Building New Models for Peroxisome Biogenesis

SIGNALS FOR TARGETING

Although one might expect that all proteins going into peroxisomes would have the same targeting signal, nearly all matrix-localized enzymes contain one of two peroxisome-targeting signals (PTSs), each of which is necessary and sufficient to direct proteins from the cytosol into peroxisomes (for review, see Olsen, 1998; Subramani et al., 2000). Examples of PTSs in plant proteins are shown in TableI. PTS1, the first PTS identified, is a carboxyl terminal tripeptide consisting of the three amino acids Ser-Lys-Leu, or related variants (Subramani et al., 2000). It is interesting to note that though PTS1s are conserved across all eukaryotes, plant PTS1s apparently exhibit more variability in sequence compared with accepted signals in animals (Mullen et al., 1997a). Even though it may seem quite remarkable that such a short sequence motif can control import into a specific organelle, the majority of peroxisomal matrix proteins carry a PTS1 signal.

The second type of PTS (PTS2) is present within the first 20 to 30 amino acids at the amino terminus and has a loosely conserved sequence of nine amino acids (Flynn et al., 1998; Olsen, 1998). In plants and mammals, the PTS2 signal is cleaved after the protein arrives in the peroxisomal matrix. Compared with the number of proteins that have PTS1 signals, few PTS2 proteins have been identified. The most well-characterized PTS2 protein is thiolase, a fatty acid β-oxidation enzyme found in plants, animals, and yeasts. Other known plant PTS2 proteins include isozymes of malate dehydrogenase, citrate synthase, amine oxidase, and Asp aminotransferase. In fact, there appear to be more PTS2 proteins in plants than in other eukaryotes.

Several peroxisomal matrix proteins contain neither a recognizable PTS1 nor a PTS2 consensus sequence (Subramani et al., 2000). Internal targeting signals for these proteins have not been fully characterized, however, and no specific interactions between internal signals and receptors have been reported. Peroxisomal membrane protein (PMP) targeting and insertion requires other types of targeting signals that will be discussed later in this review.

The discovery of multiple targeting signals for peroxisomal protein transport led to the prediction that multiple receptors interact with the matrix proteins and define separate import pathways. Support for the existence of multiple peroxisomal import pathways was provided by genetic screens for peroxisomal protein import mutants in yeasts. Several distinct classes of mutants were found: those that were defective in the import of PTS1 proteins, those that were defective in the import of PTS2 proteins, and those that were defective in import of both (for review, see Erdmann et al., 1997). The proteins initially identified in these screens to be required for peroxisome biogenesis have been termed peroxins (pex). More than 20 peroxins currently have been described; homologs for most, but not all, of the peroxins can be identified in the Arabidopsis genome.

Yeast and mammalian mutants with a defect in PEX5 are unable to import PTS1 proteins, but most are able to import proteins containing a PTS2. Mutants lacking a functional Pex7p, on the other hand, are unable to import PTS2 proteins. These data (see references in Subramani et al., 2000), along with the fact that more than one type of topogenic sequence is used by peroxisomal proteins, led to further speculation that there may be more than one receptor that interacts with peroxisomal proteins. Thus, the first revision of the simple import model included two separate pathways for protein import into the peroxisomal matrix (see Fig. 2). One pathway directs import of all PTS1 proteins to the peroxisome, whereas the other pathway handles PTS2 proteins; each pathway is mediated by separate soluble receptors that converge at common docking sites on the peroxisomal membrane. Despite some controversy (see below), most investigators now agree that Pex5p and Pex7p are the cytosolic receptors for PTS1 and PTS2 proteins, respectively. Additional proteins on the peroxisome membrane serve as docking proteins for these soluble receptors and their protein cargo. Next, we will take a closer look at these soluble receptors that interact with the targeting signals.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Separate receptors direct the import of peroxisomal matrix proteins with different targeting signals. Proteins possessing a carboxyl terminal PTS1 interact with a specific PTS1 receptor, Pex5p. Proteins possessing an amino-terminal PTS2, on the other hand, interact with a distinct PTS2 receptor, Pex7p. The soluble receptors converge on common docking complexes on the membrane after binding to their matrix protein cargo.

CYTOSOLIC RECEPTORS

The discovery of cytosolic receptors for peroxisomal matrix proteins was surprising because, at the time, most researchers expected the import pathways to resemble protein import into mitochondria and chloroplasts more closely. By now, it is widely accepted that Pex5p is the cytosolic receptor for PTS1 proteins and Pex7p is the PTS2 protein receptor. In Arabidopsis, PEX5 encodes a protein that contains seven tetratricopeptide repeat domains in its carboxyl terminus (Brickner et al., 1998); Pex5p has also been cloned from tobacco (Kragler et al., 1998) and watermelon (Wimmer et al., 1998). Proteins that possess a PTS1 tripeptide have been shown to interact with the TPR domains of Pex5p, whereas those lacking a PTS1 targeting signal do not (Terlecky et al., 1995). PEX7 encodes a protein composed almost entirely of WD-40 (β-transducin related) repeats. Pex7p, the PTS2 receptor, binds specifically to PTS2 signals and interacts with thiolase in both two-hybrid and co-immunoprecipitation experiments. In plants, Pex7p has also been cloned from Arabidopsis (Schumann et al., 1999).

Pex5p and Pex7p also appear to interact directly or indirectly with each other during import—adding another layer of complexity to the import model. Two isoforms of Pex5p have been identified in mammals. Both serve as a PTS1 protein receptor, but the longer form has also been shown to interact directly with Pex7p (Otera et al., 2000). Yeasts appear to have only the short isoform, consistent with the absence of evidence of PTS1 and PTS2 pathway interactions. Plant Pex5p is most similar to the longer mammalian isoform, suggesting that the two soluble receptors will interact with each other during matrix protein import. It is not yet clear whether the two pathways converge at the membrane, as shown in Figure 2, or in the cytosol before docking on the membrane, as suggested in Figure 4.

The exact localization of these two receptors has been somewhat controversial. Depending upon the species and the experimental techniques used, Pex5p and Pex7p have each been localized to the cytoplasm, to the peroxisomal membrane, and to the matrix (Olsen, 1998;Subramani et al., 2000). Because the predicted amino acid sequences of the two receptor proteins do not reveal obvious transmembrane domains, Pex5p and Pex7p do not appear to be typical integral membrane-bound receptors. Most researchers now believe that Pex5p and Pex7p are primarily cytosolic proteins that bind polypeptides destined for the peroxisomal matrix and that accompany their cargo to the peroxisomal membrane.

There are at least two mechanisms that could explain what happens after the receptor and its cargo arrive at the membrane. In one scenario, the receptor and its targeted protein specifically recognize the translocation machinery (including Pex13p, Pex14p, and Pex17p) on the membrane. After docking, the receptor releases its cargo and remains in the cytosol while the cargo is transported across the membrane into the peroxisome. In an alternate manner, the receptor could be transported through the translocation machinery while still bound to the peroxisomal matrix protein cargo, as suggested in Figure3. If receptors do enter the matrix of the organelle, they could be degraded or exported from the organelle to the cytoplasm in a manner similar to recycling of nuclear import receptors. In a recent landmark paper, human Pex5p was shown to translocate into the peroxisomal matrix and then be recycled to the cytosol (Dammai and Subramani, 2001), lending support to the models shown in Figures 3 and 4.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Model of PTS1 protein import showing Pex5p shuttling between the cytosol and peroxisome. The PTS1 receptor, Pex5p, binds to the PTS1 protein in the cytoplasm. The Pex5p-PTS1 protein complex then docks on a membrane-bound peroxin complex, comprising at least Pex13p, Pex14p, and Pex17p. The Pex5p-PTS1 protein complex translocates across the membrane. Dissociation of the complex occurs in the matrix of the organelle; Pex5p releases its protein cargo in the matrix and may be recycled back to the cytoplasm through an entirely hypothetical export channel. An alternative model that is not shown proposes that Pex5p dissociates from the PTS1 protein and is released from the membrane before the PTS1 protein is translocated across the membrane.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Summary model for peroxisomal matrix protein import. In the cytosol, a PTS1 protein interacts with Pex5p and a PTS2 protein interacts with Pex7p. The matrix protein receptor complexes may assemble directly in the cytosol or during docking on the membrane complex; each pathway can also occur independently. Pex7p binds to Pex14p; Pex5p binds to Pex13p and Pex14p. After docking, the receptor-cargo complexes are translocated across the membrane, through an entirely uncharacterized transport apparatus. One possibility is shown: translocation may occur through a channel formed by some or all of the zinc-binding proteins, Pex2p, Pex10p, and Pex12p. Inside the matrix, but close to the membrane, the PTS1 and PTS2 protein cargoes dissociate from their respective receptors. The PTS2 protein-targeting signal is cleaved off. Pex5p and Pex7p are recycled out of the peroxisome, perhaps through the zinc-binding proteins or another unidentified export complex (as shown in Fig. 3), for subsequent rounds of matrix protein binding and import.

DOCKING ON THE MEMBRANE

It was once thought that because there are separate PTS1 and PTS2 receptors, there might also be separate PTS1 and PTS2 import channels. Most of the evidence now supports a model in which both receptors dock at the same protein complex on the peroxisome membrane, as shown in Figure 4. Yeast peroxisome mutants again helped identify the players involved in recruiting the receptors to the membrane and then keeping them there during import. Besides the two receptors, at least six peroxins, Pex13p, Pex14p, Pex17p, Pex18p, Pex20p, and Pex21p, have been implicated primarily in the process of peroxisomal matrix protein import, though the latter three proteins listed may be specific for yeast peroxisomes (see below).

Three membrane-bound proteins, Pex13p, Pex14p, and Pex17p, interact to form a docking site for the cytosolic receptors on the organelle membrane (Fig. 3). Pex14p is a cytosolic facing, peripheral membrane protein that interacts with both Pex5p and Pex7p, usually as a dimer.pex14 mutants are unable to import both PTS1 and PTS2 proteins (see references in Subramani et al., 2000). Based on these results, Pex14p is thought to be the primary membrane docking protein where the two import pathways might converge. Pex17p is another cytosolic facing, peripheral membrane protein that interacts directly with Pex14p, and therefore indirectly with Pex5p and Pex7p. Mutations in PEX17 inhibit both the PTS1- and PTS2- dependent import pathways (Huhse et al., 1998).

Pex14p has been further shown to interact with Pex13p, an integral PMP that has an SH3 (Src homology 3) domain and is oriented in the membrane such that both the carboxyl terminus and the amino terminus face the cytosol. Pex14p and Pex13p are known to interact via the carboxyl-terminal SH3 domain because mutations in this domain result in the inactivation of Pex13p function (Elgersma et al., 1996). Pex13p is required for Pex14p to function at the membrane. Like pex14and pex17 mutants, pex13 mutants are unable to import both PTS1 and PTS2 proteins. The SH3 domain of Pex13p has also been shown to interact with the PTS1 receptor, Pex5p, but it is less clear whether Pex13p interacts with the PTS2 receptor, Pex7p (Erdmann and Blobel, 1996; Girzalsky et al., 1999).

Pex18p and Pex21p are structurally and functionally related peroxins that interact with Pex7p and seem to be required for Pex7p's targeting to the membrane. In Saccharomyces cerevisiae cells lacking both Pex18p and Pex21p, Pex7p remained cytosolic and PTS2 targeting was completely abolished (Purdue et al., 1998). The peroxin Pex20p was discovered in the yeast Yarrowia lipolytica. Y. lipolytica has no known Pex7p-like PTS2 receptor. Instead, Pex20p acts as the PTS2 receptor in this organism (Titorenko et al., 1998). Pex20p exhibits no homology to Pex7p and binds PTS2 proteins independently of the amino-terminal targeting signal. Plant homologs of Pex18p, Pex20p, and Pex21p have not yet been identified, even after thorough searches through the completed Arabidopsis genome.

Most of the components required for peroxisome matrix protein import have not been fully characterized in plants, in yeast, or in animals. In addition to the six major proteins discussed above, other peroxins including Pex2p, Pex4p, Pex10p, Pex12p, Pex19p, and Pex22p have even less well-defined roles in peroxisomal protein import. As indicated in Figure 4, Pex2p, Pex10p, and Pex12p are integral PMPs with a zinc RING finger motif at the carboxyl terminus. Their role in matrix protein import is completely unknown, but it is reasonable to propose a role for zinc binding to these proteins as a regulatory mechanism during Pex5p binding and subsequent PTS1 protein import. The zinc-binding domain of Pex12p has been shown to interact with both Pex5p and Pex10p (Okumoto et al., 2000). Recent evidence indicates that PTS1 protein import is stimulated by the addition of zinc (Terlecky and Fransen, 2000). In addition, protein import complex intermediates may provide powerful tools for identifying other components of the translocation machinery in plants (Pool et al., 1998).

ASSEMBLY OF PEROXISOME MEMBRANES

The peroxisome membrane forms an important barrier between the organelle and the rest of the cell. This membrane has been poorly characterized, mainly because of the difficulties encountered in isolating membranes free from contamination. The peroxisome membrane has a relatively low phospholipid to protein ratio and low levels of cholesterol. Although it is thinner than most other single membrane-bound organelles, its width is comparable to the endoplasmic reticulum (ER) membrane (Beard and Allen, 1968). Peroxisomal membranes contain porins that are highly permeable to small metabolites including glycolate, glycerate, and inorganic anions (for a thorough review of peroxisomal porins, see Reumann, 2000).

There is relatively little known about the mechanisms for sorting membrane proteins to peroxisomes, even though many of the peroxins characterized to date are PMPs. As mentioned previously, integral membrane proteins use signals other than PTS1 or PTS2 for targeting to the organelle (Jones et al., 2001). Different consensus sequences for membrane protein targeting have been proposed. These signals, termed membrane PTSs (mPTSs), seem to reside near the carboxyl terminus of the membrane proteins, adjacent to at least one transmembrane domain (see Table I). The most common feature of the few characterized mPTSs consists of a cluster of positively charged, or basic, amino acids (Baerends et al., 2000; Mullen and Trelease, 2000).

Studies of the mechanisms of PMPs assembly are providing some of the most exciting new results in this field. Several recent reviews have focused on these questions and present the controversies involved in more detail (Baerends et al., 2000; Subramani et al., 2000; Titorenko and Rachubinski, 2001). In Figure 5, we show a simplified model of how membrane assembly may take place. Some PMPs are inserted directly from the cytoplasm into the organelle membrane, as originally proposed. These PMPs are termed Type II proteins and include PMP22, PMP34, PMP47, and PMP70. Although their function is unknown, they do not appear to be required for or involved in the biogenesis of peroxisomes. A subset of PMPs, the Type I proteins, have been shown more recently to be targeted to the ER prior to insertion in or localization to the peroxisome membrane, probably through an intermediate vesicular compartment, sometimes referred to as a preperoxisomal vesicle. At least some of the Type I proteins are specifically involved in organelle biogenesis (e.g. Pex2p, Pex3p, Pex 15p, and Pex16p). Cottonseed pAPX, however, is a Type I PMP that has a biochemical function not related to organelle biogenesis. pAPX follows a pathway from the cytosol to a specific subdomain of the ER that defines or gives rise to a preperoxisomal vesicle (Mullen et al., 1999). Maturation of preperoxisomal vesicles or fusion between multiple vesicles leads to the more familiar mature peroxisome. Mature peroxisomes most likely divide by fission.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

The role of the ER in targeting membrane proteins to the peroxisome. PMPs are synthesized on soluble ribosomes in the cytosol. A subset of PMPs, the Type II PMPs (including PMP22, PMP34, PMP47, and PMP70), appears to be directly inserted into the membrane from the cytosol. The Type I PMPs (including peroxisomal ascorbate peroxidase [pAPX], Pex2p, Pex3p, Pex15p, and Pex16p) are first localized to a domain on the ER from which a preperoxisomal vesicle forms. Each preperoxisomal vesicle might fuse with other vesicles or with existing peroxisomes (or could conceivably mature directly), ultimately delivering the membrane proteins to the mature peroxisome. Mature peroxisomes divide by fission.

No evidence of matrix protein trafficking through the ER has yet been presented. It is not clear whether matrix proteins could be imported into preperoxisomal vesicles. That should depend on when and where the membrane translocation complex forms. As many as six separate classes of peroxisome vesicles have been isolated from the yeast Y. lipolytica; each class appears to be import competent for distinct sets of matrix proteins (Titorenko and Rachubinski, 2001). It has also been proposed that the PMP Pex16p converts preperoxisomal vesicles originating from the ER into recognizable peroxisomes by mediating the insertion of other proteins into the membrane (Eitzen et al., 1997). This in turn would allow the assembly of the matrix protein import apparatus, the subsequent import of matrix proteins, and the formation of mature peroxisomes. An Arabidopsis homolog of Pex16p has been identified (Lin et al., 1999). It is interesting that it has been implicated in ER-dependent protein and oil body biogenesis, but its role in peroxisome biogenesis has not been studied.

CHAPERONES

So far, we have focused on peroxisome-specific components required for organelle assembly. There are, however, additional cellular components that are clearly involved in peroxisome biogenesis. Chaperones, usually heat shock proteins, are often required for protein import into organelles and peroxisomes appear to be no exception. Chaperones such as Hsp70 (a 70-kD class of heat shock proteins) bind to proteins during and shortly after translation to prevent the nascent protein from misfolding and to keep the protein in an extended conformation. This loose conformation is usually necessary for the protein to be able to cross the membrane. Peroxisomes, however, have the unusual capacity to import proteins that are fully folded, assembled in oligomers, disulfide bonded, or even conjugated to 9-nm gold particles (for review, see Crookes and Olsen, 1999). So why are chaperones needed for peroxisomal protein import if fully folded proteins can enter the organelle? Studies from our laboratory suggest that although oligomeric protein can be imported in vitro, import of monomeric proteins is much more efficient (Crookes and Olsen, 1998). Antibodies against cytosolic Hsp70 have been shown to inhibit the import of peroxisomal matrix and membrane proteins (Walton et al., 1994; Crookes and Olsen, 1998; Mullen et al., 1999). We have also successfully used Hsp70 antibodies to immunoprecipitate peroxisomal proteins, suggesting that Hsp70s and peroxisomal proteins interact directly. Cytosolic Hsp70's role might be to maintain the matrix protein in an import-competent state, perhaps by participating, with other chaperones, in the formation of a cytosolic multiprotein complex similar to the one required for the activation of the glucocorticoid receptor in mammals (Pratt et al., 2001) or by simply maintaining the proteins in a loosely folded conformation.

Chaperones may facilitate protein import into plant peroxisomes at several subcellular locations and may perform many different roles. Two Hsp70 isoforms have been identified in the matrix of cucumber peroxisomes (Diefenbach and Kindl, 2000). PMP73 is immunorelated to the Hsp70 family, but its function is unknown (Corpas and Trelease, 1997). There is also evidence indicating the involvement of chaperones other than Hsp70s in peroxisomal protein import in plants. Hsp40s are chaperones that enhance Hsp70's activity, and both a membrane-bound and a cytosolic Hsp40 homolog have been identified in plants (Preisig-Muller et al., 1994; Diefenbach and Kindl, 2000). A member of the Hsp60 family, which folds transported polypeptides into active conformations after translocation across the mitochondrial and chloroplast membranes, has been reported to reside in the peroxisome matrix (Velez-Granell et al., 1995).

The first documented example of Hsp90 involvement in organelle biogenesis came from experiments showing that Hsp90 antibodies inhibit the in vitro import of PTS1 proteins (Crookes and Olsen, 1998). Hsp90s function in a “super-chaperone” complex with Hsp70s to prime mammalian steroid receptors for ligand binding (Jakob and Buchner, 1994; Pratt et al., 2001). Hsp90s also influence the assembly of protein complexes. The role of Hsp90 in peroxisomal import therefore could be to prime the cytosolic PTS receptors for binding to peroxisomal proteins. In fact, the PTS1 receptor, Pex5p, and Hsp90 have been shown to interact in co-immunoprecipitation experiments (Pratt et al., 2001). This makes sense because Hsp90 has a single binding site for tetratricopeptide repeat domains (Pratt et al., 2001) and Pex5p from Arabidopsis has seven consecutive tetratricopeptide repeats that constitute the carboxyl terminal half of the protein (Brickner et al., 1998). In an alternate manner, Hsp90s might control the assembled state of the targeted proteins or of the import complex prior to membrane translocation.

Energy, apparently in the form of ATP, is required for import of proteins into peroxisomes. GTP was also shown to be sufficient to support the in vitro import of at least one PTS1 protein into peroxisomes (Brickner and Olsen, 1998). ATP has been implicated in the formation and function of chaperone complexes (Pratt et al., 2001). The exact role of ATP in peroxisomal protein import has not been established, but it is reasonable to suspect that some energy is required for the chaperone activities. It is also possible that ATP or GTP binds to one or more of the components of the translocation apparatus.

CONCLUSIONS

During the past 10 years, models to explain peroxisome biogenesis have evolved in far more complex ways than originally imagined. The current models include multiple targeting signals to direct proteins to the peroxisome, multiple cytosolic receptors that interact with the targeted proteins, membrane components that interact with the receptors and proteins, other membrane components that have additional roles in biogenesis, chaperones that facilitate import, and complex targeting mechanisms to insert membrane proteins. There are, however, still many questions about peroxisome biogenesis that remain unanswered. Contributions from researchers using yeasts, mammals, and plants will provide us with new information to refine our understanding of these complex mechanisms, and that will cause us to continue to build and test new models for peroxisome biogenesis.