Pex20p of the Yeast Yarrowia lipolytica Is Required for the Oligomerization of Thiolase in the Cytosol and for Its Targeting to the Peroxisome

The Y. lipolytica strains used in this study are listed in Table I. The new nomenclature for peroxisome assembly genes and proteins has been used (Distel et al., 1996). Media, growth conditions, and genetic techniques for Y. lipolytica have been described (Nuttley et al., 1993; Szilard et al., 1995). Media components were as follows: (a) YEPD, 1% yeast extract, 2% peptone, 2% glucose; (b) YPBO, 0.3% yeast extract, 0.5% peptone, 0.5% K₂HPO₄, 0.5% KH₂PO₄, 1% Brij 35, 1% (wt/vol) oleic acid; (c) YNAS, 0.67% yeast nitrogen base without amino acids, 0.1% yeast extract, 18.2% sorbitol, 2% sodium acetate; (d) YND, 0.67% yeast nitrogen base without amino acids, 2% glucose; and (e) YNO, 0.67% yeast nitrogen base without amino acids, 0.05% (wt/vol) Tween 40, 0.1% (wt/vol) oleic acid. YNAS, YND, and YNO media were supplemented with leucine, uracil, lysine, and histidine, each at 50 μg/ml, as required. DNA manipulation and growth of Escherichia coli were performed as previously described (Ausubel et al., 1989).

The pex20-1 mutant was initially isolated from randomly mutagenized Y. lipolytica strain E122 by screening for the inability to use oleic acid as a sole carbon source (Nuttley et al., 1993). The PEX20 gene was isolated by functional complementation of the pex20-1 strain using a Y. lipolytica genomic DNA library in the autonomously replicating E. coli shuttle vector pINA445 (Nuttley et al., 1993). Plasmid DNA was introduced into cells by electroporation (Nuttley et al., 1993). Leu⁺ transformants recovered on YNAS agar plates were replica plated onto selective YNO agar plates and screened for restoration of their ability to use oleic acid as a sole carbon source. Total DNA was isolated from colonies that recovered growth and used to transform E. coli for plasmid recovery. Restriction fragments prepared from genomic inserts were subcloned and tested for their ability to functionally complement the pex20-1 strain. Various restriction endonuclease fragments of the PEX20 gene were cloned into the vectors pGEM5Zf (+) and pGEM7Zf (+) (Promega Corp., Madison, WI) for dideoxynucleotide sequencing of both strands. The deduced Pex20p amino acid sequence was compared with other known protein sequences using the GENEINFO (R) Blast Network Service (Blaster) of the National Center for Biotechnology Information (Bethesda, MD).

Targeted integrative disruption of the PEX20 gene was performed with the URA3 gene of Y. lipolytica. A 1.65-kbp SalI fragment containing the URA3 gene was made blunt and ligated in reverse orientation into a plasmid containing the PEX20 gene cut with NcoI and made blunt. In this way, 1,092 bp of the coding region and 2 bp of the 5′ untranslated region of the PEX20 gene were replaced with the URA3 gene flanked by 767 and 1,022 bp of the 5′ and 3′ regions, respectively, of the PEX20 gene. This construct was cleaved with HindIII/SphI to liberate the URA3 gene flanked by PEX20 sequences. The resulting linear construct was used to transform Y. lipolytica strains E122 and 22301-3 to uracil prototrophy. Ura⁺ transformants that were unable to grow on YNO agar were further characterized by Southern blotting. pex20::URA3 integrants were mated to each other, to wild-type strains, and to the pex20-1 mutant strain, and the resulting diploids were subjected to complementation and random spore analyses (Nuttley et al., 1993).

Double-labeling, indirect immunofluorescence microscopy (Szilard et al., 1995), and electron microscopy (Goodman et al., 1990) were performed as described. For morphometric analysis of random electron microscopic sections of cells, 12 × 14-cm prints and 8 × 10-cm negatives of 35–40 cell sections of each strain at 24,000–29,000 magnification were scanned and converted to digitized images with an Apple Color OneScanner (Apple Computers Inc., Cupertino, CA) and Adobe Photoshop 3.0 software (Adobe Systems Inc., San Jose, CA). Quantitation of digitized images was performed using NIH Image 1.55 software (National Institutes of Health, Bethesda, MD). Relative area of peroxisome section (%) was calculated as “area of peroxisome section/area of cell section × 100”. Peroxisomes were counted in electron micrographs, and data are expressed as the number of peroxisomes per μm³ of cell section volume.

The initial step in the subcellular fractionation of YPBO-grown cells was performed as described previously (Szilard et al., 1995) and included the differential centrifugation of lysed and homogenized spheroplasts at 1,000 g for 8 min at 4°C in a rotor (model JS13.1; Beckman Instrs., Inc., Palo Alto, CA) to yield a postnuclear supernatant (PNS) fraction. The PNS fraction was further subjected to differential centrifugation at 20,000 g for 30 min at 4°C in a rotor (model JS13.1; Beckman Instrs.) to yield pellet (20KgP) and supernatant (20KgS) fractions. The 20KgS fraction was further subfractionated by differential centrifugation at 200,000 g for 1 h at 4°C in a rotor (model TLA120.2; Beckman Instrs., Inc.) to yield pellet (200KgP) and supernatant (200KgS) fractions. The 200 KgP was further fractionated by flotation on a two-step sucrose gradient. Specifically, the 200 KgP was resuspended in 400 μl of 60% (wt/wt) sucrose in buffer H (5 mM MES, pH 5.5, 1 mM KCl, 0.5 mM EDTA, 0.1% [vol/vol] ethanol), overlaid with 2.3 ml of 50% (wt/wt) sucrose and 2.3 ml of 20% (wt/wt) sucrose (both in buffer H), and subjected to centrifugation in a rotor (model SW50.1; Beckman Instrs.) at 200,000 g for 18 h at 4°C. Gradients were fractionated from the bottom, and 18 fractions of ∼270 μl each were collected. Protein extraction and protease protection analysis of different subcellular fractions were performed as described (Szilard et al., 1995).

Highly purified peroxisomes were isolated essentially as described (Titorenko et al., 1996). 4 vol of 0.5 M sucrose in buffer H were added to the peak peroxisomal fraction 4 recovered after isopycnic centrifugation on a discontinuous sucrose gradient. Peroxisomes were sedimented through a 150-μl cushion of 2 M sucrose in buffer H by centrifugation at 200,000 g for 20 min at 4°C in a TLA120.2 rotor. The resultant pellet was resuspended in buffer H containing 1 M sorbitol and was subjected to further centrifugation on a linear 20–60% (wt/wt) sucrose gradient (in buffer H) at 197,000 g for 18 h at 4°C in a rotor (model SW41Ti; Beckman Instrs., Inc.). Peak peroxisomal fraction 5 equilibrating at a density of 1.21 g/cm³ was recovered, and peroxisomes were pelleted at 200,000 g for 20 min at 4°C in a TLA120.2 rotor, as described above. Pelleted peroxisomes were resuspended in 400 μl of 60% (wt/wt) sucrose in buffer H and subjected to flotation on a two-step sucrose gradient as described above. Peak peroxisomal fraction 11 was recovered and used for biochemical analyses. Peroxisomes isolated by this multistep method were greater than 97% pure, as judged by the presence of marker proteins of other organelles.

For pulse-chase experiments, YPBO-grown cells were pelleted at 10,000 g for 8 min at room temperature, washed three times with water, incubated in 50 mM potassium phosphate buffer, pH 7.5, containing 10 mM DTT for 10 min at 30°C, and repelleted at 10,000 g for 8 min at room temperature. Cells were resuspended at a concentration of 0.25 g/ml in 25 mM potassium phosphate buffer, pH 7.5, containing 0.55 M MgSO₄, and converted to spheroplasts by digestion at 30°C with Zymolyase 100T (ICN Biochemicals, Inc., Mississauga, Ontario, Canada) at 1 mg/ml of cells. Spheroplasts were harvested at 10,000 g for 8 min at room temperature and resuspended at a concentration of 6 OD₆₀₀/ml in YNO medium supplemented with 1 M sucrose. Spheroplasts were incubated for 90 min at 30°C, labeled with l-[³⁵S]methionine (ICN Biochemicals, Inc.) at 40 μCi/OD₆₀₀ for 1.5 min or 3 min at 30°C, and chased with an equal volume of 2× YPBO medium supplemented with 1 M sucrose and 10 mM unlabeled l-methionine. Samples were taken at various times of chase, and spheroplasts were immediately pelleted in a rotor (model F241.5; Beckman Instrs., Inc.) at 20,000 g for 2 min at 4°C. All subsequent steps were performed at 4°C, and all solutions contained unlabeled l-methionine. Spheroplasts were resuspended in 150 μl of buffer H containing 1 M sorbitol and a mixture of protease inhibitors, as described previously (Szilard et al., 1995). In protease protection experiments, protease inhibitors were omitted. Spheroplasts were osmotically lysed by addition of 300 μl of buffer H containing 0.1 M sorbitol and protease inhibitors. Lysis was greater than 90%, as determined by microscopy and enzymatic assay of the cytosolic marker G6PDH. The original osmolarity was reestablished by addition of 300 μl of buffer H containing 1.9 M sorbitol and protease inhibitors. The lysate was subjected to centrifugation at 1,000 g for 3 min at 4°C in a rotor (model F241.5; Beckman Instrs., Inc.) to yield a PNS fraction. The PNS fraction was loaded onto a 150-μl cushion of 1.2 M sucrose in buffer H containing protease inhibitors and subjected to centrifugation at 20,000 g for 20 min at 4°C in a rotor (model TLA120.2; Beckman Instrs., Inc.) to yield pellet (20KgP) and supernatant (20KgS) fractions. The 20KgS fraction was further sedimented through a 150-μl cushion of 1.2 M sucrose in buffer H containing protease inhibitors at 200,000 g for 20 min at 4°C in a rotor (model TLA120.2; Beckman Instrs., Inc.) to yield pellet (200KgP) and supernatant (200KgS) fractions. Immunoprecipitation under native conditions was performed by immunoaffinity chromatography with antibodies covalently coupled to protein A–Sepharose (Pharmacia LKB Biotechnology, Piscataway, NJ), as described previously (Szilard et al., 1995). For immunoprecipitation under denaturing conditions, SDS was added to 2%, and samples were warmed at 65°C for 10 min. Samples were allowed to cool to room temperature, and 4 vol of 60 mM Tris-HCl, pH 7.4, buffer containing 1.25% (vol/vol) Triton X-100, 190 mM NaCl, and 6 mM EDTA were added. Samples were subsequently processed for immunoprecipitation, as described (Franzusoff et al., 1991).

Chemical cross-linking of proteins was performed using dithiobis(succinimidylpropionate) (DSP) and disuccinimidyl suberate (DSS) (Pierce Chemical Co., Rockford, IL), essentially as described (Sanders et al., 1992; Marshall et al., 1996). Highly purified peroxisomes or the 200 KgP fraction were lysed in 20 mM sodium phosphate buffer, pH 7.5, containing 150 mM NaCl. Lysates were clarified by centrifugation at 200,000 g for 20 min at 4°C in a rotor (model TLA120.2; Beckman Instrs., Inc.), and the resultant supernatants were used for chemical cross-linking. Sodium phosphate buffer, pH 7.5, and NaCl were added to the supernatants to final concentrations of 20 and 150 mM, respectively. Cross-linking with DSP or DSS was initiated by the addition of cross-linker (50 mM stock in DMSO) and continued for 1 h at 4°C. Cross-linking was quenched by addition of 0.1 vol of 1 M Tris-HCl, pH 7.5, and incubation for 30 min at 4°C. SDS was added to 1.25%, and samples were warmed at 65°C for 20 min and then cooled to room temperature. 4 vol of 60 mM Tris-HCl, pH 7.4, 1.25% (vol/vol) Triton X-100, 190 mM NaCl, and 6 mM EDTA were added to the cooled samples, which were then subjected to immunoprecipitation, as described (Franzusoff et al., 1991). All immunoprecipitated samples were analyzed by SDS-PAGE under reducing or nonreducing conditions, i.e., with or without DTT in the sample buffer, respectively. Gels were treated with 22.2% 2,5-diphenyloxazole in either DMSO or glacial acetic acid (Coligan et al., 1995), dried, and then exposed to preflashed Kodak X-Omat AR X-ray film (Eastman-Kodak Co., Rochester, NY) at −80°C with intensifying screens.

Antibodies to Pex20p were raised in guinea pig and rabbit against a maltose-binding protein-Pex20p fusion, as described previously (Eitzen et al., 1995). To produce antibodies to Pex20p, the entire open reading frame of the PEX20 gene was amplified by PCR and inserted into pMAL-c2 (New England BioLabs, Beverly, MA) in-frame and downstream of the open reading frame encoding the maltose-binding protein (Eitzen et al., 1995).

Guinea pig polyclonal antibodies to Y. lipolytica isocitrate lyase (ICL), thiolase (THI), Pex2p, Pex5p, and Pex16p and to Saccharomyces cerevisiae acyl-CoA oxidase (AOX); and rabbit polyclonal anti-SKL antibodies have been described (Szilard et al., 1995; Eitzen et al., 1996, 1997). Rabbit polyclonal antibodies to S. cerevisiae malate synthase (MLS) (Eitzen et al., 1996) and to Y. lipolytica Sls1p (Boisramé et al., 1996), Kar2p (Titorenko et al., 1997) and Sec14p (Lopez et al., 1994) were described previously. Anti-MLS antibodies were kindly provided by A. Hartig (Institute of Biochemistry and Molecular Cell Biology, Vienna, Austria). Anti-Sls1p and anti-Sec14p antibodies were generous gifts of C. Gaillardin (Institut National Agronomique Paris-Grignon, Thiveral-Grignon, France). Rabbit polyclonal antibody A-9521 specific for S. cerevisiae glucose-6-phosphate dehydrogenase (G6PDH) was from Sigma Chemical Co. (St. Louis, MO).

Whole cell lysates were prepared as described (Goodman et al., 1990). The progressive permeabilization of yeast membranes with digitonin was performed as described (Zhang et al., 1993). In vitro coupled transcription/translation of the PEX20 gene was performed using the TNT T7 Quick-Coupled Transcription/Translation System (Promega Corp.). Blot overlay (Radu et al., 1995), including the denaturation/renaturation of transferred proteins by treatment with guanidinium chloride, followed by multistep dilution with water (Dingwall et al., 1995) and elution of bound proteins with urea (Radu et al., 1995), were performed according to established procedures. Enzymatic activities of catalase and cytochrome c oxidase (Szilard et al., 1995); NADPH:cytochrome c reductase, α-mannosidase, and vanadate-sensitive plasma membrane ATPase (Roberts et al., 1991); guanosine diphosphatase (Abeijon et al., 1989); fumarase (Smith et al., 1997); and alkaline phosphatase (Thieringer et al., 1991); and thiolase (Glover et al., 1994b) were determined by established methods. Inorganic phosphate liberated in assays of the activities of guanosine diphosphatase and vanadate-sensitive plasma membrane ATPase was measured as described (Lanzetta et al., 1979). Southern blot analysis (Ausubel et al., 1989), SDS-PAGE (Laemmli, 1970), and immunoblotting using a semi-dry electrophoretic transfer system (model ET-20; Tyler Research Instruments, Edmonton, Alberta, Canada) (Kyhse-Andersen, 1984) were performed as described. Antigen–antibody complexes were detected by enhanced chemiluminescence (Amersham Life Sciences, Oakville, Ontario, Canada). Quantitation of immunoblots was performed as described (Szilard et al., 1995).

We have previously reported the isolation of Y. lipolytica mutants that are unable to use oleic acid as a sole carbon source (Nuttley et al., 1993). This collection of ole⁻ mutants was screened by indirect immunofluorescence using antibodies to PTS1, to the peroxisomal matrix protein THI, which is targeted by PTS2, and to the peroxisomal membrane protein Pex2p. One mutant strain, pex20-1, exhibited a punctate pattern characteristic of peroxisomes when stained with either anti-PTS1 or anti-Pex2p antibodies (see below). However, unlike wild-type cells, a diffuse pattern of staining was seen with anti-THI antibodies, indicating that the pex20-1 mutant strain is impaired in the import of THI into peroxisomes.

The PEX20 gene was isolated by functional complementation of the pex20-1 strain. Of the 2 × 10⁴ Leu⁺ transformants screened, five strains were found to have restored growth on oleic acid. The complementing plasmids were recovered after transformation of E. coli. Restriction map analysis showed that all five complementing plasmids shared a 5.8-kbp region. Restriction fragments from this region were subcloned and tested for their ability to functionally complement the pex20-1 strain. The minimum complementing fragment was localized to a 2.9-kbp HindIII/BamHI restriction fragment. Sequencing this fragment revealed a 1,272-bp open reading frame encoding a 424-amino acid protein, Pex20p, with a predicted molecular weight of 47,274 (Fig. 1). Pex20p shows 16.3 to 28.1% identity to Pex5p peroxins of yeast and mammals, including Y. lipolytica Pex5p, but does not contain tetratricopeptide repeat motifs. Pex5p peroxins function as PTS1 import receptors (Erdmann et al., 1997). However, Pex20p is not required for the import of PTS1 proteins (see below). Despite its role in the peroxisomal import of PTS2-containing THI, Pex20p does not exhibit homology to yeast or human Pex7p peroxins, which act as import receptors for PTS2-containing proteins (Erdmann et al., 1997).

The putative PEX20 gene was disrupted by targeted integration of the Y. lipolytica URA3 gene to create the strains pex20KO-1A and pex20KO-2B in the A and B mating types, respectively (Table I). Strains containing a disrupted PEX20 gene were unable to grow on oleic acid and had the morphological and biochemical characteristics of the original pex20-1 strain (see below). The diploid strain D1-20 from the mating of wild-type strain E122 to strain pex20KO-2B, and the diploid strains D2-20 and D3-20 from the mating of strains pex20KO-1A and pex20-1, respectively, to wild-type strain 22301-3 (Table I), could grow on oleic acid-containing medium, demonstrating the recessive nature of the pex20-1, pex20KO-1A, and pex20KO-2B mutations. The diploid strain D4-20 from the mating of the original pex20-1 mutant strain to strain pex20KO-2B, and the diploid strain D5-20 from the mating of the pex20KO-1A strain to strain pex20KO-2B (Table I), were unable to grow on oleic acid-containing medium. Random spore analysis of the diploid strains D1-20 and D2-20 showed monogenic segregation for both Ura⁺ and ole⁻ phenotypes, whereas the URA3 gene invariably cosegregated with the ole⁻ phenotype. Random spore analysis of the diploid strain D4-20 and of the diploid strain D5-20 revealed no meiotic segregants with a recombinant ole⁺ phenotype. Taken together, these data indicate that the authentic PEX20 gene had been cloned.

Antibodies to Pex20p specifically recognized an ∼60-kD polypeptide in whole cell lysates prepared from the wild-type strain, but not from the pex20-1 or pex20KO mutant strains (Fig. 2 A). The difference between the predicted molecular weight of Pex20p (47,274 D) and its empirical molecular weight (∼60,000 D), as determined from SDS-PAGE, is an intrinsic feature of this protein. Indeed, in vitro coupled transcription/translation of the PEX20 gene yielded a polypeptide that comigrated identically with in vivo synthesized Pex20p on SDS-PAGE (data not shown). Why Pex20p shows reduced electrophoretic mobility on SDS-PAGE is unknown, but may be due to the large number of acidic amino acids (14.4%) present in Pex20p.

Synthesis of Pex20p is induced by growth of Y. lipolytica on oleic acid (Fig. 2,B). A small amount of Pex20p was present in wild-type cells grown in glucose-containing medium. Shifting cells to oleic acid-containing medium caused an increase in the level of Pex20p, which reached levels ∼24 times that found in glucose-grown cells 8 h after the shift (Fig. 2 B).

In wild-type cells, most (98 ± 8%) of Pex20p localized to the cytosol, whereas 6 ± 2% associated with a high-speed (200,000 g) pelletable organellar fraction (200KgP) (Fig. 2,C) enriched for high-speed pelletable (HSP) peroxisomes (see below). Cytosolic G6PDH was not detected in the 200KgP isolated under these fractionation conditions (data not shown). No Pex20p was detected in a low-speed (20,000 g) organellar fraction (20KgP) (Fig. 2,C) enriched for low-speed pelletable (LSP) peroxisomes (see below). The predominantly cytosolic location of Pex20p was confirmed by two additional approaches. Spheroplasts of oleic acid-grown wild-type cells were permeabilized by incubation with increasing amounts of digitonin, and the subsequent leakage of Pex20p and marker proteins from the cells was quantitated. Pex20p was released at low concentrations of digitonin, as was the cytosolic marker G6PDH. 50% release of both Pex20p and G6PDH was observed at 21 μg digitonin/ml (Fig. 2,D). The peroxisomal matrix proteins CAT and THI, and the mitochondrial matrix protein fumarase (FUM), were released at much higher concentrations of digitonin (Fig. 2,D). Double-labeling, indirect immunofluorescence analysis of wild-type cells grown in oleic acid-containing medium with anti-THI antibodies yielded a punctate pattern of staining characteristic of peroxisomes (Fig. 2,E). In contrast, a diffuse pattern of staining was observed with anti-Pex20p antibodies (Fig. 2 E), indicating that most Pex20p is localized to the cytosol in wild-type cells.

Pex20p localized to the 200KgP is present in membrane-associated form and is not due to aggregation. Flotation of the 200KgP from wild-type cells on a two-step sucrose gradient revealed that Pex20p, similar to the matrix proteins AOX and THI, floated out of the most dense sucrose and concentrated at the interface between 50 and 20% sucrose (Fig. 2,F). The distribution of Pex20p around peak fraction 11 coincided with the distributions of AOX and THI (Fig. 2,F), suggesting that Pex20p is localized to HSP peroxisomes. Further fractionation of the 200KgP on additional gradients showed that Pex20p cofractionated with the peroxisomal proteins AOX, THI, MLS, Pex2p, and Pex16p, but not with markers of the ER, Golgi, vacuole, or plasma membrane (data not shown). Together, these data indicate that Pex20p is localized to HSP peroxisomes and is not associated with any other organelle. Extraction of the 200KgP-associated form of Pex20p with various solubilizing agents showed that Pex20p fractionated as a peripheral membrane protein that was solubilized to a significant extent by either 1 M NaCl or 1 M urea or completely by 0.1 M Na₂CO₃, pH 11 (Fig. 2,G). Protease protection experiments revealed that the 200KgP-associated form of Pex20p was sensitive to trypsin digestion even in the absence of Triton X-100, whereas the peroxisomal matrix protein THI was degraded by trypsin only when the membranes of the 200KgP-associated organelles were disrupted by the detergent (Fig. 2 H). Taken together, these data indicate that in wild-type cells, the 200KgP-associated form of Pex20p is a peripheral membrane protein associated with the cytosolic surface of HSP peroxisomes.

Intracellular trafficking of Pex20p was studied by immunoprecipitation of pulse-labeled and chased Pex20p from the 20KgP, 200KgP, and 200KgS (cytosolic) fractions of oleic acid-grown wild-type cells. At 0 min of chase, all Pex20p was found in the cytosol (Fig. 3,A). By 3 min of chase, a minor portion of Pex20p (10% of the total pool) was chased to the 200KgP enriched for HSP peroxisomes. By 10 min of chase, the 200KgP-associated Pex20p reached its maximal level of 26% of the total pool of pulse-labeled Pex20p (Fig. 3,A). After 10 min of chase, the level of 200KgP-associated Pex20p gradually decreased and, by 90 min of chase, reached 55% of its maximal level (Fig. 3,A). No Pex20p was chased to the 20KgP (Fig. 3,A), enriched for LSP peroxisomes. All pulse-labeled Pex20p chased to the 200KgP was sensitive to trypsin digestion even in the absence of Triton X-100 (Fig. 3 B) and, therefore, was associated with the cytosolic surface of HSP peroxisomes present in the 200KgP. These data suggest that a minor fraction (not more than 26%) of newly synthesized Pex20p is targeted from the cytosol to the outer surface of HSP peroxisomes with a half-time of 4 min. These results are consistent with a model in which Pex20p shuttles between the cytosol and the cytosolic surface of HSP peroxisomes; however, we cannot rule out the possibility that the decrease in the levels of pulse-labeled Pex20p in the 200KgP over time is due to its turnover in this fraction.

We assessed the effects of mutations in the PEX20 gene on the size and number of peroxisomes in oleic acid-grown cells. EM and morphometric analysis showed that in wild-type cells, 36.2 ± 0.8% of all peroxisomes had relative areas of peroxisome section from 1.01–1.5%, whereas very small peroxisomes with relative areas of peroxisome section from 0.05 to 0.2% were much less abundant (5.3 ± 0.2% of all peroxisomes) (Fig. 4). Neither the pex20-1 nor the pex20KO mutant showed a significant decrease in the percentage of peroxisomes with relative areas of peroxisome section from 0.21 to 3.0%, but both mutants showed 3.5- to 4.9-fold increases in the percentage of very small peroxisomes with relative areas of peroxisome section from 0.05 to 0.2% (Fig. 4). This increased percentage in small peroxisomes in mutant cells was due to a large increase in their number per cell rather than to a reduction in the number of large peroxisomes. Indeed, mutations in the PEX20 gene did not significantly decrease the number of peroxisomes with relative areas of peroxisome section greater than 0.2% (Fig. 4). Rather, the pex20-1 and pex20KO mutations led to 3.5- to 4.5-fold increases in the number of very small peroxisomes with relative areas of peroxisome section less than 0.2% (7.6 ± 0.8 and 9.9 ± 1.0 peroxisomes per μm³ of cell section volume, respectively) vis-à-vis their number in wild-type cells (2.2 ± 0.3 peroxisomes per μm³ of cell section volume). Biochemical results demonstrating the selective enrichment of MLS, CAT, Pex2p, and Pex16p in HSP peroxisomes in pex20-1 and pex20KO cells (Fig. 5), combined with immunofluorescence data showing an accumulation of punctate structures decorated by anti-MLS or anti-Pex2p antibodies but not anti-SKL antibodies in these cells (see Fig. 8 A), strengthened the interpretation of EM data showing a large increase in the number of small peroxisomes in the pex20-1 and pex20KO mutants.

Subcellular fractionation of wild-type cells grown in oleic acid-containing medium showed that from 82 to 92% of peroxisomal matrix (AOX, 62- and 64-kD anti-SKL reactive proteins, ICL, CAT, MLS, and THI) and membrane (Pex2p and Pex16p) proteins was associated with a LSP organellar fraction (20KgP), whereas from 10 to 23% was localized to a HSP fraction (200KgP) (Fig. 5). Matrix proteins and the intraperoxisomal peripheral membrane protein Pex16p localized to the 20KgP and 200KgP of wild-type cells were resistant to exogenously added protease (trypsin) in the absence of Triton X-100 and were therefore in membrane-enclosed structures (data not shown). In contrast, in pex20-1 and pex20KO cells, THI was found exclusively in the 200KgS (cytosolic) fraction (Fig. 5) and was accessible to external protease (data not shown). No other peroxisomal matrix or membrane protein was mislocalized to the cytosol in pex20 mutants (Fig. 5). However, mutations in the PEX20 gene altered the relative distributions of CAT, MLS, Pex2p, and Pex16p between the 20KgP and 200KgP fractions. In wild-type cells, 82–87% of these four proteins was found in the 20KgP, whereas 8–21% was associated with 200KgP (Fig. 5). In pex20-1 and pex20KO cells, 44–74% of CAT, MLS, Pex2p, and Pex16p was localized to the 200KgP, whereas 30–45% was found in the 20KgP (Fig. 5). Neither the pex20-1 nor the pex20KO mutation altered the relative distributions of AOX, 62-kD SKL, 64-kD SKL and ICL between the 20KgP and 200KgP fractions (Fig. 5).

Pulse-chase labeling and immunoprecipitation of proteins from the 20KgP, 200KgP, and 200KgS (cytosolic) fractions showed that in wild-type cells, peroxisomal matrix proteins moved from the cytosol to the 200KgP fraction and from there to the 20KgP fraction (Fig. 6,A, data for AOX, MLS and THI; data for 62-kD SKL, 64 kD-SKL, and ICL not shown). Furthermore, in wild-type cells, the peroxisomal membrane proteins Pex2p and Pex16p were initially targeted to the 20KgP, presumably to a subpopulation of ER-derived membranes that accumulates in this fraction (Titorenko and Rachubinski, 1998), and were subsequently chased to the 200KgP and from there to the 20KgP (Fig. 6,A, data for Pex2p; data for Pex16p not shown). Mutations in the PEX20 gene selectively abolished the targeting of THI from the cytosol to the 200KgP (Fig. 6,A and B, compare data for THI). All THI accumulating in the cytosol of the pex20-1 and pex20KO mutant strains was present as the 47-kD precursor form (data not shown). In wild-type cells, maturation of the THI precursor to the 45-kD mature form occurred in the 200KgP (Fig. 6 C).

In contrast to the targeting defect for THI, the rates and efficiencies of targeting from cytosol to the 200KgP of other peroxisomal matrix and membrane proteins were unaffected in pex20 mutant strains (Fig. 6,A and B, compare half-times for the targeting of AOX, MLS, and Pex2p from the 200KgS to the 200KgP; data for 62-kD SKL, 64-kD SKL, and ICL not shown). The pex20-1 and pex20KO mutations also did not affect the rates and efficiencies of transit of AOX, 62-kD SKL, 64-kD SKL and ICL from the 200KgP to the 20KgP (Fig. 6,A and B, compare data for AOX; data for 62-kD SKL, 64-kD SKL, and ICL not shown). However, mutations in the PEX20 gene significantly decreased the rates and efficiencies of transit of MLS, Pex2p, and Pex16p from the 200KgP to the 20KgP (Fig. 6, A and B, compare data for MLS and Pex2p; data for Pex16p not shown).

The accumulation of thiolase in the cytosol and the differential transit of two subsets of peroxisomal proteins from the 200KgP to the 20KgP in pex20-1 and pex20KO mutants should lead to the accumulation in these mutants of LSP peroxisomes that completely lack THI, are partially depleted of MLS, Pex2p, and Pex16p, but contain wild-type amounts of AOX, 62-kD SKL, and 64-kD SKL and ICL. To test this prediction, the 20KgP fractions from oleic-acid grown wild-type, pex20-1, and pex20KO cells were separated by isopycnic centrifugation on a discontinuous sucrose gradient (Fig. 7). In the wild-type strain, all peroxisomal proteins were primarily in fractions 1–7, peaking in fraction 4 at a density of 1.21 g/cm³ (Fig. 7, left panel). In the pex20-1 and pex20KO mutants, all peroxisomal proteins, except THI, also peaked in fraction 4 at the same density (Fig. 7, middle and right panels, respectively). No THI was recovered in gradient fractions isolated from pex20-1 and pex20KO cells (Fig. 7, middle and right panels). Therefore, LSP peroxisomes accumulating in pex20 mutants have the same buoyant density as LSP peroxisomes present in wild-type cells, but they lack THI. AOX, 62-kD SKL, and 64-kD SKL, and ICL were present in LSP peroxisomes of the pex20-1 and pex20KO strains at levels comparable to those found in LSP peroxisomes of the wild-type strain. On the other hand, the levels of MLS, CAT, Pex2p, and Pex16p in LSP peroxisomes of the pex20-1 and pex20KO strains were 37–57% of their levels in LSP peroxisomes of the wild-type strain. Again, no THI was found in LSP peroxisomes of the pex20 mutant strains.

In immunofluorescence analysis, oleic acid-grown wild-type cells showed a punctate pattern of staining characteristic of peroxisomes with anti-THI antibodies, whereas the pex20-1 and pex20KO strains showed a diffuse pattern of staining characteristic of the cytosol (Fig. 8,A, leftmost panels). Anti-MLS and anti-SKL antibodies (Fig. 8,A, middle panels), or anti-Pex2p and anti-SKL antibodies (Fig. 8,A, rightmost panels), revealed a punctate pattern of staining in wild-type, pex20-1, and pex20KO cells. These data support the conclusion that mutations in the PEX20 gene abolish THI import into peroxisomes but not the import of other peroxisomal matrix and membrane proteins. It is interesting to note that most cells of the pex20-1 and pex20KO strains contained up to 20% of peroxisomes that were decorated by anti-MLS or anti-Pex2p antibodies, but not by anti-SKL antibodies (Fig. 8,B, arrows). Similarly decorated punctate structures were rarely observed in wild-type cells. Anti-MLS and anti-Pex2p antibodies always colocalized in punctate structures in wild-type, pex20-1, and pex20KO cells (Fig. 8 B, rightmost panels, data for pex20KO cells; data for wild-type and pex20-1 cells is not shown). Therefore, wild-type cells apparently accumulate a few peroxisomes containing both MLS and Pex2p but not anti-SKL–reactive proteins, and mutations in the PEX20 gene significantly increase the number of this type of peroxisome. These data, combined with EM results presented above showing a large increase in very small peroxisomes in pex20 cells versus wild-type cells; the selective enrichment in the 200KgP of MLS, CAT, Pex2p, and Pex16p; and the decreased rate and efficiency of transit of MLS, Pex2p, and Pex16p from the 200KgP to the 20KgP, suggest that mutations in the PEX20 gene lead to a large increase in very small peroxisomes that are enriched for MLS, CAT, Pex2p, and Pex16p. The role that Pex20p might play in the accumulation of very small HSP peroxisomes enriched for these proteins is currently under investigation.

Our data showing that (a) most Pex20p is localized to the cytosol, (b) Pex20p apparently shuttles between the cytosol and the cytosolic surface of HSP peroxisomes, and (c) Pex20p is essential for the import of THI from the cytosol into HSP peroxisomes, led to a prediction that Pex20p and THI form a complex in the cytosol. To test this prediction, pulse-labeled Pex20p and THI were immunoprecipitated from the 200KgS (cytosolic) fraction of the wild-type strain under native conditions. Both anti-Pex20p and anti-THI antibodies immunoprecipitated two radiolabeled proteins with molecular weights of 59.8 and 47.0 kD, which could be decorated with anti-Pex20p and anti-THI antibodies, respectively (Fig. 9,A, panels 1 and 6). Therefore, Pex20p and THI form a complex in the cytosol of wild-type cells. Proteins that did not bind to protein A–Sepharose and were recovered in the flow-through following native (first) immunoprecipitation with anti-Pex20p or anti-THI antibodies were subjected to a second immunoprecipitation under denaturing conditions with anti-Pex20p or anti-THI antibodies. Neither Pex20p nor THI was recovered in the flow-through when the native immunoprecipitation had been performed with anti-Pex20p antibodies (Fig. 9,A, panels 2 and 3, respectively). Therefore, all newly synthesized THI was associated with Pex20p in the cytosol of wild-type cells. On the other hand, a significant portion of Pex20p, but no THI, was recovered in the flow-through when the native immunoprecipitation had been performed with anti-THI antibodies (Fig. 9,A, panels 7 and 8, respectively). Therefore, in the cytosol of wild-type cells, Pex20p is present in two forms, namely associated or unassociated with newly synthesized THI. Newly synthesized 62-kD SKL, 64-kD SKL, and MLS were not associated with either Pex20p or THI in the cytosol of wild-type cells, since none of these proteins was coimmunoprecipitated by anti-Pex20p (Fig. 9,A, panels 1, 4, and 5) or anti-THI (Fig. 9,A, panels 6, 9, and 10) antibodies under native conditions. Native immunoprecipitation using anti-SKL antibodies did not lead to coimmunoprecipitation of Pex20p, THI, or MLS with 62-kD SKL and 64-kD SKL proteins (Fig. 9 A, panels 11–15).

We tested whether the cleaved PTS2 of THI is essential for Pex20p binding. Only the precursor form of THI (pTHI) accumulates in peroxisomes of the pex5KO mutant deficient in the PTS1 import receptor (Fig. 9,B) (Szilard et al., 1995), whereas wild-type peroxisomes contain only the mature (mTHI) form of THI lacking its PTS2 (Fig. 9,B). Radiolabeled pTHI and mTHI forms were purified from peroxisomes of pex5KO and wild-type cells, respectively, and used as probes in a blot overlay assay with unlabeled proteins of either the cytosol of wild-type cells containing Pex20p (refer to Fig. 2,A) or the cytosols of pex20-1 and pex20KO cells lacking Pex20p (refer to Fig. 2,A). Both pTHI and mTHI bound with equal efficiency to a 59.8-kD protein present in the cytosol of wild-type cells but absent in the cytosols of pex20-1 or pex20KO cells (Fig. 9,C). Neither pTHI nor mTHI interacted with any protein when the wild-type cytosol was depleted of Pex20p by native immunoprecipitation with anti-Pex20p antibodies (Fig. 9,C). Stripping of the proteins bound to the wild-type cytosol confirmed the binding of pTHI and mTHI (Fig. 9 D). These data indicate that the association of Pex20p with newly synthesized THI in the cytosol of wild-type cells is not dependent on the presence of a PTS2 in THI.

We tested whether THI remains associated with Pex20p after targeting of their complex from the cytosol to HSP peroxisomes. Proteins pulse labeled and chased for 5 min were immunoprecipitated from the 200KgP fraction of the wild-type strain under native conditions with anti-Pex20p or anti-THI antibodies. Proteins that did not bind to protein A–Sepharose and were recovered in the flow-through were subjected to a second immunoprecipitation under denaturing conditions with anti-Pex20p or anti-THI antibodies. No coimmunoprecipitation of Pex20p and THI was seen with either anti-Pex20p or anti-THI antibodies (Fig. 10, panels 1–3 and 6–8). Therefore, Pex20p and THI associated with HSP peroxisomes do not form a complex with each other. This finding is in agreement with the results of protease protection experiments showing that Pex20p associated with the 200KgP is a peripheral membrane protein bound to the cytosolic surface of HSP peroxisomes, whereas THI is located exclusively in the matrix of HSP peroxisomes (refer to Figs. 2,H and 3,B). 62-kD SKL, 64-kD SKL, and MLS associated with HSP peroxisomes also did not coimmunoprecipitate with either Pex20p or THI (Fig. 10, panels 1, 4–6, 9, and 10). Therefore, none of the matrix proteins tested associated with each other or with Pex20p in HSP peroxisomes.

As shown above, all newly synthesized THI is associated with Pex20p in the cytosol of wild-type cells. We studied the oligomeric state of THI and of the complex formed between THI and Pex20p in wild-type cytosol by chemical cross-linking. The 200KgS (cytosolic) fraction of oleic acid-grown wild-type cells pulse-labeled with l-[³⁵S]methionine was subjected to increasing concentrations of the noncleavable, amine-reactive, homobifunctional cross-linker DSS. Proteins were immunoprecipitated under denaturing conditions with anti-THI antibodies, and proteins that did not bind to protein A–Sepharose and were recovered in the flow-through were subjected to a second immunoprecipitation under denaturing conditions with anti-Pex20p antibodies. SDS-PAGE analysis of cross-linked radiolabeled proteins immunoprecipitated with anti-THI antibodies revealed a 217-kD complex whose amount increased with increasing DSS, and a 47-kD (monomeric) form of THI that concomitantly decreased and eventually disappeared (Fig. 11,A). Immunoprecipitation of Pex20p from the flow-through of the first immunoprecipitation with anti-THI antibodies showed that as the concentration of DSS increased, the level of unbound Pex20p decreased (Fig. 11,B). Immunoprecipitation of cytosolic proteins with anti-Pex20p antibodies revealed that as the DSS concentration increased, a prominent 217-kD complex appeared and increased (Fig. 12,B). The electrophoretic mobility of this complex was identical to that of the 217-kD complex immunoprecipitated with anti-THI antibodies when samples were separated on the same gel (data not shown). Immunoprecipitation of THI from the protein A–Sepharose flow-through after immunoprecipitation with anti-Pex20p antibodies showed that as the DSS concentration increased, unbound, monomeric THI decreased and eventually disappeared (Fig. 11,D). Taken together, these data indicate that in wild-type cytosol, all newly synthesized THI and a significant portion of Pex20p are components of a 217-kD complex. The fact that only THI and Pex20p can be coimmunoprecipitated under native conditions with either anti-THI or anti-Pex20p antibodies (refer to Fig. 9,A) suggests that they are the only protein components of the 217-kD complex. That this was the case was confirmed by experiments using the thiol-cleavable, homobifunctional cross-linker DSP. Treatment of cytosolic proteins with increasing concentrations of DSP, followed by immunoprecipitation with anti-THI antibodies and evaluation of radiolabeled proteins by nonreducing SDS-PAGE and fluorography, showed that even at low concentrations of DSP, a 217-kD complex appeared, whereas there was concomitant disappearance of 47-kD monomeric THI (Fig. 11,E, left). When DSP was cleaved with DTT, and the immunoprecipitated proteins were resolved by SDS-PAGE under reducing conditions, only two proteins of 59.8 and 47.0 kD were detected in the samples treated with DSP (Fig. 11,E, right). To quantitate the stoichiometry of Pex20p and THI in the 217-kD complex, the densities of the signals for Pex20p and THI in Fig. 11 E, right, lane 1.0 were divided by 17 and 8, respectively, the number of methionine residues in Pex20p and THI, respectively. The ratio obtained was 1.1 ± 0.1 (data from four independent experiments), indicating that Pex20p and THI are present in equimolar amounts in the complex. Based on this observation and on a consideration of the molecular weights of Pex20p, pTHI, and the complex formed by these two proteins (59.8-, 47.0-, and 217-kD, respectively), we conclude that in the cytosol of wild-type cells, all newly synthesized THI is associated with Pex20p in a heterotetrameric complex composed of two polypeptide chains of each protein.

We also tested the effects of the pex20-1 and pex20KO mutations on the oligomerization of THI in the cytosol. Interestingly, most THI in the cytosol of pex20 mutants is present as the monomeric form, even at the highest concentration of DSS (Fig. 11, F and G). Therefore, mutations in the PEX20 gene affect oligomerization of newly synthesized THI in the cytosol and prevent the formation of the heterotetrameric complex between Pex20p and THI.

We also evaluated the oligomeric state of mature THI in LSP and HSP peroxisomes of oleic-acid grown wild-type cells. Both forms of peroxisomes were treated with increasing concentrations of DSS or DSP and then subjected to immunoprecipitation with anti-THI antibodies. Treatment of DSS cross-linked proteins under reducing conditions and DSP cross-linked proteins under nonreducing conditions revealed a prominent 86-kD protein that increased with increasing concentration of cross-linker (Fig. 11, H and J, and I and K [left panels]). No monomeric THI was detected in either LSP or HSP peroxisomes in samples treated with the highest concentration of cross-linker. On the other hand, only monomeric THI was seen in DSP-treated samples cleaved with DTT and analyzed by SDS-PAGE under reducing conditions (Fig. 11, I and K, right panels). From these data, we conclude that the mature form of THI is present exclusively as a homodimer in LSP and HSP peroxisomes of the wild-type strain.

Chemical cross-linking of pulse-labeled proteins, followed by the immunoprecipitation with anti-ICL or anti-AOX antibodies, was used to study the oligomeric state of ICL and AOX in the cytosol and in peroxisomes of the wild-type and pex20 mutant strains. Treatment with increasing concentrations of either DSS or DSP showed that ICL and AOX were present primarily as complexes rather than as monomers in the 200KgS (cytosolic) fraction (Fig. 12, A, C, G, and I), in LSP peroxisomes (Fig. 12, B, E, H, and K) and in HSP peroxisomes (data not shown) of the wild-type strain. The electrophoretic mobilities of ICL- and AOX-containing complexes from the cytosol were identical to those from LSP and HSP peroxisomes when resolved by SDS-PAGE on the same gel (data not shown). DTT treatment and reducing SDS-PAGE analysis of DSP cross-linked samples showed that only the monomeric forms of ICL and AOX could be recovered from cytosol (Fig. 12, D and J), LSP peroxisomes (Fig. 12, F and L) and HSP peroxisomes (data not shown) of the wild-type strain. The oligomerization of ICL or AOX was unaffected in both the pex20-1 and pex20KO mutants (Fig. 12, data for the pex20KO strain; data for the pex20-1 strain not shown). From these data, we conclude that newly synthesized ICL and AOX rapidly oligomerize in the cytosol of wild-type cells and are imported into peroxisomes as homooligomers. Mutations in the PEX20 gene do not affect the formation of the homooligomers of ICL and AOX in the cytosol and do not impair their subsequent import into peroxisomes.

Our kinetic data show that the oligomerization of THI in the cytosol precedes its import into the peroxisome. At the end of a short pulse with radiolabeled methionine, all labeled THI was found exclusively in a cytosolic heterotetrameric complex of two THI and two Pex20p polypeptide chains. Labeled THI was chased to peroxisomes and was imported exclusively as a homodimer. Mutations in the PEX20 gene not only affected the oligomerization of THI in the cytosol but also abolished import of THI into peroxisomes. Two interpretations for the observed effects of pex20 mutations on the oligomerization and peroxisomal targeting of THI can be proposed. First, Pex20p could be required only for the oligomerization and maintenance of the import-competent oligomeric state of THI and, therefore, oligomerization of THI in the cytosol not only precedes but is also required for its targeting to the peroxisome. Second, Pex20p could not only act as a cytosolic chaperone assisting oligomerization of THI in the cytosol, but could also function as a cytosolic THI import receptor that interacts with a specific docking site(s) on the peroxisomal membrane, thereby mediating docking of THI to the peroxisome.

Our data also show that not only THI, which contains a PTS2, but also ICL, which is targeted by PTS1 (Barth and Scheuber, 1993), and AOX, which lacks any conserved variant of PTS1 or PTS2 (Nicaud et al., 1988), are imported into peroxisomes of Y. lipolytica as oligomers. Our results extend recent observations that peroxisomes can translocate folded, disulfide-bonded and oligomeric proteins, and even 9-nm colloidal gold particles, across their membranes (Glover et al., 1994,a; McNew and Goodman, 1994; Walton et al., 1995; Elgersma et al., 1996; Häusler et al., 1996; Leiper et al., 1996; Lee et al., 1997). Although polypeptide chains lacking a PTS can be translocated into the peroxisomal matrix in a piggyback fashion on PTS1- and PTS2-containing proteins (for review see Rachubinski and Subramani, 1995; McNew and Goodman, 1996; Subramani, 1998), we did not see evidence of piggybacking of THI, ICL, or AOX into the peroxisomal matrix in Y. lipolytica. According to our data, THI penetrates the peroxisome exclusively as a homodimer rather than as a component of a cytosolically formed heterotetrameric complex of THI and Pex20p. Our data also show that newly synthesized ICL and AOX rapidly homooligomerize in the cytosol and are imported as homooligomers, which suggests that they are not piggybacked into peroxisomes.

Mutations in the PEX20 gene abolish the assembly of newly synthesized monomers of THI into a heterotetrameric complex of THI and Pex20p, suggesting that Pex20p may act as a cytosolic chaperone that assists in the oligomerization of THI in vivo. Molecular chaperones were originally defined as temporary partners that assist in the oligomeric assembly of unassembled subunits of oligomeric proteins (for review see Hendrick and Hartl, 1993). Our results indicate that two polypeptides of Pex20p bind temporarily to two polypeptide chains of THI and that this heterotetrameric complex dissociates upon binding to the outer surface of the peroxisomal membrane, releasing THI homodimer into the peroxisomal matrix. Moreover, since oligomerization of THI in the cytosol (see above) is a prerequisite for its targeting to the peroxisome, we propose that Pex20p acts as a putative cytosolic chaperone to maintain the import-competent oligomeric state of THI. Our data also show that the substrate specificity of Pex20p as a putative cytosolic chaperone is limited to THI. Pex20p does not bind the newly synthesized cytosolic form of any peroxisomal protein other than THI, and it is not essential for the targeting of any peroxisomal protein other than THI from the cytosol to the peroxisome nor is it required for the oligomeric assembly of ICL and AOX in the cytosol. Pex20p is also apparently not required for the folding of newly synthesized monomers of THI in the cytosol, as our results show that newly synthesized monomeric THI is enzymatically active (data not shown), does not form insoluble aggregates and is not degraded in the cytosol of the pex20 mutant strains. This step of THI maturation in the cytosol may be assisted by other cytosolic chaperone(s) such as TRiC (TCP1 ring complex), the eukaryotic homologue of GroEL (for review see Horwich and Willison, 1993; Bukau et al., 1996; Hartl, 1996), or cytosolic Hsp70. TriC has been shown to assist in the folding of the peroxisomal matrix protein, firefly luciferase (Frydman et al., 1994; Frydman and Hartl, 1996), and to bind to the peroxisomal membrane protein, PMP22 (Pause et al., 1997). Cytosolic Hsp70 also binds to firefly luciferase (Frydman et al., 1994; Frydman and Hartl, 1996) and has been shown to play an essential role in peroxisomal protein import (Walton et al., 1994).

The results of this study can be summarized in the following model for Pex20p-assisted oligomerization and peroxisomal import of THI. Monomeric Pex20p binds newly synthesized monomeric THI in the cytosol and promotes the formation of a heterotetrameric complex composed of two polypeptide chains of each protein. The association of Pex20p with THI is independent of the PTS2 in THI. The heterotetrameric complex binds to the peroxisomal membrane. Translocation of the THI homodimer into the peroxisomal matrix may release Pex20p monomers back to the cytosol, thereby permitting a new cycle of binding- oligomerization-targeting-release for Pex20p and THI. Binding of Pex20p to THI is PTS2 independent because, as we have shown, Pex20p interacts with PTS2-containing and PTS2-lacking forms of THI with equal efficiency. No receptor that binds PTS2 directly and is required for import of PTS2-containing THI has yet been identified in Y. lipolytica. However, PTS2 receptors of the Pex7p peroxin family have been identified in S. cerevisiae (Marzioch et al., 1994; Zhang and Lazarow, 1995, 1996; Rehling et al., 1996), Kluyveromyces lactis (Motley et al., 1997), Pichia pastoris (Elgersma et al., 1998), mouse (Braverman et al., 1997) and human (Braverman et al., 1997; Motley et al., 1997; Purdue et al., 1997). The exact subcellular location of the PTS2 receptor is a matter of debate and has been reported as mainly cytosolic and partially peroxisomal (Marzioch et al., 1994; Rehling et al., 1996) or exclusively intraperoxisomal (Zhang and Lazarow, 1995, 1996). Accordingly, two alternative models for PTS2 receptor action have been proposed. The first proposes that Pex7p is a mobile receptor, shuttling between the cytosol and the peroxisome (Marzioch et al., 1994; Rehling et al., 1996). The second proposes that Pex7p acts from within the peroxisome to pull THI inside the organelle and predicts the existence of another peroxin to interact with PTS2-containing THI, either in the cytosol or at the outer surface of the peroxisomal membrane (Zhang and Lazarow, 1995, 1996). Both models could accommodate the existence of a cytosolic peroxin such as Pex20p that not only maintains the oligomeric (import-competent) state of THI but which could also stabilize a conformation of THI that makes PTS2 more accessible to Pex7p. In the first model, cytosolic Pex20p could interact with a region of THI outside of the PTS2, whereas cytosolic Pex7p would bind the PTS2 directly. The cooperative action of Pex7p and Pex20p in the cytosol would permit correct targeting of THI to the peroxisomal docking protein Pex14p (Albertini et al., 1997), followed by translocation of THI into the matrix. In the second model, cytosolic Pex20p would maintain the oligomeric (import-competent) state of THI and could stabilize a conformation of THI that makes PTS2 more accessible to Pex7p, thereby mediating targeting of THI to the peroxisomal membrane. Subsequent interaction of intraperoxisomal Pex7p with an exposed PTS2 would initiate pulling of THI inside the organelle. It should be noted that in the blot overlay assay, no cytosolic protein other than Pex20p was seen to interact with pTHI or mTHI. The inability to detect a cytosolic PTS2 receptor that interacts only with pTHI but not with mTHI may be the result of Pex20p having a significantly higher affinity for pTHI than does any other cytosolic protein. An alternative explanation might be that the Y. lipolytica PTS2 receptor is exclusively intraperoxisomal. We are currently developing genetic screens aimed at identifying Pex7p of Y. lipolytica so as to critically evaluate both models in regards to a cooperative action of Pex7p and Pex20p in the peroxisomal import of THI.

In conclusion, the results described herein provide evidence that the oligomerization of THI in the cytosol requires the peroxin Pex20p and is a prerequisite for THI import into peroxisomes in Y. lipolytica. THI and Pex20p interact in a PTS2-independent manner and form a heterotetrameric cytosolic complex composed of two polypeptide chains of each protein. After targeting of the complex to the peroxisomal membrane, the homodimer of THI is translocated into the peroxisomal matrix, and Pex20p monomers are envisaged to recycle to the cytosol to permit a new cycle of binding-oligomerization-targeting-release between Pex20p and THI.

We thank H. Chan (University of Alberta, Edmonton, Alberta, Canada) for help with electron microscopy.

This work was supported by grant MT-9208 from the Medical Research Council (MRC) of Canada to R.A. Rachubinski. R.A. Rachubinski is an MRC Senior Scientist and an International Research Scholar of the Howard Hughes Medical Institute.

AOX

acyl-CoA oxidase

CAT

catalase

DSP

dithiobis(succinimidylproprionate)

DSS

disuccinimidyl suberate

FUM

fumarase

G6PDH

glucose-6-phosphate dehydrogenase

HSP

high-speed (200,000 g) pelletable

ICL

isocitrate lyase

LSP

low-speed (20,000 g) pelletable

20KgP

20,000 g pellet

20KgS

20,000 g supernatant

200KgP

200,000 g pellet

200KgS

200,000 g supernatant (cytosol)

MLS

malate synthase

PNS

postnuclear supernatant

PTS

peroxisomal targeting signal

THI

thiolase