPTS2). When DB-ScPTS2 was coexpressed with AD-PpPex7p, we found again a strong induction of the HIS3 gene and β-galactosidase (shown for HIS3 expression in Fig. 4 B). These results indicate that the ScPTS2 sequence alone is sufficient for interaction with PpPex7p in vivo, supporting a receptor-like function for this protein.

To study the function of Pex7p, we made a PEX7 deletion mutant (pex7Δ) in which the entire open reading frame was replaced by the PpARG4 gene. Subcellular fractionation of this mutant showed the same phenotype as was observed for the pex7.1 mutant: a specific import deficiency of the PTS2-containing enzyme thiolase (shown later in Figs. 5,B and 6, B and D).

Like the original pex7.1 mutant, pex7Δ did grow on methanol medium, suggesting that peroxisomal import of PTS2-containing enzymes is not necessary for growth on methanol (Fig. 2,A). Conversely, pex7Δ did not grow on oleate, indicating that import of at least one PTS2-containing protein is required for growth on this medium (Fig. 2,B). We investigated whether the growth defect on oleate was caused solely by the mistargeting of thiolase. Therefore, a construct (Fox3p-SKL) expressing P. pastoris Fox3p (Koller, A., S. Subramani, unpublished results) fused to the PTS1 sequence (SKL) under the control of the ACO1 promoter was integrated in the genome of pex7Δ and fox3Δ cells. This protein restored growth on oleate plates to nearly wild-type levels when expressed in Ppfox3Δ, indicating that Fox3p-SKL is active and imported into peroxisomes (Fig. 5,A). Expression of Fox3p-SKL in pex7Δ only partially restored the growth on oleate plates. This suggests that the inability to import thiolase is not the only reason why pex7Δ cells do not grow on oleate, but that other PTS2-containing proteins are also required for growth on this medium (Fig. 5 A).

The import efficiency of Fox3p-SKL was studied by subcellular fractionation (Fig. 5,B). We observed a considerable amount of endogenous thiolase in the cytosolic fraction in the control (untransformed wild-type) cells, whereas thiolase was exclusively cytosolic in untransformed pex7Δ cells. The control enzyme, Aco1p, was predominantly present in the organellar pellet fractions (Fig. 5,B). These results suggest that thiolase easily leaks out of peroxisomes during subcellular fractionation. Alternatively, the PTS2 import pathway in P. pastoris is not very efficient. This is supported by the observation that PTS2-GFP is also largely cytosolic and only partially peroxisomal (Fig. 1, E and G), although we cannot rule out that this localization is caused by the chimeric protein itself. Interestingly, Fox3p-SKL was fully imported into peroxisomes of fox3Δ cells, with no protein detectable in the cytosolic fraction. In pex7Δ cells expressing Fox3p-SKL, thiolase was equally distributed between the pellet and supernatant fractions (Fig. 5,B). From the small difference in molecular weight between Fox3p-SKL and endogenous Fox3p, we conclude that Fox3p-SKL is fully imported into these cells via its PTS1 sequence, whereas endogenous thiolase is almost exclusively cytosolic, as in the untransformed pex7Δ control cells (Fig. 5 B). These results suggest that the PTS2 import pathway in P. pastoris may be less efficient than the PTS1 import pathway (see Discussion).

The lack of useful antibodies against endogenous Pex7p has greatly hampered the localization analysis of this protein in S. cerevisiae and in human cells. We therefore wanted to raise antibodies against PpPex7p to analyze its localization. We expressed the full-length Pex7p in E. coli fused to a (His)₆ tag, but the expression of this protein was very poor. We were unable to increase its expression by fusing it to dihydrofolate reductase, nor could we increase the expression by fusing three different truncated versions of Pex7p to dihydrofolate reductase. Good expression in E. coli was obtained for a truncated protein corresponding to amino acids 1–143 fused to the (His)₆ epitope. This purified protein, (His)₆-Pex7p(143), was used to immunize rabbits. The antiserum that was obtained recognized several proteins in the wild-type strain and in the pex7Δ strains (Fig. 6,A, lanes 1 and 2). Because this could indicate that a (nonfunctional) pseudogene of PEX7 might be present, we checked this possibility with PCR and low stringency Southern blotting, but did not find any evidence for the existence of such a gene (data not shown). We purified the antiserum with an immunodepletion column, using a lysate obtained from pex7Δ cells. The flow-through of this column was used to purify the antiserum on an affinity column using the coupled(His)₆-Pex7p(143) protein. The eluted antibody recognized two prominent bands in lysates of wild-type cells (Fig. 6, A, lane 4). The signal obtained from the lower molecular weight (cytosolic) protein was present at the same intensity in pex7Δ cells (Fig. 6, lane 3) and could be eliminated by using more stringent washing conditions (see figures below). The higher molecular weight band was largely reduced in pex7Δ cells, indicating that this signal is from Pex7p (Fig. 6,A, lane 3 versus lane 4). This could be confirmed by overexpressing Pex7p from the acyl-CoA oxidase promoter (Fig. 6, lane 5). A 25-fold dilution of this lysate (compared to a wild-type and pex7Δ lysate) gave only one signal at the expected molecular weight (Fig. 6 A, lane 6). Because there is still a weak signal present at this molecular weight in a pex7Δ lysate (from a mitochondrial protein, see below), in our subsequent analyses, we included the pex7Δ control and separated the peroxisomes from mitochondria where necessary.

The intracellular location of Pex7p was determined by subcellular fractionation. Wild-type and pex7Δ cells were fractionated, and the organellar pellet and the cytosolic supernatant were analyzed. Aco1p, Pex3p, and thiolase (Fox3p) were present in the organellar (27,000 g) pellet fraction of wild-type cells, although varying portions of some of these proteins were also present in the cytosolic (27,000 g) supernatant fraction, possibly because of leakage (Fig. 6,B). A similar distribution was observed for pex7Δ cells, with the exception of thiolase, which was exclusively present in the cytosolic fraction, reflecting the PTS2 import deficiency of pex7Δ (Fig. 6,B). Pex7p was present in both the organellar pellet (varying from 10 to 30%) and the cytosol fraction of wild-type cells, suggesting that about a quarter of the protein is associated with peroxisomes. In Fig. 6,B, the mitochondrial protein that sometimes cross-reacts with the anti-Pex7p antibodies (see Fig. 6, A and D) would not affect our conclusion that Pex7p is partially found in the cytosol because (a) we did not observe the cross-reacting band in this particular experiment (see H and P fractions of control pex7Δ cells), and (b) the cross-reacting mitochondrial band was never found in the supernatant fraction during subcellular fractionation. The protein band found in the cytosolic fraction can therefore be only Pex7p.

To determine whether the pelletable Pex7p was indeed associated with peroxisomes, we used a postnuclear supernatant for further analysis on a Nycodenz gradient. A portion of Pex7p comigrated with the peroxisomal marker enzymes Aco1p and Fox3p, suggesting that at least some Pex7p is associated with peroxisomes (Fig. 6,C). We also found significant amounts of αPex7p–cross-reacting material at the top of the gradient (fractions 14–20), representing mitochondrial and cytosolic proteins. In these gradients, the mitochondrial protein (Fig. 6,D, fractions 14–18) that cross-reacted with the antibody against Pex7p was never found at the top of the gradient in fractions 19 and 20, where cytosolic proteins would be found. The presence of Pex7p in fractions 19 and 20 of the gradient from wild-type cells (Fig. 6,D) shows that some Pex7p is indeed cytosolic. Because of the cross-reacting protein present in the mitochondrial fractions (see fractions 14–18 in the pex7Δ control gradient, Fig. 6 D), however, it is impossible to estimate the exact amount of Pex7p present in the cytosol in this experiment.

To study the localization of Pex7p by an independent biochemical method, we selectively permeabilized spheroplasts of wild-type cells, pex7Δ and pex3Δ (a mutant disturbed in the import of PTS1- and PTS2-containing proteins; Wiemer et al., 1996) by digitonin, and we followed the subsequent leakage of Pex7p and control proteins from the cells. In this procedure, cytosolic enzymes are released at low concentrations of digitonin, whereas peroxisomal matrix enzymes are released at much higher concentrations of digitonin (Zhang and Lazarow, 1995). As can be seen in Fig. 7, the cytosolic enzyme glucose-6 phosphate dehydrogenase (G6PDH) was completely released at low concentrations (50 μg/ml) of digitonin. This was also observed for catalase in pex3Δ, as well as thiolase in the pex7Δ and pex3Δ strain, which reflects the peroxisomal import deficiency of these mutants. In contrast, thiolase and catalase were released at significantly higher concentrations of digitonin in wild-type cells, starting at 150 μg/ ml digitonin, and release was not complete until 300–400 μg/ml digitonin was added. The membrane protein Pex3p could be completely released at only very high concentrations of digitonin (>600 μg/ml; Fig. 7). Using the αPex7p antibody, we observed release of the mitochondrial αPex7p–cross-reacting protein in pex7Δ cells at digitonin concentrations of ⩾400 μg/ml. In wild-type cells, a substantial amount of Pex7p was already released before release of peroxisomal matrix proteins took place (before 150 μg/ml digitonin). In contrast to the cytosolic markers, however, release was not complete until 300–400 μg/ml digitonin, which is similar to what was observed for the release of Cta1p and Fox3p. In pex3Δ cells, Pex7p was completely released at 50 μg/ml. These results confirm that Pex7p is distributed between the peroxisome and the cytosol. Furthermore, since complete release of Pex7p is observed at digitonin concentrations <400 μg/ml, these results also suggest that some Pex7p is present in the lumen of peroxisomes.

If Pex7p functions solely as an intraperoxisomal receptor, it probably has a PTS that directs it efficiently to the peroxisomal matrix. Some experiments in S. cerevisiae suggest that this is indeed the case (Zhang and Lazarow, 1996). To test whether PpPex7p can direct a reporter protein to the peroxisome, we fused PpPex7p to the peroxisomal malate dehydrogenase of S. cerevisiae, which lacks its PTS1 (NH-Mdh3pΔPTS1). It has previously been shown that NH-Mdh3pΔPTS1 is cytosolic and that it can not complement the growth defect on oleate of S. cerevisiae mdh3Δ cells (Elgersma et al., 1996b). When PpPex7p was fused to NH-Mdh3pΔPTS1 (NH-Mdh3p-PpPex7p) and expressed in S. cerevisiae mdh3Δ cells under the control of the ScCTA1 promoter, we observed some complementation of the mdh3Δ mutant (data not shown). This suggests that PpPex7p does indeed have a signal that can direct it to the peroxisomal matrix.

To study the effect of overexpressing and/or epitope-tagging Pex7p, the expression level of Pex7p was increased by expressing it under the control of the ACO1 promoter. Western blot analysis revealed that this resulted in an ∼25-fold higher expression level (see Fig. 6,A). The overexpressed protein fully complemented the growth deficiency of pex7Δ cells (Fig. 8,A). Full complementation was also observed when we epitope tagged Pex7p by fusing it to the NH tag (Elgersma et al., 1996b) and expressed it under the control of the ACO1 promoter (NH-Pex7p/ACO1) (Fig. 8,A). Neither tagging nor overexpression affected its ability to restore the peroxisomal import of thiolase in pex7Δ cells, as shown by subcellular fractionation (Fig. 8,B). Interestingly, we found on average >90% of overexpressed Pex7p in the cytosol (Fig. 8,B). Comparison of the absolute amount present in the organellar pellet revealed that the amount of pelletable Pex7p was only increased 5–10-fold (Fig. 8 C) compared to untransformed cells. These experiments suggest that targeting of PpPex7p to the peroxisomal matrix is not as efficient as the PTS1 or PTS2 import pathway, and that an unknown component is limiting the import of Pex7p.

The localization of overexpressed NH-Pex7p was also addressed by fluorescence microscopy of semithin sections of yeast cells labeled with the NH antibody. This antibody and epitope tag have previously been shown to be very suitable for immuno-EM analysis (Elgersma et al., 1996b). Analysis of control cells expressing NH-tagged malate dehydrogenase of S. cerevisiae (NH-Mdh3p) under the control of the ACO1 promoter showed a punctate labeling, indicating efficient import of NH-Mdh3p into peroxisomes (Fig. 9, A and B). In contrast, cells expressing NH-Pex7p showed only limited punctate staining and a high cytosolic staining (Fig. 9, C and D). This supports the biochemical data that a large amount of overexpressed Pex7p is cytosolic. To verify the nature of the punctate staining, we used the cells expressing NH-Mdh3p or NH-Pex7p for immuno-EM analysis. This revealed that a portion of NH-Pex7p is indeed present in the peroxisomal matrix (Fig. 10, B–D). These results confirm the predominantly cytosolic and partially intraperoxisomal location of overexpressed NH-Pex7p.

The human PEX7 gene has recently been cloned and shown to be responsible for RCDP (Braverman et al., 1997; Motley et al., 1997; Purdue et al., 1997). The most frequently occurring mutations were the L292ter and A218V mutations. Five patients were genetic compounds of the L292ter as the first allele and a G217R mutation as the second allele (Braverman et al., 1997). Expression studies showed that the L292ter and A218V mutations impaired Pex7p function, but the functional consequence of the G217R mutation was undetermined. To test whether the mutated proteins were nonfunctional because of instability, mistargeting, or failure to bind the PTS2 sequence, we made the PpPEX7 equivalents of the RCDP G217R, A218V, and L292ter mutations (A248R, G249V, and C347ter, respectively, as indicated in Fig. 3). The mutated PpPex7p proteins were expressed under the control of the PEX7 and ACO1 promoters, and they were tested for the ability to correct the growth deficiency of pex7Δ on oleate. As shown in Fig. 11 A, all mutated proteins were unable to restore the growth defect on oleate. When (over)expressed under the control of the ACO1 promoter, we observed some oleate usage by Pex7p(G249V), as judged by the formation of a halo on an oleate plate (data not shown). This correlates with the finding that patients with this corresponding allele have a phenotype milder than in classical RCDP (Braverman et al., 1997).

We made lysates of pex7Δ cells expressing the mutated Pex7p. As shown in Fig. 11,B, all proteins were expressed stably, suggesting that the deficiency is not caused by protein instability. We then tested whether the mutated proteins were able to bind the PTS2 sequence, using the two-hybrid system as described above. Interestingly, all mutated proteins had lost the ability to bind the PTS2 sequence (Fig. 11 C). This suggests that the mutations lead to impaired PTS2 binding.

To test whether the mutations affected the peroxisomal targeting of Pex7p, we expressed the mutated proteins in pex7Δ cells and compared their targeting to that observed for wild-type Pex7p by Nycodenz gradient analysis. We did not observe significant differences between the distribution of wild-type or mutated proteins (data not shown). This indicates that peroxisomal targeting of the mutated Pex7p proteins is not affected. Moreover, this suggests that binding of the PTS2 sequence is not a prerequisite for targeting Pex7p to the peroxisome.

Using a new screening procedure devised to yield mutants specifically compromised in the PTS2 import pathway, we have obtained a P. pastoris pex7 mutant, cloned the PpPEX7 gene, and shown in several ways that PpPex7p is the homologue of ScPex7p, the PTS2 receptor for protein import (Rehling et al., 1996, Zhang and Lazarow, 1996): (a) PpPex7p has high sequence identity with ScPex7 (43%), (b) PpPEX7 complements the ScPex7 mutant, (c) Pppex7Δ is specifically disturbed in PTS2 protein import, and (d) PpPex7p interacts with the PTS2 sequence in the two-hybrid system. The observation that Pppex7Δ is unaffected in its growth on methanol medium suggests that there are no PTS2-containing enzymes or peroxins required for growth on this medium. Furthermore, it explains why screening procedures for pex mutants, based on their inability to grow on methanol, did not result in the isolation of this mutant.

The role of Pex7p in PTS2 import has been a matter of debate. Zhang and Lazarow (1995, 1996) concluded that ScPex7p is an intraperoxisomal receptor and functions by pulling PTS2 proteins inside, analogous to the role of mitochondrial HSP70. Although they found significant amounts of a triple-tagged Pex7p (Pex7p-HA3) in the cytosol upon fractionation, they observed in digitonin experiments a strikingly similar pattern of latency of Pex7p-HA3 compared to peroxisomal matrix proteins. Therefore, they concluded that the cytosolic pool found in fractionation studies was probably caused by leakage of Pex7p-HA3. In contrast, others proposed that Pex7p functions as a mobile cytosolic receptor because most of the (overexpressed and tagged) Pex7p protein was found in the cytosol by biochemical fractionation (Marzioch et al., 1994; Rehling et al., 1996) or by immunofluorescence (Braverman et al., 1997).

We raised antibodies against the endogenous Pex7p and studied its localization. Subcellular fractionation and digitonin experiments suggest that Pex7p is largely cytosolic but that some of the protein is peroxisomal. Evidence that the peroxisome-associated protein is present in the matrix was deduced from digitonin experiments and immuno-EM using NH-tagged Pex7p, which showed clear peroxisomal matrix labeling and no obvious membrane labeling. Interestingly, a 25-fold overexpression of Pex7p results in a modest 5–10-fold increase in peroxisome-associated Pex7p and in a large increase of cytosolic Pex7p. This explains why others found an almost exclusively cytosolic localization for Pex7p when they studied the localization of overexpressed Pex7p (Marzioch et al., 1994; Rehling et al., 1996; Braverman et al., 1997).

Our analyses indicate that most of Pex7p is cytosolic (varying from 70 to 90%), suggesting a role in the cytosol where it binds to PTS2-containing proteins and delivers them at the peroxisomal membrane. The identification of Pex14p as a putative peroxisomal docking protein for Pex7p supports this model (Albertini et al., 1997). A similar model has also been proposed for the PTS1 receptor Pex5p, which binds to the peroxisomal membrane protein Pex13p (Elgersma et al., 1996a; Erdmann and Blobel, 1996; Gould et al., 1996).

In addition to the cytosolic pool of Pex7p, we found also a significant fraction of Pex7p in the peroxisomal matrix (varying from 10 to 30%). Therefore, it is possible that Pex7p is also involved in the subsequent translocation step. In that case, we expect that Pex7p has to get recycled to the cytosol because thiolase expression is at least 25-fold higher than Pex7p expression. The observation that the amount of intraperoxisomal Pex7p did not increase proportionally upon overexpression suggests that the Pex7p import signal is not very efficient. Therefore, it is conceivable that intraperoxisomal Pex7p does not have a role at all, but that some Pex7p enters the peroxisome coincidentally and remains in the peroxisomal matrix without any function. Further kinetic experiments are required to address this point.

The ability of Pex7p to enter the peroxisomes, albeit at low efficiency, appears to be real. Some amounts of the wild-type and overexpressed Pex7p were intraperoxisomal, as judged by digitonin permeabilization and immuno-EM (Figs. 7 and 10). Furthermore, NH-Mdh3p-PpPex7p was able to complement an mdh3Δ strain of S. cerevisiae (data not shown). The mechanism involved in this targeting is unknown. Because protein unfolding is not essential for the import of proteins into the peroxisomal matrix, Pex7p might enter the matrix solely by association with PTS2-containing proteins. This is unlikely, however, because the A248R, G249V, and C347ter mutants cannot interact with the PTS2 sequence and show the same subcellular location as wild-type Pex7p. Moreover, within experimental error, we did not observe a different distribution of Pex7p when we deleted the FOX3 gene or when we raised its expression twofold by overexpressing it from the ACO1 promoter. An alternative possibility is that Pex7p has a new targeting signal or that it enters the peroxisome in association with some other protein. Such a targeting or protein association signal might reside in the first 56 amino acids of ScPex7p, as suggested by the experiments of Zhang and Lazarow (1996).

The cloning of yeast PEX genes has tremendously facilitated the characterization of their human orthologues (for review see Subramani, 1997). For instance, the human PEX7 gene was cloned by screening databases using the yeast PEX7 gene (Braverman et al., 1997; Motley et al., 1997; Purdue et al., 1997). Our studies show a new role for the yeast model system: the rapid analysis of the effect of the mutations found in the PEX genes of patients. We introduced the mutations found in the RCDP patients at the corresponding positions in the PpPEX7 gene and found that this resulted in impaired growth on oleate. Furthermore, we found that the mutated Pex7p proteins in P. pastoris are stably expressed and not disturbed in peroxisomal targeting, but that they are unable to bind to the PTS2 sequence. The application of a similar strategy to other human PEX genes involved in disease is likely to provide useful insights regarding the effect of mutations.

Pppex7Δ did not grow on oleate plates, but this deficiency could be partially suppressed by expressing Fox3p-SKL in these cells. This indicates that the lack of thiolase import is possibly the main, but certainly not the only, reason that pex7Δ cells do not grow on oleate. Hence, we expect that other PTS2-containing proteins are also required for oleate metabolism. Surprisingly, it appeared that Fox3p-SKL was either imported or retained in peroxisomes much more efficiently than endogenous thiolase. Moreover, it should be noted that we did not observe significant import of endogenous thiolase in pex7Δ cells expressing Fox3p-SKL. This indicates that the formation and import of heterodimers of Fox3p-SKL with endogenous thiolase did not take place, in contrast to what has been observed for S. cerevisiae cells expressing thiolase with and without a PTS2 (Glover et al., 1994b). This may suggest that the import of Fox3p-SKL via the PTS1 pathway is much faster than the import of endogenous thiolase via the PTS2 pathway, and that import of Fox3p-SKL takes place before the folding and oligomerization of Fox3p-SKL is completed. Further kinetic experiments, as conducted by Ruigrok et al. (1996), are necessary to validate this hypothesis.