Towards solving the mystery of peroxisomal matrix protein import

Abstract

Peroxisomes are vital metabolic organelles that import their lumenal (matrix) enzymes from the cytosol using mobile receptors. Surprisingly, the receptors can import even folded proteins, but the underlying mechanism has been a mystery. Recent results reveal how import receptors shuttle cargo into peroxisomes. The cargo-bound receptors move from the cytosol across the peroxisomal membrane completely into the matrix by a mechanism that resembles transport through the nuclear pore. The receptors then return to the cytosol through a separate retrotranslocation channel, leaving the cargo inside the organelle. This cycle concentrates imported proteins within peroxisomes, with the energy for cargo import being supplied by receptor export. Peroxisomal protein import thus fundamentally differs from other previously known mechanisms for translocating proteins across membranes.

The mystery of peroxisomal matrix protein import

Peroxisomes are organelles found in most eukaryotic cells [1]. They consist of a single membrane surrounding a dense matrix that houses crucial metabolic pathways [2]. These generally include the oxidation of fatty acids and other biomolecules [3], as well as the decomposition of hydrogen peroxide and other reactive oxygen species [4]. Yet peroxisomes are remarkably versatile and often perform additional reactions [5–9], which in humans include the production of essential precursors for the biosynthesis of bile salts and of lipids contained in myelin [10,11]. These diverse functions require a wide range of enzymes to be post-translationally imported into the peroxisomal matrix from the cytosol [12]. Congenital defects in this process cause life-threatening peroxisome biogenesis disorders – most notably the Zellweger spectrum – which affect multiple organs and for which no cure is known [13,14].

Seminal research from many groups has delineated the peroxisomal matrix protein import pathway, and has identified the cellular factors, called peroxins (PEX), that mediate the process. Various aspects of peroxisomal matrix protein import are unusual, with arguably the most enigmatic being that peroxisomes can import folded proteins and even protein complexes [15,16]. This surprising property fundamentally differs from protein translocation into the endoplasmic reticulum (ER), mitochondria, and chloroplasts, which can only import proteins as monomers in an unfolded conformation [17–19]. Recent results have clarified this longstanding question and have led to a new model of peroxisomal matrix protein import. In this review, we will describe this emerging paradigm and discuss the experimental data supporting it.

Overview of protein import into peroxisomes

Most peroxisomal matrix proteins contain a peroxisome targeting signal called PTS1, which resides at their C terminus and comprises the amino-acid sequence Ser-Lys-Leu (SKL) or variants of it [20]. This signal is recognized in the cytosol by the soluble receptor PEX5 (Fig. 1A, step 1). PEX5 consists of a long unstructured N-terminal region followed by a globular tetratricopeptide-repeat (TPR) domain (Fig. 1B). The TPR domain directly binds the PTS1 peptide [21]. Cargo-bound PEX5 is recruited to peroxisomes by a complex that includes the conserved membrane proteins PEX13 and PEX14 (Fig. 1A, step 2). The receptor then shuttles the cargo across the peroxisomal membrane through a conduit formed by PEX13 (Fig. 1A, step 3). PEX5 is driven into the matrix by an interaction between a lumenal domain of PEX14 and motifs in the receptor’s N-terminal segment (Fig. 1A, step 4). These motifs are characterized by the amino-acid sequence WxxxF/Y (where “x” denotes any residue) (Fig. 1B), and exist in variable numbers in all known peroxisomal import receptors [1,22].

Figure 1. Import of peroxisomal matrix proteins by the receptor PEX5.

(A) Model of peroxisomal matrix protein import in metazoans. Step 1: PEX5 binds cargo proteins in the cytosol. Step 2: Cargo-bound PEX5 is recruited to peroxisomes by a complex containing the membrane proteins PEX13 and PEX14. Step 3: Cargo-bound PEX5 traverses the membrane through a conduit formed by multiple copies of PEX13. The conduit contains a dense meshwork formed from PEX13’s YG domain (pink), into which PEX5 partitions using its WxxxF/Y motifs (dark blue). Step 4: PEX5 is drawn into the matrix by favorable lumenal interactions (represented by pink dotted lines) between the receptor’s WxxxF/Y motifs and PEX14 oligomers. Step 5: PEX5 spools its flexible N terminus into the cytosol through a pore in the PEX2-PEX10-PEX12 ubiquitin ligase complex, which then monoubiquitinates the receptor on a conserved cysteine (red). Step 6: Monoubiquitinated PEX5 is pulled out of the matrix through the ligase pore by the PEX1-PEX6 AAA ATPase, which unfolds the receptor and causes cargo to be stripped off inside the matrix. Step 7: PEX5 refolds in the cytosol and ubiquitin is removed by deubiquitinases, resetting the receptor for another import cycle. (B) Diagram illustrating key features of metazoan PEX5, including the site of monoubiquitination (Cys), amphipathic helices (AH) 1 and 2 required for recycling, WxxxF/Y motifs (dark blue), the TPR domain that binds PTS1 cargo, and the binding site for the adapter PEX7 that binds PTS2 cargo. Residue coordinates in the human protein are labeled on top.

PEX5 must then return to the cytosol. This recycling step requires PEX5 to be monoubiquitinated on a conserved cysteine (Fig. 1B) near the receptor’s N terminus [23,24]. The reaction is catalyzed by a membrane-embedded ubiquitin ligase complex, which consists of three RING finger-containing proteins called PEX2, PEX10, and PEX12 (Fig. 1A, step 5).

Monoubiquitinated PEX5 is next pulled out of the matrix by a hexameric AAA ATPase assembled from alternating copies of PEX1 and PEX6 (Fig. 1A, step 6). This extraction step releases the receptor into the cytosol, while the cargo is left behind inside the organelle. Finally, the ubiquitin is removed by deubiquitinating enzymes (Fig. 1A, step 7), thereby resetting the receptor for a new round of import.

Some matrix proteins contain an alternative peroxisome targeting signal called PTS2 that resides at their N terminus. The PTS2 is recognized by an adapter called PEX7 [25], which in most organisms binds to a conserved motif within the unstructured region of PEX5 [26,27] (Fig. 1B). In certain yeasts, PEX7 binds instead to PEX5 paralogs that have lost the TPR domain but have retained the N-terminal unstructured region with its WxxxF/Y motifs and conserved cysteine [22]. These paralogs import PTS2 cargo by a similar mechanism as PEX5 [28]. We will therefore focus specifically on PEX5, and explore more thoroughly the individual steps by which this receptor delivers proteins into peroxisomes.

PEX5 crosses the membrane through a conduit formed by PEX13

PEX13 has long been recognized to be essential for peroxisomal matrix protein import [29], although its role has only recently become clear. The hallmark of PEX13 is a YG domain (Fig. 2A), an inherently unstructured region enriched in aromatic and small amino acids, predominantly tyrosine (Y) and glycine (G). These unusual sequence properties had originally been noted by multiple groups [30–32]. They are preserved despite low overall sequence conservation [33–35] and are sufficient to identify even distantly related PEX13 homologs [35]. The tyrosines in particular are essential, as mutating them to serines in yeast abolishes matrix protein import [33,34]. Interestingly, the exact position of the tyrosines is not critical as long as a sufficient number are present [33,34]. The YG domain resides near the N terminus of PEX13 and is followed by a long amphipathic helix (AH); certain species (e.g., green algae) contain a second YG domain downstream of the AH (Fig. 2A). A transmembrane (TM) segment and an SH3 domain additionally occur in PEX13 from opisthokonts (i.e., fungi and animals) and some protozoa (Fig. 2A), but are otherwise not universally conserved [1].

Figure 2. The YG domain of PEX13 forms a nuclear pore-like hydrogel.

(A) Domain organization of PEX13 from organisms representing the indicated eukaryotic clades (in order from top to bottom: Saccharomyces cerevisiae, Tetrahymena thermophila, Chlamydomonas reinhardtii, and Arabidopsis thaliana; UniProt accession codes P80667, Q23KA1, A0A2K3CZW8, and Q9SRR0, respectively). The locations of the YG domain and the long amphipathic helix (AH) are indicated, as are the positions of the transmembrane (TM) segment and the SH3 domain that occur in some species. The amino-acid sequence of the YG domain from A. thaliana PEX13 is shown below, with aromatic residues highlighted in pink. (B) Hydrogel formation by the purified YG domain of PEX13 from A. thaliana. A concentrated solution (40 mg/ml) of the protein was pipetted into silicone tubing and allowed to gel, then squeezed out onto a colored surface and photographed. Gelation was observed with the wild-type (WT) protein, whereas a mutant (Y→S) whose tyrosines were all converted into serines remained fluid. (C) The scheme on the left depicts how the YG hydrogel consists of a meshwork held together by cohesive interactions between the domain’s Y residues. PEX5 partitions into this meshwork by transiently disrupting the cohesive interactions using its WxxxF/Y motifs (dark blue). The scheme on the right shows how hydrogels formed by nucleoporin FG repeats are primarily held together by their phenylalanine (F) residues. Nuclear transport receptors (NTRs) diffuse through the FG meshwork by transiently binding the F residues using hydrophobic patches.

The sequence properties of the YG domain uncannily resemble those of nucleoporin FG repeats. These unstructured segments consist of recurring phenylalanine and glycine motifs that assemble into a dense meshwork within the central channel of nuclear pores. This meshwork functions as a permeability barrier (a selective phase) that restricts passage of most proteins and other large molecules, yet allows nuclear transport receptors to rapidly move through with bound cargo [36]. Notably, the FG meshwork can be reconstituted in vitro as a hydrogel from purified FG repeats. FG hydrogels are mostly held together by hydrophobic interactions, primarily between their phenylalanine residues [37], and faithfully recapitulate the selective permeability of nuclear pores [38–42].

The YG domain of PEX13 indeed forms a similar selective phase in the peroxisomal membrane. This conclusion was originally based on the observation that the purified YG domain of PEX13 from Arabidopsis thaliana can form hydrogels (Fig. 2B) [33]. The hydrogels were prepared using methodology developed for FG repeats, and assembled spontaneously even in the presence of urea (which was added during purification to delay gelation). YG hydrogels remained stable after the urea was washed away. Cohesive interactions between tyrosines in the YG domain likely hold the gel together, as mutating these residues to serines abolished hydrogel formation (Fig. 2B). Hydrogels could be assembled at YG-domain concentrations up to 40 mg/ml, because otherwise the recombinant polypeptide gelled too quickly to be handled. However, the concentration of YG domains within the actual conduit in the peroxisomal membrane has been estimated to be much higher [33]. Analogously, the FG-repeat concentration within nuclear pores has been calculated to exceed 200 mg/ml [36,38].

PEX5 fused to green fluorescent protein (GFP) was observed to rapidly partition into YG hydrogels that had been equilibrated in physiological buffer [33]. GFP alone did not accumulate, while a model cargo composed of GFP fused to a PTS1 was readily dragged into the hydrogel by PEX5. The partitioning required the receptor’s WxxxF/Y motifs. Presumably, the motifs’ aromatic residues transiently replace the cohesive interactions between the tyrosines holding the hydrogel together (Fig. 2C), thereby enabling the receptor to diffuse through with bound cargo.

A recent study confirms the formation of a condensed phase by the YG domain of PEX13 [34]. In this case, a region of yeast PEX13 encompassing the YG domain was found to form liquid droplets in vitro, into which PEX5 could partition together with bound cargo. To what extent the physiological conduit is liquid rather than gel-like is debatable, as some fluidity would be required to accommodate a translocating import receptor. Nonetheless, the droplets were only studied in the presence of the crowding agent polyethylene glycol; whether the droplets would persist and allow partitioning of PEX5 in its absence has not been ascertained. YG hydrogels are by contrast stable in physiological buffers [33], and their sieve-like properties are consistent with the selective permeability of the peroxisomal membrane [43–46].

The peroxisomal translocon

A meshwork formed from multiple YG domains was shown to occur on peroxisomes in vivo using disulfide-mediated crosslinking [33]. By introducing single cysteines into the YG domain of PEX13 in yeast and treating the corresponding membranes with an oxidant, efficient dimerization of the protein was observed. Introducing two cysteines revealed a ladder of oligomeric forms, consistent with assembly of the YG domain into a dense and multivalent meshwork. Mutating the conserved tyrosines in the YG domain into serines caused the meshwork to collapse and consequently abolished import.

The YG meshwork is suspended in the peroxisomal membrane by PEX13 molecules of opposite orientations. Such dual topology is an unusual property among membrane proteins [47]. The configuration was demonstrated for PEX13 by incorporating a specific protease-cleavage site at either end of the protein in yeast [33]. Only half of the PEX13 molecules were accessible to the protease and thus had the cleavage site facing the cytosol. Disulfide-mediated crosslinking further demonstrated that the YG domains of both populations associate with one another [33]. The dual topology of PEX13 thus explains how the YG domains meet within the membrane to form a meshwork that spans the lipid bilayer. In the case of algal PEX13 (Fig. 2A), the two YG domains flanking the AH presumably meet inside the membrane too.

The dual topology of PEX13 reconciles a longstanding controversy about the protein’s orientation. Both the N and the C terminus of PEX13 could be immunolabeled in cells under conditions that only perforated the plasma membrane, suggesting that both termini faced the cytosol [31,32,48]. By contrast, protease-protection experiments with intact peroxisomes had identified a C-terminal fragment of the protein that was protected from digestion and thus presumably faced the matrix [49]. Considering that the protein actually resides in the membrane in two orientations, these previously contradictory results in fact reinforce each other.

The dual topology of PEX13 is established by the AH. The AH is the only other conserved element of the protein apart from the YG domain, and is likewise essential for import: mutations along either the hydrophilic or the hydrophobic face of the AH in yeast were shown to abolish protein import into peroxisomes [33]. In humans, a number of polymorphisms have been identified within the AH, some of which are predicted to be pathogenic (https://www.uniprot.org/uniprotkb/Q92968). While it remains unclear how the AH governs the insertion of PEX13 into the membrane in two orientations, the process involves the peroxisomal membrane protein insertion machinery epitomized by PEX3 [33].

The AH segments of multiple PEX13 molecules likely form the membrane-spanning wall that encircles the aqueous YG meshwork. A ring-like structure is indeed predicted by AlphaFold-Multimer [50] using 10 or 12 copies of the AH (Fig. 3). This number of subunits reflects the largest oligomeric state of PEX13 determined by disulfide-mediated crosslinking [33]. The AH segments in the predicted assembly are highly tilted, and are arranged in alternating orientations (Fig. 3A) consistent with the dual topology of PEX13. Presumably, the dual topology enforces proper packing of the AH segments.

Figure 3. Model for pore formation by the amphipathic helix of PEX13.

(A) Twelve copies of the amphipathic helix (AH; residues 141–214) from yeast PEX13 are predicted to form a ring by AlphaFold-Multimer (version 1.0). In the model, the helices are highly tilted and arranged in alternating orientations, which are colored light and dark green for two of the subunits (“N” and “C” designate their N and C termini, respectively). Side and oblique views of the ring assembly are shown, with the height and inner-diameter dimensions labeled. (B) The hydrophobic residues (orange) of all 12 helices are predicted to face the outside of the ring, which would correspond to the hydrophobic lipid environment of the membrane. The helices’ hydrophilic residues (cyan) are instead predicted to face the center, corresponding to the aqueous central channel. Side and oblique views are the same as in (A). (C) The scheme on the left shows a side view of the predicted PEX13 ring, illustrating how it would sit in the membrane to form a pore. The alternating orientations of the AH segments are colored different shades of green. Also shown are the alternating orientations of the downstream TM segment and the SH3 domain (outlined by dashed black lines) found in some PEX13 proteins. The scheme on the right shows a top view of the ring assembly, indicating how the YG domains (pink) from each subunit would form a dense meshwork inside the pore.

The predicted ring-like assembly of PEX13 resembles a membrane pore. The hydrophobic surface of each AH faces the outside (Fig. 3B), corresponding to the lipid environment; the hydrophilic side is instead oriented toward the center (Fig. 3B), corresponding to the aqueous channel filled by the YG meshwork (Fig. 3C). The inner diameter of the assembly ranges from 8–10 nm, which should be sufficiently wide to accommodate cargoes known to be imported into peroxisomes [15,16,51] together with the accompanying receptor. The height of the pore would also be compatible with the thickness of a lipid bilayer. However, this model remains speculative; the actual number and arrangement of AH segments that comprise the translocon are currently unknown.

The additional TM helix in some PEX13 proteins also adopts two transmembrane orientations [33]. This completely hydrophobic segment would presumably reside among the lipids outside the predicted pore (Fig. 3C). The downstream SH3 domain would thereby be positioned around the perimeter of the pore on either side of the membrane (Fig. 3C). The SH3 domains facing the matrix might promote the association of the translocon with PEX14 [52,53], whereas those facing the cytosol might instead recruit import receptors [29]. Given that the SH3 domain is only found in PEX13 from opisthokonts [1], import receptors might directly partition into the YG meshwork without a prior docking event.

The proposed nuclear pore-like translocon compared to other models

A recent study proposes that the PEX13 translocon assembles only during cargo translocation [34]. This conclusion is based on fluorescence correlation spectroscopy (FCS) experiments in live yeast cells, which show that PEX13 briefly coincides with incoming cargo in small patches on the peroxisomal surface. However, it remains unclear whether the observed coincidence reflects actual clustering of PEX13, or rather the more conservative interpretation that incoming cargo would necessarily associate with the translocon. While the PEX13 translocon might conceivably consist of a variable number of subunits, it seems unlikely that they would assemble de novo for each translocation event. The AH of PEX13 would not be expected to span the membrane as a monomer, because its hydrophilic surface would contact the hydrophobic lipids. The majority of PEX13 has in fact been demonstrated to be oligomeric at steady state by disulfide-mediated crosslinking [33], consistent with earlier data supporting PEX13 oligomerization [54–57]. The YG domain also forms oligomers and assembles into a meshwork in the absence of PEX5 [33], the predominant import receptor in yeast, although a possible role for the remaining import receptors cannot be readily excluded. Regardless, definitive visualization of a PEX13 pore by electron microscopy is still lacking.

An alternative to the nuclear pore-like mechanism envisions that PEX5 itself forms the translocon together with PEX14 [58]. According to this model, PEX5 would behave like a pore-forming toxin, transiently assembling within the peroxisomal membrane into a channel. While PEX5 can adhere to synthetic membranes in vitro [59,60], the absence of hydrophobic segments in the receptor befogs how it would insert into the lipid bilayer. PEX5 does contain several amphipathic helices that might interact with lipids [61], but only two of them (AH1 and AH2; see Fig. 1B) are actually required for import [26,62,63] and the remainder likely function instead to bind specific cargoes [64–67]. The two essential amphipathic helices in fact mediate receptor recycling and play no role in translocation across the membrane [62,68] (see the section below on receptor recycling for a more thorough discussion).

It is also unlikely that an aqueous conduit could be formed by PEX14. This peroxin contains only a single, completely hydrophobic transmembrane segment and forms stable rod-like bundles that project into the cytosol [69,70]. PEX14 moreover does not seem to be absolutely required for import in all organisms. Residual import of certain cargoes has been reported to occur in the yeast Hansenula polymorpha in the absence of PEX14 [71,72]; in an A. thaliana mutant that lacks detectable PEX14 mRNA and protein [73]; as well as in a mutant of the filamentous fungus Podospora anserina missing both PEX14 and its paralog PEX33 [74]. However, these studies lack a rigorous demonstration that the cargo was actually localized inside the matrix. By contrast to PEX14, the absence of PEX13 in these organisms abolishes import of all cargoes that have been tested [74,75].

A beguiling exception appears to be the import of PEX8. This peroxin only occurs in fungi [1] and is localized within peroxisomes [76]. In the absence of PEX13, some of the peroxisome-localized PEX8 was shown to still be protease-protected and thus assumed to be inside the matrix [77], suggesting that PEX13 may be dispensable for PEX8 import. However, protease protection does not immediately imply localization within the matrix. In fact, the lack of PEX13 potently induces pexophagy [78] and in yeast, peroxisomes devoid of PEX13 are enclosed by multiple membranes [79]. It is thus conceivable that the additional membranes prevent access of exogenously added protease to PEX8 stuck on the cytosolic surface of peroxisomes.

Interestingly, PEX8 consists almost entirely of alpha-helical motifs resembling ARM and HEAT repeats. These protein domains notably occur in many nuclear transport receptors, where they mediate the interaction with the nucleoporin FG meshwork inside nuclear pores [80]. Perhaps, by analogy to nuclear transport, PEX8 can directly partition into the YG meshwork within the PEX13 conduit and thereby enter peroxisomes. The reported interaction of PEX8 with peroxisomal import receptors [76,81–84] might facilitate translocation [77,85] or prevent retrograde diffusion of PEX8 back into the cytosol.

Biased diffusion of PEX5 into the matrix

Diffusion of PEX5 through the YG meshwork should in principle be bidirectional, analogously to how nuclear transport receptors traverse the FG meshwork inside nuclear pores. The directionality of nuclear transport is established by the Ran•GTP/GDP gradient between the nucleoplasm and cytoplasm [86]. Peroxisomes rely instead on an interaction between PEX5 and PEX14 inside the matrix to bias diffusion inward (Fig. 1A, step 4). The WxxxF/Y motifs of PEX5 bind with high affinity to an N-terminal domain of PEX14 [87–89], which has by now been well established to face the matrix [33,49,62,90–92]. This interaction is likely potentiated by avidity, as PEX5 proteins usually contain multiple WxxxF/Y motifs [1,93] and PEX14 forms oligomers [69,91,94,95]. The favorable interaction with PEX14 would outcompete the low-affinity interaction with the YG meshwork, and evict PEX5 from the meshwork on the lumenal side. Inward diffusion might be further biased by lumenal interactions between PEX5 and other import components such as PEX13 [29] or in yeast, PEX8 [54].

PEX5 completely enters the matrix

While PEX5 has long been known to reach into the matrix during import [96], whether the receptor completely enters peroxisomes has been contentious. Recent data reveal that the entire PEX5 molecule, and not just the WxxxF/Y motifs or the cargo-binding domain, transits through the matrix during import. Pivotal for reaching this conclusion was a novel cell-free system that recapitulates matrix protein import in Xenopus egg extract [97,98]. This system allows PEX5 to be replaced with mutants containing a cleavage site for a peroxisomal matrix protease at different positions (N terminus, middle, or C terminus). All mutants were shown to be cleaved [62], revealing that the entire PEX5 molecule had accessed the matrix. A similar strategy was originally used to demonstrate lumenal entry of the receptor’s N terminus in mammalian cells [99]. Complete lumenal entry of PEX5 is reinforced by protease-protection experiments (see below), and agrees with previous studies demonstrating that PEX7, the PTS2-cargo adapter, also enters the matrix [100–103].

PEX5 recycling through the ubiquitin ligase complex

To return from the matrix back into the cytosol, PEX5 must again cross the membrane. A clue to how this step might occur was obtained in the Xenopus system using protease-protection experiments, which identified a prominent retrotranslocation intermediate of PEX5 in a transmembrane orientation [62]. The first 20–30 amino acids of PEX5 in this state were accessible to exogenously-added protease and thus faced the cytosol, whereas the remainder, including the C-terminal folded TPR domain, were protected from proteolysis and thus located inside the matrix. Importantly, the abundance of this transmembrane state was increased by mutating the receptor’s conserved cysteine needed for recycling, revealing that this state corresponded to a stalled recycling intermediate. A PEX5 intermediate with a similar orientation has been described previously in rat-liver peroxisomes [104,105].

The identity of the retro-translocon was revealed by immunoprecipitating the intermediate and analyzing bound proteins by mass spectrometry [62]. Major interacting partners turned out to be all three components of the PEX2–10-12 ubiquitin ligase complex. The interaction with the ligase was disrupted by mutating a conserved amphipathic helix (AH2; see Fig. 1B) in the N-terminal unstructured region of PEX5. In the absence of this helix, PEX5 became trapped inside the matrix [62]. Similarly, a PEX5 mutant lacking the region encompassing AH2 had been shown to enter rat-liver peroxisomes, but could not return to the cytosol [68]. AH2 is thus an export signal for PEX5 that allows the receptor’s N terminus, including the conserved cysteine, to emerge in the cytosol for monoubiquitination. Receptors for PTS2 cargo likely use the same mechanism, as they have comparable features to PEX5 [22,62] and pass through a similar protease-protected recycling intermediate [106].

A recent cryo-EM structure of the ubiquitin ligase complex shows that it indeed forms a retrotranslocon [107,108]. Each subunit of the complex contributes five transmembrane segments that assemble into a channel with an open pore approximately 10 Å in diameter (Figs. 4A and B). The three RING fingers (RFs) form a tower above the pore on the cytosolic side. The channel is likely constitutively open because it seems to have rigid walls and lacks features that could plug the pore. The channel also lacks a lateral opening toward the surrounding lipids. PEX5 can therefore access the RFs only by moving from the matrix through the channel and into the cytosol (Fig. 4A), rather than sideways from the membrane. Consistent with this translocation path, peroxisomal import is compromised by mutations in the complex that occlude the pore [107]. However, direct evidence that receptors transit through the pore is still lacking.

Figure 4. Structure of the peroxisomal ubiquitin ligase complex.

(A) Surface model of the structure of the ubiquitin ligase complex from Thermothelomyces thermophilus, solved by cryo-EM (PDB accession code 7T92). A cross-sectional view through the complex is shown, with the three subunits (PEX2, PEX10, and PEX12) differently colored. The complex consists of two domains: a transmembrane portion that is shown embedded in the peroxisomal membrane with the cytosolic and matrix sides labeled; and a cytosolic tower formed from each subunit’s RING finger. The RING finger of PEX2 (RF2) is colored, and the surface map is overlaid by a cartoon model of RF2. The complex contains an open pore spanning the membrane, through which PEX5 is likely retrotranslocated from the matrix into the cytosol. (B) Scheme illustrating the overall architecture of the complex, with each subunit differently colored and the probable retrotranslocation path of PEX5 delineated by a blue arrow. Shown is a cross-sectional view as in (A). (C) Views of the complex from the cytosolic side and from within the peroxisomal matrix. Shown is a surface model with each subunit colored as in (A), and the open pore indicated. The black line across the pore represents a loop which is invisible in the density map.

Receptor monoubiquitination during recycling is likely mediated by the RF of PEX2, which sits right above the ligase pore whereas the RFs of the other two subunits (PEX10 and PEX12) are more distant (Fig. 4A). The monoubiquitination mechanism remains poorly understood, however, as does the reason why import receptors are monoubiquitinated on a cysteine. Ubiquitination of a cysteine in a protein is unusual [109]; the formation of a ubiquitin thioester bond is predominantly found as a transient modification in ubiquitin-activating and ubiquitin-conjugating enzymes [110].

The RFs of PEX10 and PEX12 function together in a separate polyubiquitination pathway, which kicks in when PEX2-mediated recycling is blocked [107]. Polyubiquitination provides a backup mechanism for retrieving stalled import receptors and targets them for degradation. This pathway has accordingly been termed RADAR for Receptor Accumulation and Degradation in the Absence of Recycling [111], and has been reviewed elsewhere [108].

Extraction of PEX5 into the cytosol

Monoubiquitinated PEX5 is next extracted from the matrix into the cytosol by the PEX1-PEX6 ATPase complex (Fig. 1A, step 6). PEX1 and PEX6 each consist of two ATPase domains that co-assemble into two stacked hexameric rings [112–114] (Fig. 5). Each subunit additionally contains two N-terminal domains (Fig. 5A) that likely recognize monoubiquitinated PEX5 [115], although it remains unclear whether the interaction involves a cofactor. The receptor is then likely translocated through both ATPase rings by processive threading [116,117] (Figs. 5B and C), which resembles substrate processing by other hexameric ATPases [118]. Translocation would pull PEX5 out of the matrix through the pore of the ubiquitin ligase complex.

Figure 5. Structure of the PEX1-PEX6 AAA ATPase.

(A) Domain organization of PEX1 and PEX6. Both proteins contain two N domains (N1 and N2) and two AAA ATPase domains (D1 and D2); the N1 domain of PEX1 (dashed black outline) is invisible in the cryo-EM structure [112]. The D1 and D2 ATPase domains of PEX1 and PEX6 co-assemble into two stacked hexameric rings (the D1 ring and the D2 ring, respectively); each ring consists of alternating copies of the two subunits (right panels). (B) Structural model of the ATPase complex from S. cerevisiae. The model was generated by fitting AlphaFold predictions of yeast PEX1 and PEX6 (accession codes AF-P24004-F1 and AF-P33760-F1, respectively) into the cryo-EM density map of the intact complex (EMDB accession code EMD-6359). The side view of the complex on the left is rendered as a surface, overlaid by cartoon models of two of the subunits, with their D1 and D2 domains colored as in (A); the other subunits are not colored. The N domains are outlined in black, but are otherwise transparent. On the right is a cross-sectional view, indicating the open pore through the middle of both ATPase rings. The blue arrow designates the path by which an import receptor would be translocated through the complex. (C) Views of the complex from the top (D1) side and the bottom (D2) side. All six subunits are shown as surface models, with their D1 and D2 domains colored as in (A).

PEX5 is likely unfolded by the PEX1-PEX6 ATPase during recycling, as the pore of the ligase complex is too narrow to accommodate a folded domain. Experiments in the Xenopus system indeed support PEX5 unfolding [62]. GFP lacking a peroxisome targeting signal was shown to accumulate in peroxisomes in the presence of PEX5 fused to a GFP-binding nanobody, whereas no accumulation was observed with a PEX5 mutant unable to be recycled. The interaction between GFP and the nanobody must therefore have been broken, likely by unfolding of the nanobody. Unfolding must have occurred during each import cycle to allow GFP to accumulate in the matrix. The receptor’s TPR domain must also have been unfolded because it resided upstream of the nanobody in the fusion protein. This conclusion is braced by experiments with rat-liver peroxisomes, which showed that cysteines buried within the TPR domain become exposed during recycling [115] and that recycling is impeded after fusing PEX5 to tightly folded domains [115,119]. These results imply that pulling by the PEX1-PEX6 ATPase unfolds the cargo-binding TPR domain, causing cargo to be stripped off and left inside the organelle as the receptor emerges in the cytosol. The TPR domain presumably refolds in the cytosol to reset the receptor for the next import cycle.

A recent study reveals that the ubiquitin conjugated to PEX5 is also unfolded during extraction [120]. This result was demonstrated in a cell-free system [121] using ubiquitin that contained a buried cysteine [122]. When a thiol-reactive maleimide reagent was added during PEX5 extraction, the ubiquitin conjugated to PEX5 became modified and must therefore have been unfolded. By contrast, little modification was observed when the reagent was added after extraction, suggesting that the ubiquitin had refolded once the receptor was released into the cytosol. Refolding would be necessary for the ubiquitin to be recognized and removed by cytosolic deubiquitinases [123,124]. PEX5 extraction is accordingly blocked by ubiquitin that cannot be unfolded because of an intramolecular crosslink [120].

The driving force for translocation

Translocation of PEX5 into the matrix does not require nucleotide hydrolysis [125,126]. The receptor is drawn into the matrix by a strong lumenal interaction with PEX14 (and possibly other peroxins), as discussed above. Sustained import thus requires these lumenal binding sites to be continually vacated by the ATP-dependent retrieval of PEX5 from the matrix, as suggested previously [126,127]. This mechanism explains how cargo proteins can be concentrated inside the matrix against their chemical gradient.

Since all of the features required for PEX5 recycling occur near the receptor’s N terminus [62,68], the PEX1-PEX6 ATPase could conceivably start extracting the receptor before the TPR domain has fully entered the matrix. The remainder of PEX5 could thus be pulled into the matrix together with cargo, as envisioned by the export-driven import model [106,127]. While this mechanism might not be necessary for import of all cargoes [102,128], it could facilitate the translocation of larger cargoes whose passage through the YG meshwork might be energetically unfavorable.

Whether PEX5 translocation can occur in the absence of cargo is unclear. Such futile cycling would be wasteful, as it would consume ATP without concomitant cargo import. The association of PEX5 with peroxisomes has been reported to require cargo proteins [129], suggesting that the receptor might be autoinhibited in their absence. However, this property does not appear to be recapitulated in purified PEX5, which can readily interact in vitro with the PEX13 YG phase [33,34] and with the lumenal domain of PEX14 [93].

Peroxisomal protein import compared to other protein-translocation systems

Translocation into peroxisomes most resembles nuclear transport. In both cases, mobile receptors shuttle folded cargo through a selective phase, which is formed from unstructured protein domains and held together by cohesive interactions between aromatic residues (Fig. 2C). However, an important distinction is that the peroxisomal phase resides within the membrane and forms an aqueous pore that is shielded from the surrounding lipids (Fig. 6A). The nuclear-pore phase is by contrast suspended in the aqueous space between the nucleoplasm and the cytoplasm (Fig. 6B).

Figure 6. Peroxisomal matrix protein import compared to other protein-translocation mechanisms.

(A) The peroxisomal conduit is formed by multiple copies of PEX13 (green), and is filled by a dense meshwork formed from the subunits’ YG domains (pink). Note how the conduit is embedded directly in the membrane. The peroxisomal import receptor PEX5 diffuses through the meshwork together with bound cargo. (B) The nuclear pore (light gray) is similarly filled by a meshwork formed from nucleoporin FG repeats (dark gray), through which nuclear transport receptors (peach-colored) enter and exit the nucleus together with bound cargo. Note how the nuclear pore sits outside the membrane at the highly curved edges of the nuclear envelope. (C) The peroxisomal ubiquitin ligase complex forms a retrotranslocon with a constitutively open pore spanning the membrane, through which PEX5 is recycled into the cytosol. (D) During ER-associated protein degradation (ERAD), a protein substrate (peach-colored) is retrotranslocated into the cytosol through a complex containing Hrd1 and Der1, which form “half-channels” open either to the cytosol or to the lumen, respectively. These half-channels are connected by a thinned membrane region, through which substrates pass as a loop. (E) Proteins are translocated into the ER as unfolded polypeptides through the Sec61 translocon, which maintains a tight seal (dark gray ring) during translocation and has a plug domain that opens only in response to a substrate. Arrows designate the direction of translocation in each panel.

Peroxisomal import conceptually also resembles the twin-arginine protein translocation (Tat) system, which transports folded proteins across the bacterial plasma membrane and the thylakoid membrane of chloroplasts. However, the Tat system relies on membrane thinning to lower the energy required for hydrophilic proteins to traverse the hydrophobic lipid bilayer [130,131], as opposed to the aqueous pore used by peroxisomes.

Retrotranslocation of peroxisomal import receptors resembles ER-associated protein degradation (ERAD). In both cases, substrate proteins cross the membrane through a multi-spanning ubiquitin ligase that ubiquitinates them on the cytosolic side, and are subsequently extracted by a AAA ATPase. However, the peroxisomal ligase complex forms an open channel (Fig. 6C) whereas ERAD ligases do not form channels at all. As revealed by the cryo-EM structure of the Hrd1 ubiquitin ligase complex [132], the membrane components Hrd1 and Der1 form “half-channels” open either to the cytosol or to the lumen, respectively (Fig. 6D). These half-channels are connected by a thinned membrane region through which substrates pass as a loop [133].

The peroxisomal ubiquitin ligase complex also resembles the classic Sec61 translocon (SecY in prokaryotes), which mediates protein translocation across the ER membrane (Fig. 6E). In both cases, polypeptides transit through an aqueous pore. Whereas the peroxisomal ubiquitin ligase complex has a constitutively open pore, the pore of Sec61 is initially closed by a plug and opens only upon arrival of a substrate protein [17]. Sec61 additionally has a lateral gate that allows hydrophobic segments to escape sideways into the surrounding lipids, whereas a similar gate is absent in the peroxisomal retrotranslocon.

The peroxisomal import and export mechanisms would not work in other membranes. Both the PEX13 conduit and the PEX2-PEX10-PEX12 retrotranslocon allow the free flow of ions and small solutes, which is consistent with the known permeability of peroxisomes to molecules below 800 Da [44]. In contrast, other organelles must sustain ionic or chemical gradients across their membranes [134–137], and therefore require mechanisms that maintain a tighter seal during protein translocation.

Concluding remarks

Elucidating the mechanism of peroxisomal matrix protein import has revealed two novel ways by which proteins are moved across membranes. The recent advances have clarified the longstanding question of how folded proteins are translocated into peroxisomes, and have led to a more comprehensive understanding of the pathway. Many questions remain unresolved (see Outstanding Questions), which are bound to orient future research into this fascinating topic. Crucially, this mechanistic insight will help illuminate how mutations in import components cause disease, and hopefully accelerate the development of appropriate therapies.

Outstanding questions.

• What is the structure of the peroxisomal translocon? Structural modeling suggests a plausible arrangement of PEX13 molecules into a channel filled by the YG meshwork, but the actual structure remains to be determined.

• How are import receptors recruited to peroxisomes? Do receptors partition directly into the YG meshwork, or do they initially engage other domains of the import machinery that face the cytosol?

• What is the mechanism of receptor monoubiquitination? Does the receptor transit through the pore of the peroxisomal ubiquitin ligase complex, and can this process be reconstituted in vitro using purified components?

• How are monoubiquitinated import receptors recognized by the PEX1-PEX6 ATPase? Might an unidentified cofactor be involved?

• How does PEX7 return to the cytosol? The adapter is brought into the matix by import receptors together with cargo, but lacks the receptors’ unstructured segments and conserved cysteine. It therefore cannot use the ubiquitin ligase as a retrotranslocon and must recycle by a different mechanism.

• What is the function of PEX14? PEX14 forms stable bundles in the membrane and projects a long coiled rod into the cytosol. The role of this cytosolic domain remains enigmatic.

• How are the membrane components of the import machinery inserted into the peroxisomal membrane? How is the dual topology of PEX13 established?

Highlights.

• Peroxisomes have the unusual ability to import folded or oligomeric proteins, but the underlying mechanism has been mysterious.

• Recent studies reveal that peroxisomal import resembles nuclear transport: mobile receptors shuttle cargo into the peroxisomal lumen through a membrane pore that is filled by a meshwork serving as a selective phase.

• The meshwork is formed by the YG domain of the peroxisomal membrane protein PEX13, and is suspended in the membrane by multiple PEX13 molecules arranged in alternating orientations.

• The receptors are pulled out of the lumen through a retrotranslocon and thereby unfolded, causing cargo to be stripped off and left behind inside the organelle.

• Peroxisomal import is thus driven by the retrieval of import receptors out of the organelle.