Pex19p Binds Pex30p and Pex32p at Regions Required for Their Peroxisomal Localization but Separate from Their Peroxisomal Targeting Signals*

The assembly of proteins in the peroxisomal membrane is a multistep process requiring their recognition in the cytosol, targeting to and insertion into the peroxisomal membrane, and stabilization within the lipid bilayer. The peroxin Pex19p has been proposed to be either the receptor that recognizes and targets newly synthesized peroxisomal membrane proteins (PMP) to the peroxisome or a chaperone required for stabilization of PMPs at the peroxisomal membrane. Differentiating between these two roles for Pex19p could be achieved by determining whether the peroxisomal targeting signal (PTS) and the region of Pex19p binding of a PMP are the same or different. We addressed the role for Pex19p in the assembly of two PMPs, Pex30p and Pex32p, of the yeast Saccharomyces cerevisiae. Pex30p and Pex32p control peroxisome size and number but are dispensable for peroxisome formation. Systematic truncations from the carboxyl terminus, together with in-frame deletions of specific regions, have identified PTSs essential for targeting Pex30p and Pex32p to peroxisomes. Both Pex30p and Pex32p interact with Pex19p in regions that do not overlap with their PTSs. However, Pex19p is required for localizing Pex30p and Pex32p to peroxisomes, because mutations that disrupt the interaction of Pex19p with Pex30p and Pex32p lead to their mislocalization to a compartment other than peroxisomes. Mutants of Pex30p and Pex32p that localize to peroxisomes but produce cells exhibiting the peroxisomal phenotypes of cells lacking these proteins demonstrate that the regions in these proteins that control peroxisomal targeting and cell biological activity are separable. Together, our data show that the interaction of Pex19p with Pex30p and Pex32p is required for their roles in peroxisome biogenesis and are consistent with a chaperone role for Pex19p in stabilizing or maintaining membrane proteins in peroxisomes.

The synthesis of nuclear-encoded proteins begins in the cytosol. A protein that is destined for an organelle is directed from the cytosol to the organelle by a sequence or sequences within the protein that contain information essential for targeting the protein to the organelle. These sequences are recognized by receptors in the cytosol or at the target membrane. Having arrived at its specific organelle, the protein is then either translocated across the organelle membrane or inserted into the membrane lipid bilayer.
Peroxisomes are ubiquitous organelles that perform a variety of essential biochemical functions, notably in fatty acid metabolism. Mature peroxisomes usually appear as spherical structures in electron micrographs, with diameters of 0.5-1.0 m, and surrounded by a single membrane. All of the peroxisomal proteins are encoded by nuclear genes, synthesized on cytoplasmic polysomes, and posttranslationally targeted to the peroxisome. They are sorted to the peroxisome by specific peroxisomal targeting signals (PTS). 3 PTSs are recognized in the cytosol by their cognate receptors, which direct the targeting and docking of proteins at the peroxisome membrane. Most peroxisomal matrix proteins are targeted by PTS1, an extreme carboxyl-terminal tripeptide consisting prototypically of Ser-Lys-Leu, and its shuttling receptor, the peroxin (protein required for peroxisome assembly) Pex5p. Strikingly, Pex5p has been shown to enter the peroxisome matrix together with its cargo and to recycle back to the cytosol following dissociation from its cargo. A few matrix proteins are targeted by PTS2, an amino-terminal nonapeptide, and its cytosolic shuttling receptor, Pex7p, whereas a limited number of matrix proteins are targeted by largely uncharacterized internal PTSs (reviewed in Refs. [1][2][3][4][5][6]. The sorting of peroxisomal membrane proteins (PMPs) is less well understood than the sorting of matrix proteins but appears to be independent of matrix protein import (2)(3)(4)(5)(6). Although a limited number of PMPs first target the endoplasmic reticulum and then traffic to peroxisomes (7)(8)(9)(10)(11), most PMPs are inserted into the peroxisomal membrane directly from the cytosol (12)(13)(14)(15)(16). The direct insertion of PMPs into the peroxisomal membrane is facilitated by Pex19p, a peroxin that is localized predominantly to the cytosol and to a much lesser extent at the surface of peroxisomes (17)(18)(19). Pex19p interacts essentially with all PMPs via a consensus motif that is found at least once in a PMP (20,21). The interaction of Pex19p and a PMP could either prevent the aggregation of the PMP or serve to guide the PMP to the peroxisomal membrane. Accordingly, Pex19p has been proposed to function either as an import receptor for PMPs or as a chaperone acting in the assembly or stabilization of PMPs at the peroxisomal membrane (20,(22)(23)(24)(25)(26). Here, we address the role of Pex19p in the assembly of two PMPs, Pex30p and Pex32p, of the yeast Saccharomyces cerevisiae (27). Pex30p and Pex32p represent excellent candidates with which to study the role of Pex19p in PMP targeting or assembly at the peroxisomal membrane, because Pex30p and Pex32p control peroxisome number and size but are dispensable for peroxisome formation and function (27), whereas the absence or mislocalization of most PMPs results in cells that are unable to form functional peroxisomes (the classical pex phenotype). We show that although Pex30p and Pex32p interact with Pex19p in regions that do not overlap with their experimentally identified PTSs, Pex19p is still required to localize Pex30p and Pex32p to peroxisomes. Pex30p and Pex32p mutants that localize to peroxisomes but produce cells exhibiting the peroxisomal phenotypes of cells lacking these proteins demonstrate that the regions in Pex30p and Pex32p controlling their peroxisomal targeting and cell biological activity are separable. Our results show that Pex30p and Pex32p must interact with Pex19p to function in peroxisome biogenesis and are consistent with Pex19p acting as a chaperone to stabilize or maintain Pex30p and Pex32p at the peroxisomal membrane.
Plasmids-The plasmid pDsRed-PTS1 has been described (28). Genes to be overexpressed were amplified by PCR and cloned into the plasmid YEp13 (29). For overexpression, the PEX19 gene included 400 base pairs of upstream sequence and 347 base pairs of downstream sequence. The plasmid pGFP/HIS5 was constructed from the plasmid pProtA/HIS5 (30) by exchange of the sequence encoding protein A (prA) with sequence encoding an improved version of GFP (GFPϩ) from Aequoria victoria (31). The plasmid pGFP/NAT was constructed from pGFP/HIS5 by exchange of the HIS5 auxotrophic marker for the gene encoding resistance to the drug neurseothricin (NAT). pNLS-PEX19 and pNLS-GFP-PEX19 were constructed in pRS316 and contained sequence encoding either the nuclear localization signal of Mad1p (NLS)-Pex19p or NLS-GFP-Pex19p under the control of the PHO4 promoter.
Tagging Genes-Genes were genomically tagged with sequence encoding Staphylococcus aureus protein A or GFP by homologous recombination with a PCR-based integrative transformation of parental BY4742 or pex19⌬ haploid cells (32).
Construction of Yeast Strains Mutated in the PEX30 or PEX32 Gene-In-frame deletions of the regions of the PEX30 or PEX32 gene coding for putative Pex19p recognition sequences or constructs coding for fusions between a Pex19p recognition sequence and GFP were made by overlapping PCR followed by genomic integration of the mutant genes at the PEX30 or PEX32 locus of the haploid parental strain.
Microscopy-Strains synthesizing GFP chimeric proteins or transformed with the plasmid pDsRed-PTS1 were grown in synthetic minimal medium for 12 h and then incubated in YPBO medium for 8 h. Images were captured on a LSM510 META (Carl Zeiss) laser scanning microscope or on an Olympus BX50 microscope equipped with a digital fluorescence camera (Spot Diagnostic Instruments). The cells were processed for immunofluorescence microscopy (33) and electron microscopy (34).
Two-hybrid Analysis-Physical interactions between Pex19p and full-length or partial constructs of Pex30p or Pex32p were detected

Role of Pex19p in Sorting Peroxisomal Membrane Proteins
using the Matchmaker two-hybrid system (BD Biosciences Clontech). Chimeric genes were made by amplifying the sequences encoding Pex19p and full-length or partial constructs of Pex30p or Pex32p and ligating them in-frame and downstream of the DNA encoding the transcription-activating domain (AD) and the DNA-binding domain (BD) of the GAL4 transcriptional activator in the plasmids pGAD424 and pGBT9, respectively. Cells of S. cerevisiae strain HF7c were transformed simultaneously with a pGAD424-derived plasmid and a pGBT9-derived plasmid. Transformants were grown on synthetic minimal agar medium lacking tryptophan, leucine, and histidine. Growth on this medium indicates an interaction between two chimeric proteins. Subcellular Fractionation and Isolation of Peroxisomes-Subcellular fractionation of oleic acid-incubated cells was done as described (28) and involved the isolation of a postnuclear supernatant fraction and 20,000 ϫ g supernatant (20KgS) and pellet (20KgP) fractions enriched for cytosol and for peroxisomes and mitochondria, respectively. Peroxisomes were purified from the 20KgP fraction by isopycnic density centrifugation on Nycodenz gradients (28).
Antibodies-Antibodies to peroxisomal thiolase have been described (35). Protein A chimeras were detected by immunoblotting with affinity-purified rabbit antibodies against mouse IgG (Jackson).
Analytical Procedures-Extraction of nucleic acid from yeast lysates and manipulation of DNA were performed as described (36). Immunoblotting was performed using a wet transfer system (36), and antigen-antibody complexes in immunoblots were detected by enhanced chemiluminescence. Protein concentration was determined using a commercially available kit (Bio-Rad) and bovine serum albumin as standard.

Amino Acids 230 -250 of Pex30p and 159 -179 of Pex32p Function as
Peroxisomal Targeting Signals-Pex30p and Pex32p are two recently identified integral PMPs of S. cerevisiae that are not required for the formation of peroxisomes per se but that regulate the number and size of peroxisomes (27). Pex30p and Pex32p are 523 and 413 amino acids in length, respectively. To delineate sequences within Pex30p and Pex32p required for their targeting to peroxisomes, we constructed strains expressing genomically integrated constructs of Pex30p or Pex32p progressively truncated at their carboxyl termini and tagged at their carboxyl termini with either protein A for biochemical analysis or GFP for analysis by confocal microscopy. Subcellular fractionation demonstrated that full-length Pex30p or Pex32p preferentially associated with a 20KgP fraction enriched for peroxisomes and not a 20KgS fraction enriched for cytosol (Fig. 1). The truncated proteins Pex30(1-282)p and Pex30(1-250)p also preferentially associated with the 20KgP fraction.
Further truncations of Pex30p accumulated primarily in the 20KgS fraction (Fig. 1A), suggesting that information for targeting Pex30p to peroxisomes is present between amino acids 230 and 250. In the case of Pex32p, the truncated forms Pex32(1-203)p and Pex32(1-179)p, like full-length Pex32p, preferentially localized to the 20KgP fraction enriched for peroxisomes. Pex32p (1-159)p preferentially accumulated in the 20KgS fraction (Fig. 1B), suggesting that information for targeting Pex32p to peroxisomes lies between amino acids 159 and 179.
A carboxyl-terminal PTS1 is sufficient to direct a reporter protein to peroxisomes. A fluorescent chimera between Discosoma sp. red fluorescent protein (DsRed) and the PTS1 Ser-Lys-Leu has been shown to target to peroxisomes of S. cerevisiae (28,37). Genomically encoded chimeras of full-length and truncated forms of Pex30p and Pex32p linked to GFP at their carboxyl termini were localized in oleic acidincubated cells by fluorescence microscopy with DsRed-PTS1 to identify peroxisomes (Fig. 2, A and B). To analyze whether these 20-amino acid sequences are essential for the targeting of Pex30p and Pex32p to peroxisomes, we constructed strains expressing genomically integrated GFP fusions of Pex30p deleted for amino acids 230 -250 (Pex30(⌬230 -250)p) and of Pex32p deleted for amino acids 159 -179 (Pex32(⌬159 -179p) (Fig. 2C). Cells expressing Pex30(⌬230 -250)p showed both a punctate pattern of fluorescence that colocalized with the peroxisomal marker DsRed-SKL and substantial cytosolic fluorescence, suggesting that amino acids 230 -250 of Pex30p, although not essential for the targeting of Pex30p to peroxisomes, do significantly increase the efficiency of peroxisomal targeting of Pex30p. In contrast, cells expressing Pex32(⌬159 -179)p exhibited a diffuse pattern of cytosolic fluorescence, suggesting that amino acids 159 -179 are absolutely required for the targeting of Pex32p to peroxisomes.
To determine whether amino acids 230 -250 of Pex30p and 159 -179 of Pex32p were sufficient to target a reporter protein to peroxisomes, chimeric proteins consisting of these 20 amino acids (Pex30(230 -250)p and Pex32(159 -179)p) preceded by an initiating methionine residue and linked to GFP at their carboxyl termini were expressed from a genomic integration at the respective PEX30 and PEX32 gene loci. As expected, punctate fluorescence exhibited by DsRed-SKL demonstrated that functional peroxisomes were still assembled in these cells, because cells deleted for either the PEX30 or PEX32 gene have been shown to assemble functional peroxisomes (27). Both Pex30(230 -250)p and Pex32(159 -179)p showed extensive cytosolic fluorescence but also localized to punctate structures that colocalized with the peroxisomal marker DsRed-SKL (Fig. 2D). These results suggest that although amino acids 230 -250 of Pex30p and 159 -179 of Pex32p are sufficient to target a reporter protein to peroxisomes and can therefore be considered as bona fide PTSs, they function inefficiently as PTSs. That Pex30(230 -250)p and Pex32(159 -179)p should target inefficiently to peroxisomes is not unexpected, because these constructs lack a transmembrane domain (27) that has been shown to facilitate the anchoring of proteins within the peroxisomal membrane (21, 37-40, 42). were grown in YPD medium for 16 h, transferred to oleic acid-containing YPBO medium, and incubated for 8 h in YPBO medium, and their homogenates were subjected to differential centrifugation to yield a postnuclear supernatant (PNS), a 20KgS fraction enriched for cytosol, and a 20KgP fraction enriched for peroxisomes. An equal percentage of each subcellular fraction was separated by SDS-polyacrylamide gel electrophoresis and subjected to immunoblotting with affinity-purified rabbit antibodies against mouse IgG to detect protein A chimeric constructs.

The Pex19p Recognition Sites in Pex30p and Pex32p Do Not Overlap
Their Peroxisomal Targeting Signals-The peroxin Pex19p has been implicated in peroxisome membrane biogenesis and functions either as an import receptor for PMPs or as an assembly/disassembly factor, i.e. chaperone, for PMPs at the peroxisomal membrane (20,(22)(23)(24)(25)(26). Interactions between Pex19p and Pex30p or Pex32p were investigated by yeast two-hybrid analysis. Chimeric genes were constructed between the open reading frames of peroxins in-frame and downstream of sequences encoding either the AD or BD of the GAL4 transcriptional activator and transformed into S. cerevisiae strain HF7c for selective growth on agar plates lacking histidine (Fig. 3, B and C). Because fusion of Pex19p to the GAL4 BD resulted in the self-activation of transcription and growth of transformants on medium lacking histidine, only Pex19p fused to AD of GAL4 was used to probe the interactions of Pex19p with various peroxins. Chimeras of BD with all other peroxins did not show self-activation of transcription and growth of transformants on medium lacking histidine (data not shown). Pex19p exhibited a known interaction with the PMP Pex11p (21) (Fig. 3B). Pex19p interacted with both Pex30p and Pex32p (Fig. 3, B and C).
Recently, a consensus sequence for recognition of a protein by Pex19p has been defined (21). Two putative Pex19p recognition sequences were identified in Pex30p at amino acids 24 -38 and 97-111 (Fig. 3A). Two putative Pex19p recognition sequences were also identified at amino acids 93-107 and 135-149 in Pex32p (Fig. 3A). The PTSs identified in Pex30p and Pex32p do not overlap with the putative Pex19p recognition sequences in Pex30p and Pex32p. In the yeast two-hybrid system, Pex19p interacted with truncation mutants of Pex30p that did (Pex30(1-250)p) or did not (Pex30(1-229)p) contain the PTS (Fig. 3B). These results show that the Pex30p PTS is not required for Pex19p binding. Pex19p interacted with Pex32(1-179)p, which contains the PTS, but not with Pex32(1-124)p, which does not contain the PTS or the second putative Pex19p recognition sequence at amino acids 135-149 (Fig. 3B). Thus, the first putative Pex19p recognition sequence in Pex32p at amino acids 93-107 apparently does not function in interacting with Pex19p.
Substitution of a proline for any amino acid within the consensus Pex19p recognition sequence has been shown to disrupt the interaction of the PMPs Pex11p and Pex13p with Pex19p (21). Introduction of proline within the Pex19p-binding region likely disrupts its ␣-helical conformation that is thought to promote the association of a partner protein with Pex19p. Substitution of proline for isoleucine 31 (Pex30(I31P)p) in the first Pex30p putative Pex19p recognition sequence marginally reduced the interaction between Pex30p and Pex19p (Fig. 3C). However, substitution of proline for leucine 104 (Pex30(L104P)p) in the second Pex30p putative Pex19p recognition sequence abolished the interaction between Pex30p and Pex19p but did not abolish the known interaction between Pex30p and Pex29p (27), showing that the lack of an interaction between Pex30(L104P)p and Pex19p was not due to nonexpression of Pex30(L104P)p (Fig. 3C). Substitution of proline for leucine 142 (Pex32(L142P)p) in the first putative Pex19p recognition sequence of Pex32p abolished the interaction between Pex32p and Pex19p (Fig. 3C). Again, this lack of interaction between Pex32(L142P)p was not due to nonexpression of Pex32(L142P)p, because the known interaction of Pex32p with Pex28p (27) was maintained by the Pex32(L142P)p mutant (Fig. 3C). Because both Pex30(L104P)p and Pex32(L142P)p still contain their experimentally identified PTSs, these PTSs do not bind Pex19p. As is the case for the recognition of Pex11p and Pex13p by Pex19p, Pex19p recognizes Pex30p and Pex32p at one site only, and these sites do not overlap with the PTSs of these proteins.
Pex19p Functions in the Subcellular Localization of Pex30p and Pex32p-We have shown that Pex19p interacts with both Pex30p and Pex32p; however, the Pex19p-binding regions in these proteins do not overlap with their PTSs. We therefore asked the question of whether Pex19p has a role in localizing Pex30p or Pex32p within the cell. To address this question, we took advantage of the observation in mammalian cells that attachment of NLS to Pex19p redirects Pex19p to the nucleus, which in its turn redistributes a Pex19p-targeted protein from peroxisomes to the nucleus (25,43). We therefore wanted to determine whether a nuclear-targeted chimera of Pex19p, NLS-Pex19p, containing the NLS of the spindle assembly checkpoint protein Mad1p (44,45) could redirect the fluorescent chimeras Pex30p-GFP and Pex32p-GFP to the nucleus in cells deleted for the PEX19 gene. NLS-GFP-Pex19p localized to the nucleus in pex19⌬ cells, as expected (Fig. 4, A and B). NLS-Pex19p was able to direct Pex30p-GFP (Fig. 4A) and Pex32p-GFP (Fig. 4B) to the outer surface of the nucleus, most probably the nuclear envelope. Complete import of Pex30p-GFP and Pex32p-GFP into the nucleus may have failed because of steric or conformational incompatibility for nuclear import of NLS-Pex19p complexed with Pex30p-GFP or Pex32p-GFP. Moreover, GFP chimeras of Pex30(L104P)p and Pex32(L142P)p, which are abrogated in their interaction with Pex19p, localized to punctate structures when genomically expressed in wildtype BY4742 cells (Fig. 4C). These punctate structures resemble the punctate structures targeted by Pex30p-GFP and Pex32p-GFP in pex19⌬ cells (Fig. 4C). However, these structures are not peroxisomes, because peroxisomes are absent in pex19⌬ cells (18,20,46). Subcellular fractionation confirmed that these punctate structures are not peroxisomes and are of lighter density than peroxisomes, which contain the enzyme thiolase (Fig. 4, D and E). Our results show that although Pex19p does not bind the PTSs of Pex30p and Pex32p, its interactions with Pex30p and Pex32p do have a role in localizing these proteins to peroxisomes in the cell.
Mapping Regions of Pex30p and Pex32p that Function in Peroxisome Biogenesis-We next sought to identify regions of Pex30p and Pex32p required for their function in peroxisome biogenesis. Cells deleted for the PEX30 gene exhibit increased numbers of peroxisomes, whereas cells deleted for the PEX32 gene exhibit enlarged peroxisomes (27). Analysis by electron microscopy showed that any deletion of Pex30p and Pex32p at their carboxyl termini resulted in the peroxisomal phenotypes exhibited by cells of the respective pex30⌬ and pex32⌬ mutants, irrespective of their still being targeted to peroxisomes. Cells genomically expressing Pex30(1-282)p (Fig. 5B) or Pex30(1-250)p (Fig.  5C) exhibited the increased numbers of peroxisomes exhibited by pex30⌬ cells (Fig. 5A) as compared with wild-type BY4742 cells (Fig. 5F), even though both Pex30(1-282)p and Pex30(1-250)p are targeted to peroxisomes (Figs. 1A and 2A). Likewise, Pex32(1-307)p (Fig. 6B), Pex32(1-203)p (Fig. 6C), and Pex32(1-179)p (Fig. 6D), although targeted to peroxisomes (data not shown and Figs. 1B and 2B), exhibited the enlarged peroxisome phenotype of pex32⌬ cells (Fig. 6A). Thus, the functioning of both Pex30p and Pex32p in peroxisome biogenesis requires that their carboxyl termini be intact and is distinct from their ability to be targeted to peroxisomes. Interaction of Pex30p or Pex32p with Pex19p is important for the function of these peroxins in peroxisome biogenesis, because cells genomically expressing point mutant forms of Pex30p and Pex32p that are unable to interact with Pex19p exhibit the mutant peroxisome phenotype of cells deleted for their respective genes (Figs. 5E and 6E).

DISCUSSION
The assembly of proteins at the peroxisomal membrane is a multistep process requiring their recognition in the cytosol, their targeting to the peroxisomal membrane, their insertion into that membrane, and their stabilization within the lipid bilayer. Although there is a general agreement that Pex19p is required for the assembly of the peroxisomal mem-  (21). Two amino acid sequences in each of Pex30p and Pex32p that conform to the consensus sequence for Pex19p recognition are presented, along with their positions within the primary sequence of Pex30p or Pex32p. B, yeast two-hybrid analysis. Full-length genes encoding Pex11p, Pex30p, and Pex32p, as well as genes encoding truncated forms of Pex30p and Pex32p, were fused to the GAL4 binding domain (GAL4-BD) in the vector, pGBT9. The resulting plasmids were cotransformed into S. cerevisiae strain HF7c with a pGAD424derived plasmid expressing a fusion gene between the GAL4 activation domain (GAL4-AD) and PEX19. Transformants were tested for evidence of protein-protein interaction as shown by growth on agar medium lacking histidine. No growth was observed when the parental vector pGBT9 (-) was cotransformed with the plasmid expressing GAL4-AD-Pex19p. C, yeast two-hybrid analysis shows greatly reduced interaction between Pex19p and Pex30p or Pex32p containing proline substitutions in their second Pex19p consensus recognition sequences. Genes encoding Pex30p with a isoleucine to proline substitution at amino acid position 31 within the first Pex19p consensus recognition sequence (Pex30(I31P)p), Pex30p with a leucine to proline substitution at amino acid position 104 within the second Pex19p consensus recognition sequence (Pex30(L104P)p), Pex32p with a leucine to proline substitution at amino acid position 142 within the second Pex19p consensus recognition sequence (Pex32(L142P)p), full-length Pex30p, and fulllength Pex32p were fused to the GAL4-binding domain (GAL4-BD) in the vector, pGBT9. The resulting plasmids were cotransformed into S. cerevisiae strain HF7c with a pGAD424-derived plasmid expressing a fusion gene between the GAL4 activation domain (GAL4-AD) and PEX19, PEX28, or PEX29. Expression of the proline substitution constructs Pex30(L104P)p and Pex32(L142P)p was confirmed by positive two-hybrid interactions with the known respective partners of Pex30p and Pex32p, Pex29p and Pex28p (27).
brane, there is still considerable debate as to whether Pex19p functions in PMP targeting, i.e. it functions as a receptor for PMPs, or whether it functions as a chaperone, aiding in the assembly or stabilization of PMPs at the peroxisomal membrane (20,(22)(23)(24)(25)(26). Determining whether the PTS and Pex19p recognition sequence of a PMP are the same or different would help to differentiate between the two proposed roles for Pex19p in peroxisomal membrane assembly. Here, we have addressed the role of Pex19p with respect to two recently identified PMPs, the peroxins Pex30p and Pex32p, of S. cerevisiae. Most investigations regarding the role of Pex19p in peroxisomal membrane assembly have had to rely on mutation of a PMP to identify sequences for peroxisomal targeting or Pex19p binding but whose absence in peroxisomes leads to a severe pex phenotype in which peroxisomes fail to assemble. This failure of cells to form functional peroxisomes because of the mislocalization of a PMP is a major drawback in attempts to study the role of Pex19p in peroxisomal membrane assembly. Pex30p and Pex32p represent excellent PMPs with which to define PTSs and Pex19p-binding sites and potentially differentiate between them, because they are dispensable for peroxisome formation per se, as cells lacking one or both of these proteins can still form functional bona fide peroxisomes (27).
Microscopical and biochemical analysis of cells expressing deletion mutants of Pex30p and Pex32p identified a PTS between amino acid residues 230 and 250 in Pex30p and 159 and 179 in Pex32p. However, these PTSs were not sufficient to target a reporter protein to peroxisomes. This was not unexpected, because PTSs for PMPs have been shown to require additional amino acids that constitute a transmembrane domain for efficient targeting to peroxisomes (21, 38 -42). Whether Pex30p and Pex32p have PTSs additional to the ones we have identified remains unknown. Recent evidence has suggested that there could be multiple, nonredundant targeting modules on PMPs and that they could function as independent PTSs in a cell depending on its metabolic status (47).
A consensus sequence for recognition by Pex19p has recently been determined (21). Pex30p has two sequences that conform to this sequence at amino acids 24 -38 and 97-111, whereas Pex32p also has two Pex19p consensus recognition sequences at amino acids 93-107 and 135-149. Yeast two-hybrid analysis showed that Pex19p interacts with both Pex30p and Pex32p and that in both cases Pex19p binding was dependent on the second, more carboxyl-terminal consensus recognition site. Importantly, the sequences recognized by Pex19p in both Pex30p and Pex32p are different from the sequences experimentally demonstrated to function as PTSs in these two PMPs. These findings are inconsistent with Pex19p functioning as a peroxisomal targeting receptor for Pex30p and Pex32p.
If the PTSs of Pex30p and Pex32p do not correspond to the sites recognized by Pex19p, does Pex19p have a role in localizing Pex30p and Pex32p to peroxisomes? In pex19⌬ cells, Pex30p and Pex32p are targeted to punctate structures that are not peroxisomes, because peroxisomes are absent in pex19⌬ cells (18,20,46). Cells expressing Pex30(L104P)p and Pex32(L142P)p, which are mutated in their Pex19p recognition sites and abrogated in their interaction with Pex19p, are also mistargeted to small punctate structures that do not correspond to peroxisomes as demonstrated by subcellular fractionation. Moreover, mutations in Pex30p and Pex32p that disrupt their interaction with Pex19p would be predicted to give rise to the peroxisomal phenotype exhibited by the respective gene deletion mutant cells, and this was indeed shown by electron microscopy to be the case. These findings, combined with the observation that a chimera of Pex19p and the NLS of the nuclear protein Mad1p efficiently targeted the nuclei of yeast cells and directed GFP-tagged variants of Pex30p and Pex32p to the nuclear surface, presumably the nuclear envelope, are consistent with Pex19p having an impor-tant role in localizing Pex30p and Pex32p to peroxisomes, even though its interaction with these two PMPs is independent of their PTSs.
pex30⌬ and pex32⌬ cells show increased numbers of peroxisomes and enlarged peroxisomes, respectively, as compared with wild-type cells (27). Cells expressing forms of Pex30p and Pex32p truncated at their carboxyl termini but still containing their PTSs and Pex19p-binding sites exhibited the abnormal peroxisomal phenotypes of their respective gene deletion strains. Thus, domains of Pex30p and Pex32p required for their functioning in peroxisome biogenesis are separable from those required for peroxisomal targeting and Pex19p binding. It is interesting to note that Pex30p, Pex32p, and the related family member Pex31p contain conserved dysferlin domains at their carboxyl termini. Dysferlin domains were identified in dysferlin, a skeletal muscle fiber membrane protein whose deficiency results in specific forms of muscular dystrophy (48,49). What function the dysferlin domain plays in proteins and what, if any, role it would have in Pex30p, Pex31p, and Pex32p are not known at this time, although it has been proposed that dysferlin domains are calcium-sensitive binders of phospholipid (50).
In summary, we have shown that Pex19p interacts with Pex30p and Pex32p in regions that do not overlap with their experimentally identified PTSs. Nevertheless, interaction of Pex30p and Pex32p with Pex19p is still necessary for Pex30p and Pex32p localization to peroxisomes. Pex30p and Pex32p mutants that localize to peroxisomes but produce cells that exhibit the peroxisomal phenotypes of pex30⌬ and pex32⌬ cells show that the regions in Pex30p and Pex32p that control their peroxisomal targeting and their cell biological activity are separable. Altogether, our data show that the interaction of Pex19p with Pex30p and Pex32p is required for their roles in peroxisome biogenesis and are consistent with a chaperone role for Pex19p in stabilizing or maintain-  ing Pex30p and Pex32p in peroxisomes rather than in targeting them to peroxisomes.