The N-domain of Pex22p Can Functionally Replace the Pex3p N-domain in Targeting and Peroxisome Formation*

Pex3p is a central component of the import machinery for peroxisomal membrane proteins (PMPs) that can reach peroxisomes via the endoplasmic reticulum (ER) and even trigger de novo peroxisome formation from the ER. Pex19p is the import receptor for type I PMPs, whereas targeting of type II PMPs, of which Pex3p so far represents the only species, does not require Pex19p. Pex3p possesses two domains with distinct function: a short N-terminal domain, which harbors the information for peroxisomal (and ER) targeting, and a C-terminal domain, which faces the cytosol and serves as a docking site for Pex19p, thereby delivering newly synthesized PMPs to the peroxisome. Here we show that the N-terminal domain of Pex3p can be functionally replaced by the N-terminal peroxisomal membrane targeting signal (mPTS) of Pex22p, a supposedly unrelated component of the import machinery for peroxisomal matrix proteins. An exchange of the mPTS of Pex22p by that of Pex3p likewise fully preserved the function of Pex22p. Neither of the two mPTS interacted with Pex19p, and in the absence of Pex19p, colocalization of Pex3p and Pex22p was observed, indicating that also Pex22p is targeted to peroxisomes by a type II mPTS. When a type I mPTS was hooked to the C-terminal domains of Pex22p and Pex3p, function was retained in the case of Pex22p and in part even for Pex3p. The C-terminal domain of Pex3p thus contains the relevant information required for de novo peroxisome formation, thereby challenging the concept of the N terminus of Pex3p being key in that process.

Peroxisome biogenesis requires a set of proteins called peroxins, most of which accomplish the faithful import of matrix proteins (1)(2)(3). Two types of targeting signals are known to direct proteins to the peroxisomal lumen: a C-terminal peroxisomal targeting signal type 1 (PTS1) 2 comprising the C-terminal tripeptide SKL and conservative variants thereof, and an N-terminal PTS2. The import of peroxisomal membrane proteins (PMPs) follows a distinct route and depends on Pex3p and Pex19p in all of the species analyzed (4) and on Pex16p in mammals (5,6).
The majority of PMPs are recognized by Pex19p during or early after synthesis in the cytosol. Pex19p binding not only protects these hydrophobic proteins from aggregation but also delivers them to the peroxisomal membrane (7)(8)(9)(10). Docking of the cargo-loaded Pex19p receptor at the peroxisomal membrane occurs via the C-terminal cytoplasmic domain of Pex3p (10 -13). In a subsequent, poorly understood step, PMPs are released from Pex19p for insertion into the peroxisomal membrane. Although some controversy exists in the literature on the nature of a peroxisomal membrane protein targeting signal (mPTS), it requires one or more transmembrane spans and a targeting-specific sequence that is identical to the Pex19pbinding site in at least a subset of PMPs (4,10,11).
In mutants with a specific block in matrix protein import, PMPs are normally imported into so-called peroxisomal remnants. In contrast, in the absence of the peroxins required for PMP import, i.e. Pex3p, Pex16p, or Pex19p, peroxisomes are thought to be truly missing (14). Nonetheless, upon reintroduction of the respective missing gene, peroxisome formation can be restored in these strains (5,15,16). Recent work highlights the endoplasmic reticulum (ER) as a cornerstone in this process (17)(18)(19), but other scenarios such as the existence of a protoperoxisome still appear conceivable (20,21).
Consent exists on Pex3p being essential for PMP import, and it is widely considered as an early peroxin that is able to trigger de novo peroxisome formation by virtue of its N-terminal mPTS, which comprises ϳ45 amino acids (22)(23)(24). In mammals targeting of Pex3p turned out to be Pex19p-independent, whereas all other PMPs tested did require Pex19p (10). In the yeast Saccharomyces cerevisiae, Pex3p was reported to reside in the ER in the absence of Pex19p, poised to form mature peroxisomes upon reintroduction of Pex19p (17).
That Pex3p sorts to peroxisomes via the ER also in wild-type cells is supported by a transient localization of conditionally expressed Pex3p to the ER (17), as well as by a set of experiments using a Pex3p variant with an appended ER-targeting signal (25). Vesicular transport of Pex3p to peroxisomes would also explain how peroxisomes are supplied with membrane lipids (18,26). Notably, however, none of these experimental setups monitored endogenous Pex3p, and as a consequence, a direct insertion of Pex3p from the cytosol in the presence of peroxisomes cannot be totally ruled out. Evidence for a direct targeting was indeed reported for the Arabidopsis thaliana (27) and human (10) orthologs of Pex3p.
Pex3p possesses two domains with distinct function: the short N-terminal domain harbors the mPTS including a short luminal domain and one transmembrane domain (TMD) and is also thought to create the basis for the formation of peroxisomes, and the C-terminal domain faces the cytosol and serves as a docking site for Pex19p, thereby delivering newly synthesized PMPs to the peroxisome (4). Pex22p is involved in a late step of matrix protein import and does not play a role in the biogenesis of the peroxisomal membrane (28). Although functionally distinct, the topology of Pex22p resembles that of Pex3p, with a single TMD, a short N-terminal region facing the peroxisomal lumen, and a large C-terminal domain exposed to the cytosol. This latter domain provides the docking site for the ubiquitin-conjugating enzyme Pex4p (28), which accomplishes the export of the PTS1 receptor Pex5p via ubiquitination of a conserved cysteine residue (29 -31). In Pichia pastoris, targeting of Pex22p depended on its N terminus (28), but it is currently unknown whether Pex19p is directly involved in the recruiting of Pex22p to the peroxisomal membrane.
In this study we compared the targeting of Pex3p with that of Pex22p and examined the role of Pex19p in this process. The mPTS sequences were mutually exchanged and analyzed for their ability to restore function in pex22⌬ and pex3⌬ mutants. The gathered data not only led to the identification of a second, functionally unrelated PMP that is targeted to peroxisomes via a Pex19p-independent type II mPTS but also challenge the concept of the N terminus of Pex3p being key in peroxisome formation.
Plasmids-The plasmids and oligonucleotides used are listed in supplemental Tables S1 and S2, respectively. Unless otherwise stated, genes and gene fragments were amplified from genomic DNA of the S. cerevisiae wild-type strain UTL-7A. The respective PCR primers, restriction sites, and target vectors used for cloning are also listed in supplemental Table S1. The sequence of all PCR-generated fragments was verified by automated sequencing (MWG Biotech, Ebersberg, Germany). To clone pMDE24/2, DsRed.T1 was amplified from pMDE21 (37) by using PCR primers RE1685/1686 and inserted into pMDE23 (38) after SacI/SpeI digestion (pMDE24/1). Following SacI/ KpnI digestion of pMDE24/1, the entire MET25 Prom -DsRed.T1-Ant1-CYC1 Term cassette was transferred into appropriately digested pRS414. Plasmid pAH78 was created by replacing ANT1 with CYB5; pHPR346 was digested with SpeI/ XhoI and the isolated CYB5 was transferred to the similarly treated pMD24/2. To construct pAH54, the isolated MET25 promoter was transferred from pUG35 to pRS414 after SacI/ SpeI digestion of both plasmids. The CYC1 terminator was lifted from pUG36 by XhoI/KpnI digestion and inserted into the similarly treated pAH54 resulting in pAH76. DsRed.T1 was amplified from pUG34DsRed.T1-SKL (38) by using PCR primers RE 1715/1716 and inserted into EcoRI/XhoI-digested pAH76 resulting in pAH77. PCR amplification of PEX22 by virtue of primer pair RE1708/1709 and subsequent insertion into BamHI/EcoRI-digested pAH77 finally led to pAH73.
Fluorescence and Electron Microscopy-Live cell imaging was carried out essentially as described (11). Prior to inspection, yeast cells were grown on plates containing ethanol as sole carbon source. The cells were analyzed for GFP and DsRed by virtue of a Zeiss Axioplan 2 microscope with a Zeiss ␣ Plan-FLUAR 100ϫ/1.45 oil objective. The micrographs were recorded with an Axiocam MR digital camera and processed with Axio Vison 4.2 software (Zeiss, Jena, Germany). Electron microscopy was carried out as described (33,39). To induce peroxisome proliferation, the cells were incubated in oleic acidcontaining medium over night. The cells were fixed in 2% paraformaldehyde and 0.5% glutaraldehyde, stained with 1.5% KMnO 4 and 1% uranyl acetate, and inspected with a Philips EM 300 transmission electron microscope (Philips Electron Optics, Eindhoven, The Netherlands).
Subcellular Fractionation by Sucrose Density Gradient Centrifugation-A post-nuclear supernatant (total protein content, 10 mg) prepared from cells that had been induced overnight in Rytka medium was loaded on top of a linear sucrose gradient (20 -54% w/w) and subjected to centrifugation at 38,000 ϫ g in a vertical rotor (SV-288, Sorvall RCB5; DuPont, Bad Nauheim, Germany) for 90 min. Subsequently, the gradients were fractionated from the bottom to the top, and the resulting fractions were analyzed for catalase activity (EC 1.11.1.6) using an established method (40). In addition, aliquots of each fraction were precipitated, resolved in standard Laemmli SDS buffer, and subjected to immunoblot analysis to determine the distribution of peroxisomal Pex14p and mitochondrial porin. Polyclonal antibodies directed against Pex14p (41) and porin (42) were described previously.
Yeast Two-hybrid Assays-The assay was carried out as described (11), based on the method of Fields and Song (43). PEX3, PEX22, and truncations thereof were fused to the Gal4p DNA-binding domain in vector pPC86, whereas PEX4 and PEX19 were fused to the transcription-activation domain of Gal4p in pPC97 (44). Appropriate plasmid combinations were cotransformed into yeast strain PJ69-4A and selected on synthetic dextrose plates lacking tryptophan and leucine. Transformed PJ69-4A was tested for concomitant histidine and adenine prototrophy by growth on selective plates lacking leucine, tryptophan, histidine, and adenine.
Miscellaneous-Purification of GST-Pex19p and peptide scan analysis were carried out as described previously (11).

Topology and Targeting Signals Are Similar for ScPex3p
and ScPex22p-To learn more about the topogenesis of Pex3p in S. cerevisiae, we compared its targeting with that of Pex22p. The latter PMP bears a similar topology ( Fig. 1A) but is not involved in membrane biogenesis (28). A C-terminal GFP fusion of Pex22p gave rise to a punctate staining pattern that was congruent with that of PTS2-DsRed (Fig. 1B), a synthetic peroxisomal marker protein (36). Peroxisomal localization was also visible for the N-terminal 35 amino acids of Pex22p, although some perinuclear ER staining was additionally discernible (Fig. 1C). The complementary fragment comprising the large cytosolic domain of Pex22p (amino acids 36 -180) led to a diffuse staining pattern typical for a cytosolic localization and expected for a fragment that lacks a TMD (Fig. 1D).
Similarly, and consistent with published work (22), peroxisomal targeting of Pex3p was observed for the full-length protein (Fig. 1E) and its N-terminal 45 amino acids (Fig. 1F), whereas the complementary C-terminal fragment (amino acids 46 -441) was localized to the cytosol (Fig. 1G). Thus, the targeting signals of Pex3p and Pex22p are comprised of the extreme N termini and include the single TMD (Fig. 1A), whereas the large C-terminal domains are not required for targeting.

The mPTS of Pex3p and Pex22p
Do Not Interact with Pex19p-Common among the mPTS of several PMPs is the presence of at least one Pex19p-binding site, which is typically composed of a linear sequence of ϳ15 amino acids enriched for hydrophobic and basic residues (10,11). Such sites are readily detectable by peptide scan analysis (11,45); however, purified Pex19p bound neither to overlapping 20-mer peptides covering the first 50 amino acids of Pex3p or Pex22p (i.e. their mPTS) nor to 15-mer peptides representing the entire proteins (not shown).
A yeast two-hybrid assay was used to test for Pex19p interaction in vivo. As shown in Fig. 2 (A and B), neither the mPTS of Pex3p (amino acids 1-45) nor that of Pex22p (amino acids 1-35) showed an interaction with Pex19p. The strong interaction of full-length Pex3p with Pex19p reflects the established function of Pex3p as a docking protein for Pex19p and is apparently not linked to its targeting, similar to observations made in mammals (10). An interaction of Pex19p with Pex22p was not observed. That the fusion proteins employed were correctly folded can be inferred from the interaction of full-length Pex22p (amino acids 1-180) and its cytosolic domain (amino acids 36 -180) with Pex4p, a known interaction partner of Pex22p (Fig. 2C). Pex3p did not interact with Pex4p, confirming that the C termini of Pex3p and Pex22p serve distinct functions.
The Pex19p interaction data suggested that targeting of Pex22p and Pex3p is independent of Pex19p and thus differs markedly from that of other PMPs. Notably, the mPTS sequences of Pex3p and Pex22p are not only similarly positioned but also quite similar in sequence; within the first 31 amino acids, ϳ42% of the residues are identical (Fig. 2D).
Pex3p and Pex22p Accumulate in the Same Compartment in the Absence of Pex19p-Localization of Pex3p and Pex22p was analyzed in pex19⌬ cells lacking peroxisomes. In this mutant strain, Pex3p is believed to accumulate in distinct regions of the ER (17,18) and poised to form peroxisomes, whereas Pex19pdependent PMPs are mislocalized or degraded. Labeling of the ER with Cyb5p-DsRed revealed that the majority of Pex3p and Pex22p was localized to punctate structures within or in vicinity to the ER (Fig. 3, A and B) (18). Because of the similar staining pattern of Pex3p and Pex22p, a DsRed fusion of Pex22p was coexpressed with Pex3p-GFP in a pex19⌬ mutant. A clear congruent staining was discernible, demonstrating colocalization of the two proteins (Fig. 3C). These data support the hypothesis that Pex3p and Pex22p share the same targeting pathway.
The mPTS of Pex3p and Pex22p Are Interchangeable-The semblance in mPTS might point to a common import pathway for Pex3p and Pex22p. We therefore tested whether an  1C). From these results it was concluded that the mutual exchange of the mPTS did not compromise the targeting of Pex3p and Pex22p.
The Pex3p N -Pex22p C Chimera Can Functionally Complement a pex22⌬ Mutant-We next examined the potential of the Pex3p-Pex22p chimeras to functionally replace Pex22p. Because of its essential role in Pex4p recruitment to the peroxisomal membrane, matrix protein import is disabled in the absence of Pex22p, which can be visualized by a cytosolic distribution of the synthetic peroxisomal marker protein DsRed-SKL (Fig. 4B). Membrane protein targeting, on the other hand, is not affected in the pex22⌬ mutant, as demonstrated by a congruent, punctate staining pattern for Pex3p and Ant1p (Fig. 4C). Expression of Pex22p restored peroxisomal localization of the matrix protein marker (Fig. 4D), whereas Pex3p (Fig. 4B) and Pex22p 36 -180 (Fig. 4E) failed to do so. Interestingly, the Pex22p chimera harboring the mPTS of Pex3p and the cytoplasmic domain of Pex22p clearly provoked a relocation of DsRed-SKL into peroxisomes (Fig. 4F).
The ability of Pex3p 1-45 -Pex22p 36 -180 to complement the import defect of a pex22⌬ mutant was also analyzed by a test for growth on solid agar medium containing oleic acid as the sole carbon source. Because in yeast, peroxisomes are essential for the breakdown of fatty acids, cells devoid of functional peroxisomes as in the pex22⌬ mutant cannot utilize oleic The diffuse staining of DsRed-SKL indicated a cytosolic localization, whereas DsRed-Ant1p exhibited a punctate staining, congruent with that of Pex3p-GFP. D-F, test for complementation of the pex22⌬ matrix protein import defect by the Pex3p N -Pex22p C chimera. Pex22⌬ strains coexpressing C-terminal GFP fusions of Pex22p (D), the cytoplasmic domain of Pex22p (E), and the Pex3p N -Pex22p C chimera (F) together with DsRed-SKL were inspected for their intracellular localization by fluorescence microscopy. Pex22p and the Pex3p N -Pex22p C -chimera clearly reconstituted matrix protein import, evident by a punctate staining pattern for DsRed-SKL (D and F). Bar, 2 m. G, test for complementation of the pex22⌬ growth defect on oleic acid by Pex22p C chimeras. The indicated strains were spotted in serial 1:10 dilutions on plates containing oleic acid as sole carbon source and incubated at 30°C for 5 days. Strong colony growth accompanied by halo formation indicates proper fatty acid utilization. Pex22p as well as the Pex3p N -Pex22p C chimera clearly restored growth. The expression of a Pex13p mPTS -Pex22p C chimera also led to a restoration of growth of the pex22⌬ mutant (see also Fig. 6).
acid and as a consequence neither grow efficiently nor form the typical halo on such plates. Halo formation was restored upon expression of Pex22p and Pex3p 1-45 -Pex22p 36 -180 but not of Pex22p 1-35 -Pex3p 46 -441 (Fig. 4G). The combined results indicated that Pex3p 1-45 -Pex22p 36 -180 had inserted into the membrane with the correct topology. Furthermore, it demonstrated that the Pex22p-specific function is contained within its cytoplasmic domain, whereas its N-terminal domain is required for targeting and insertion into the peroxisomal membrane.
The Pex22pN-Pex3pC Chimera Restores Peroxisome Formation in pex3⌬ Cells-In the absence of Pex3p, peroxisomes are considered to be missing, yet the reintroduction of Pex3p allows their restoration (15)(16)(17)46). The N terminus of Pex3p has been implicated as being the key component in this process because it promotes the proliferation of precursor vesicles that eventually mature into peroxisomes upon expression of full-length Pex3p (22,23). We therefore reasoned that expression of Pex3p 1-45 -Pex22p 36 -180 in pex3⌬ cells should allow the formation of such vesicles but not of mature peroxisomes, whereas the Pex22p 1-35 -Pex3p 46 -441 chimera should be mislocalized in the absence of peroxisomes. Upon coexpression of both chimeras, however, the vesicles formed by Pex3p 1-45 -Pex22p 36 -180 might suffice as a target membrane for the insertion of the Pex22p 1-35 -Pex3p 46 -441 chimera, which in turn would serve as a docking site for cargo-loaded Pex19p. Bulk insertion of PMPs would then allow the formation of peroxisomes.
Restoration of peroxisome formation in pex3⌬ cells was achieved by episomal expression of Pex3p, again indicated by the punctate fluorescence pattern of the peroxisomal marker protein (Fig. 5B). By contrast, expression of the N termini of Pex3p (Fig. 5C) and Pex22p (Fig. 5E) or the cytoplasmic domain of Pex3p (Fig. 5D) gave rise to a diffuse staining pattern of DsRed-PTS1 that is typical for cells with a matrix protein import defect. Opposite to our expectation, expression of Pex22p 1-35 -Pex3p 46 -441 was clearly proficient in restoring matrix protein import in cells lacking Pex3p (Fig. 5F). Furthermore, electron microscopy revealed the existence of normally sized and electron dense peroxisomes when Pex22p 1-35 -Pex3p 46 -441 was expressed in pex3⌬ cells but not in the pex19⌬ control strain (Fig. 6A).
To gain further evidence for the formation of peroxisomes by Pex22p 1-35 -Pex3p 46 -441 , a post-nuclear supernatant of the pex3⌬ transformant was subjected to density gradient centrifugation. A clear peak at a density of 1.22 g/cm 3 was discernible for catalase that coincided with the peroxisomal marker PMP Pex14p, whereas in the pex3⌬ control strain, catalase activity was only recovered from the cytosol-representing light fractions (Fig. 6B). A growth test on solid agar plates containing oleic acid as sole carbon source finally demonstrated that peroxisomes of the pex3⌬ strain expressing Pex22p N -Pex3p C represented fully functional peroxisomes. The strain grew similarly to the wild-type strain and the pex3⌬ mutant expressing full-length Pex3p, whereas Pex3p N -Pex22p C failed to complement the growth defect of the pex3⌬ strain (Fig. 6C).
Complementation of pex22⌬ with a Fusion of a Pex19p-dependent mPTS and Pex22p C -In light of the similarity of the N termini of Pex3p and Pex22p, the observed rescue could be limited to targeting signals of the Pex3p type, i.e. N-terminal and independent of Pex19p. We therefore analyzed whether the exchange with a Pex19p-dependent targeting signal would likewise lead to functional proteins. To this end, the C-terminal domains of Pex22p and Pex3p were attached to the mPTS of Pex13p so that they will face the cytosol (Fig. 7A). According to current knowledge, the mPTS of Pex13p should allow a faithful targeting to peroxisomal remnants existing in pex22⌬ cells, whereas in a pex3⌬ strain, mistargeting should occur because of the absence of a peroxisomal membrane.
The Pex13p 166 -310 -Pex22p 36 -180 chimera localized to punctate structures and occasionally to the ER (Fig. 7, B and C). It was indeed able to complement the import defect of the pex22⌬ mutant because coexpressed DsRed-SKL was found in the same punctate structures (Fig. 7,  B and C). In a pex3⌬pex19⌬ strain lacking peroxisome remnants, mPTS Pex13p -Pex22p C localized to the ER, and the peroxisomal marker remained cytosolic (Fig. 7D). Electron microscopy revealed the appearance of peroxisome clusters in the transformed pex22⌬ strain that were absent in the nontransformed strain (Fig. 7E). Importantly, growth on oleic acid was also restored by the Pex13p 166 -310 -Pex22p 36 -180 fusion protein (Fig.  4G). The effect of this chimera corroborated the conclusion that the cytoplasmic domain possesses all that is required from Pex22p to achieve matrix protein import. It also showed that Pex22p can be directed to peroxisomes via the Pex19p-dependent pathway and thereby retain function.
The mPTS Pex13p -Pex3p C Fusion Can Provoke Peroxisome Formation in pex3⌬ Cells-Expression of the Pex13p 166 -310 -Pex3p 46 -441 chimera in a wild-type strain revealed that it is prone to degradation and only some cells exhibited GFP-dependent fluorescence (not shown). The fraction of fluorescent cells increased when the protein was expressed in pex3⌬ cells. A complex staining pattern was discernible that included punctate structures, diffuse cyto-FIGURE 6. Expression of Pex22p N -Pex3p C in a pex3⌬ strain leads to bona fide peroxisomes. A, ultrastructure of a Pex22p N -Pex3p C -expressing pex3⌬ strain. The cells were grown for 14 h on oleic acid-containing medium and processed for electron microscopy. The arrow denotes a cluster of peroxisomes. The image of a pex19⌬ cell expressing the same chimera is shown as a negative control. Magnification, 30,000ϫ. B, subcellular distribution of peroxisomal catalase in a pex3⌬ strain expressing Pex22p N -Pex3p C . Post-nuclear supernatants of oleic acid-induced pex3⌬ cells were loaded on top of a linear sucrose gradient (20 -54% w/w) and subjected to centrifugation at 38,000 g for 1.5 h. Fractions of each gradient were obtained from the bottom (fraction 1) to the top (fraction 27) and assayed by immunoblot analysis for the distribution of the peroxisomal marker protein Pex14p and mitochondrial porin. In addition, each fraction was assayed for catalase activity, which is presented in percent relative to the peak fraction. In contrast to the pex3⌬ strain, where catalase remained on top of the gradient, a clear peroxisomal catalase peak was discernible for the Pex22p N -Pex3p C -expressing strain. C, complementation of the pex3⌬ growth defect on oleic acid by Pex3p chimeras. Serial dilutions of the indicated strains were spotted onto oleic acid-containing plates and incubated at 30°C for 5 days. Pex3p as well as the Pex22p N -Pex3p C chimera clearly restored growth. Expression of a Pex13p 166 -310 -Pex3p 46 -441 chimera led to only a minimal improvement of the growth of the pex3⌬ mutant (see also Fig. 7).
plasmic as well as some perinuclear ER staining. Nonetheless, this fusion protein was able to restore peroxisomal matrix protein import, because a punctate staining pattern was observed for DsRed-SKL (Fig. 8A). In several cells peroxisomes were present, although the Pex3p chimera was not detectable, suggesting that only small amounts of Pex3p C are required to generate import-competent peroxisomes (not shown). The observed peroxisome formation was dependent on Pex19p, because DsRed-SKL labeling was only diffuse in a pex3⌬pex19⌬ double-deletion strain (Fig. 8B). Electron micrographs clearly revealed the existence of peroxisomes in this strain (Fig. 8C). However, mPTS Pex13p -Pex3p C was virtually unable to restore growth on fatty acids of a pex3⌬ strain (Fig. 6C), indicating that the peroxisomes formed were not fully functional. This lack of complementation notwithstanding, our data strongly suggest that peroxisome formation is driven by the function of the cytoplasmic domain of Pex3p.

DISCUSSION
In the work presented we compared the targeting of the PMPs Pex3p and Pex22p from S. cerevisiae and provide evidence that both proteins harbor a Pex19-independent mPTS. In addition, we could show that targeting of Pex3p is not as unique as thought.
Pex3p seems to be the central molecule for peroxisome biogenesis and maintenance (15,16,47). At least three crucial functions are described for this peroxin. First of all, it is directly involved in the insertion of PMPs in pre-existing peroxisomes by acting as a membrane receptor for soluble Pex19p-PMP complexes (7,13). The other two functions are directly associated with the special targeting route of Pex3p. In contrast to the majority of PMPs, yeast Pex3p does not seem to target to peroxisomal membranes directly from the cytosol but detours via the ER and is supposed to reach its peroxisomal destination via a vesiclemediated transport (17). In wildtype cells, these Pex3p-containing vesicles are believed to supply peroxisomes with membrane lipids (18). In case of cells devoid of peroxisomes, Pex3p is thought to additionally induce de novo formation of peroxisomes at special subdomains of the ER. The close correlation between targeting and function is underlined by the ability of the mPTS of Hansenula polymorpha Pex3p to induce formation of vesicles competent to mature into peroxisomes upon expression of the full-length protein (22,23). Because the unique ability of Pex3p to initiate organelle formation is intertwined with its targeting, one might expect that the mPTS of Pex3p is unique as well. To test this hypothesis, we compared the function of the Pex3p mPTS with that of Pex22p. The latter PMP seemed to be a good candidate for this approach, because it is functionally unrelated to Pex3p yet possesses the same membrane topology (Fig. 1A), as well as the same location of the mPTS at the extreme N terminus (Fig. 1, B-G). An N-terminal location of the mPTS was previously described for Pex22p in P. pastoris (28).
A straightforward approach to test for eventual differences in the targeting properties of the two mPTS is their mutual exchange. As shown in Fig. 4 (A and F), respectively, the Pex3p-mPTS could indeed functionally replace that of Pex22p in a wild type as well as in a pex22⌬ background. At first sight one could argue that this result was predictable because of the intrinsic property of the Pex3p mPTS to be targeted to membranes of both functional peroxisomes as well as of peroxisomal remnants. However, Pex3p was published to first target to the ER from where it is subsequently released in a Pex19p-dependent manner so as to reach peroxisomes (17,22,25). In the Pex3p N -Pex22p C fusion protein, however, neither the Pex3p-N terminus nor the Pex22p C-terminal domain bound Pex19p, shown in vitro by a peptide scan as well as in vivo by a two-hybrid-assay (Fig. 2, A  and B). The observations made with this chimera and with Pex22p clearly suggested that Pex3p and Pex22p can reach peroxisomes independently of Pex19p and thus harbor a dis-tinct class of mPTS, a so-called type 2 mPTS, as defined by Gould and co-workers (10).
Expression of the vice versa construct, i.e. the Pex22p N -Pex3p C -chimera, gave rise to a stunning result. This fusion protein was not only transported to peroxisomes in wildtype cells but beyond that also in pex3⌬ cells (Fig. 5, A and F), showing that the mPTS of Pex3p is replaceable. Most interestingly, the Pex22p N -Pex3p C -chimera was shown to be fully functional by its ability to complement the defect of a pex3⌬ strain for import of peroxisomal membrane and matrix proteins (Fig. 6, A and B) as well as for the defect in oleic acid utilization (Fig.  6C). This result clearly indicated the dispensability of the N terminus of Pex3p for its function in peroxisome biogenesis; the C terminus of Pex3p contains all the information required to form peroxisomes.
Peroxisomal targeting of yeast Pex3p is strongly favored to occur via the ER (17). Although not the focus of this study, the fact that in a pex19⌬ mutant Pex22p accumulated in the same compartment as Pex3p (Fig. 3C) clearly suggested that this also holds true for Pex22p. Further work employing time lapse microscopy will be required to directly demonstrate the topogenesis of this peroxin.
Replacing the Pex22p mPTS with the Pex19p-dependent mPTS of Pex13p (Fig. 7A) (11) still allowed the faithful delivery of a major fraction of the chimera to peroxisomal remnants in a pex22⌬ strain, with a minor portion being localized at the ER (Fig. 7, B and C). The latter staining was probably due to a mistargeting of the fusion protein, because the ER localization persisted in the absence of Pex19p (Fig.  7D). The mPTS Pex13p -Pex22p C chimera could also restore peroxisomal functions in cells devoid of Pex22p, indicated by the ability to import a synthetic peroxisomal matrix protein (Fig. 7, B and C) and to restore growth on oleic acid (Fig. 4G). This finding not only corroborated that the function of Pex22p is independent of its N terminus but also showed that Pex22p retained function when targeted by a Pex19p-dependent mPTS.
The mPTS Pex13p -Pex3p C chimera proved not all that stable and was localized to multiple intracellular sites including the ER (Fig. 8A), making interpretations of its targeting more difficult. Nonetheless, the Pex3p-C terminus was able to induce formation of peroxisome-like structures in a Pex19pdependent manner (Fig. 8, A and B). These structures had a normal morphology (Fig. 8C) and were competent for import of at least a subset of peroxisomal matrix proteins Pex3⌬ cells coexpressing a C-terminal GFP fusion of the Pex13p 166 -310 -Pex3p C chimera and DsRed-SKL were inspected by fluorescence microscopy. DsRed-SKL gave rise to a punctate staining pattern that was partially superimposable with that of the Pex3pC chimera, which exhibited also some cytosolic and ER-staining. B, test for function of Pex13p 166 -310 -Pex3p C in the absence of peroxisomes. Coexpression of the chimera and DsRed-SKL in a pex3⌬pex19⌬ double knock-out strain led to a cytosolic localization of DsRed-SKL, indicating that Pex19p is essential for the peroxisomal restoration process. Bar, 2 m. C, ultrastructure of Pex13p 166 -310 -Pex3p C -expressing pex3⌬ cells. The cells were grown for 14 h on oleic acid-containing medium and processed for electron microscopy. The arrows denote the appearance of peroxisomes. A conjunction between ER and a peroxisome appears to be visible in the right panel. Magnification, 30,000ϫ (left panel) and 18,500ϫ (right panel), respectively. (Fig. 8A). In contrast to the Pex22p N -Pex3p C -chimera, however, mPTS Pex13p -Pex3p C virtually failed to restore fatty acid utilization in a pex3⌬ strain (Fig. 6C). Given that substitution of the mPTS of Pex3p with that of Pex22p retained function of Pex3p, one might interpret this improper peroxisome formation to be due to an aberrant targeting of the C-terminal domain of Pex3p by the Pex19p-dependent mPTS of Pex13p. Carrying on this thought, the mPTS of the Pex3p/Pex22ptype might interact with putative membrane factors in the absence of peroxisomes and thereby determine the correct site for peroxisome development. By contrast, an mPTS not belonging to this type would provoke a more random membrane integration of the cytoplasmic domain of Pex3p. This might still trigger peroxisome formation, albeit in an imprecise fashion with some components required for peroxisome function missing. However, considering the ambiguity of mPTS Pex13p -Pex3p C localization even in a wild-type strain, where the mPTS of Pex13p should direct Pex3p C to existing peroxisomes, such an interpretation needs to be handled with caution. Nonetheless, our data strongly suggest that the cytoplasmic domain of Pex3p can drive peroxisome formation, provided that it is endowed with an mPTS.
In summary, our results are in line with the following scenario. The mPTS of Pex3p and Pex22p belong to the same type of targeting signal accomplishing Pex19p-independent targeting to peroxisomes that is likely to occur via the ER but might also happen directly from the cytosol. In the absence of peroxisomes, the mPTS of either Pex3p or Pex22p recruit Pex3p C to membrane segments of the ER competent for peroxisome formation. Subsequently, by virtue of the C-terminal domain of Pex3p, Pex19p-dependent import of PMPs takes place. This is accompanied by a release of peroxisomal precursor vesicles that takes place by a so far unknown mechanism. Future experiments will have to show which components (if any) define the particular ER sections required for origination of fully functional peroxisomes, whether these components directly interact with the type 2 mPTS present in Pex3p and Pex22p, and whether these components are also required in cells harboring normal peroxisomes.