A peroxisomal C-tail-anchored type-II membrane protein, Pex26p, recruits AAA ATPase Pex1p–Pex6p complexes to peroxisomes. We herein attempted to gain mechanistic insight into Pex26p function. Pex26pΔ33-40 truncated in amino-acid residues at 33-40 abolishes the recruiting of Pex1p–Pex6p complex to peroxisomes and fails to complement the impaired phenotype of pex26 CHO cell mutant ZP167, thereby suggesting that peroxisomal localization of Pex1p and Pex6p is indispensable for the transport of matrix proteins. In in vitro transport assay using semipermeabilized CHO cells, Pex1p is targeted to peroxisomes in a manner dependent on ATP hydrolysis, while Pex6p targeting requires ATP but not its hydrolysis. This finding is confirmed by the assay using Walker-motif mutants. Transport of Pex1p and Pex6p is temperature-dependent. In vitro binding assays with glutathione-S-transferase-fused Pex26p, Pex1p and Pex6p bind to Pex26p in a manner dependent on ATP binding but not ATP hydrolysis. These results suggest that ATP hydrolysis is required for stable localization of Pex1p to peroxisomes, but not for binding to Pex26p. Moreover, Pex1p and Pex6p are altered to a more compact conformation upon binding to ATP, as verified by limited proteolysis. Taken together, Pex1p and Pex6p are most likely regulated in their peroxisomal localization onto Pex26p via conformational changes by the ATPase cycle.
Peroxisome, a single membrane-bounded ubiquitous organelle, is present in a wide variety of eukaryotic cells from yeast to humans, and functions in various metabolic pathways, including β-oxidation of very long-chain fatty acids and the synthesis of ether lipids (1).
Up to now, we and other groups have revealed 14 different complementation groups (CGs) in peroxisome-deficient phenotypes in mammals, by CG analysis between peroxisome-deficient CHO cell mutants and fibroblasts from patients with peroxisome biogenesis disorders (PBDs). Therefore, at least 14 genes (PEX) or their products, termed peroxins, are required for peroxisome biogenesis in mammals (2–6). Import of peroxisomal matrix proteins is mediated by two types of peroxisomal-targeting signals (PTSs): PTS1, identified in numerous enzymes (7–9), and PTS2, present in several proteins such as 3-ketoacyl-CoA thiolase (thiolase) (10,11). In the cytoplasm, Pex5p and Pex7p function as the PTS1 and PTS2 receptors, respectively, and translocate respective PTS-cargoes into peroxisomes via peroxisomal membrane importers (12–15). In contrast to the molecular mechanisms underlying the matrix protein import to peroxisomes, those of peroxisome membrane biogenesis are less understood (16–18).
PEX26 is identified the most recently identified gene responsible for PBDs of 13 CGs (2,19). PEX26 encodes a 34-kDa type-II peroxisomal membrane protein with one C-terminal transmembrane segment, exposing its N-terminal domain to the cytosol (2). Pex26p interacts with Pex1p–Pex6p complexes and recruits them to peroxisomes (2,20–22). Yeast Pex15p, a functional homolog of mammalian Pex26p, binds Pex6p (23) and forms a complex with Pex1p and Pex6p (24). Pex1p and Pex6p are members of the AAA ATPase family (5,25–29) and show a dual subcellular localization, mostly in peroxisomes and partly in the cytosol (21,24,30,31). Pex1p and Pex6p are required for the export of Pex5p from peroxisomes. However, it remains unclear how the AAA peroxins export Pex5p to the cytosol. Furthermore, ATP requirements for the Pex1p–Pex6p complex and targeting mechanism of Pex1p and Pex6p to peroxisomes remain obscure.
As a further step toward understanding the molecular mechanisms underlying the targeting of Pex1p and Pex6p to Pex26p and its regulation, we here developed an in vitro assay system for targeting of Pex1p and Pex6p to Pex26p using semi-intact cells. Moreover, we provide several lines of evidence that peroxisomal targeting of Pex1p and Pex6p takes place in an ATP-dependent manner and that nucleotide binding to Pex1p and Pex6p triggers their conformational changes.
Region of Pex26p required for peroxisomal matrix protein import and peroxisomal localization of Pex1p–Pex6p complexes
Pex1p and Pex6p are mostly localized to peroxisomes and partly in the cytosol (21). Pex26p, the recruiter of Pex1p–Pex6p complexes to peroxisomes, is a C-tail-anchored peroxisomal membrane protein, extruding its N-terminal domain to the cytosol. Therefore, the N-terminal cytosolic domain of Pex26p is likely involved in binding to the Pex1p–Pex6p complexes.
We first attempted to define the region of Pex26p required for peroxisomal localization of the Pex1p–Pex6p complex. We constructed several N-terminally truncated Pex26p variants and verified their activity in recruiting the Pex1p–Pex6p complexes in pex26 ZP167 cells. Hemagglutinin (HA)-tagged Pex26p variants, Flag-tagged Pex1p and Myc-tagged Pex6p were coexpressed in ZP167 cells and subcellular localization of Flag-Pex1p and Myc-Pex6p was verified by immunofluorescent staining with respective antibodies. In cells expressing wild-type Pex26p-HA or a Pex26p mutant truncated in amino-acid residues at positions 1-32, termed Pex26pΔ1-32-HA, Flag-Pex1p and Myc-Pex6p were detected in a manner superimposable on Pex14p-positive particles, hence indicating that Flag-Pex1p and Myc-Pex6p were localized to peroxisomes (Figure 1A, a–h; 1B, b–d and f–h). In contrast, Flag-Pex1p and Myc-Pex6p failed to translocate to peroxisomes and remained in the cytosol in ZP167 cells expressing Pex26pΔ1-40-HA or Pex26pΔ33-40-HA, where Pex26pΔ1-40-HA and Pex26pΔ33-40-HA were normally localized to peroxisomes (Figure 1A, i–p; 1B, j–l and n–p). These results suggested that Pex1p and Pex6p were transported to peroxisomes in a Pex26p-dependent manner and that amino-acid residues at 33-40 of Pex26p were required for peroxisomal localization of Pex1p and Pex6p. Next, we assessed the complementing activity of these Pex26p variants in pex26 ZP167 cells. In ZP167 cells expressing wild-type Pex26p-HA or Pex26pΔ1-32-HA, catalase-, PTS1- and PTS2 thiolase-positive particles were discernible (Figure 1B, a and e; 1C, b and f; 1D, b and f), suggesting that Pex26pΔ1-32-HA was as active as Pex26p-HA in restoring the impaired import of peroxisomal matrix proteins. In contrast, Pex26pΔ1-40-HA and Pex26pΔ33-40-HA did not restore peroxisomal matrix protein import (Figure 1B, i and m; 1C, j and n; 1D, j and n). Together, these data showed that amino-acid residues at 33-40 of Pex26p were essential for peroxisomal matrix protein import as well as for the recruitment of Pex1p and Pex6p to peroxisomes.
Pex1p–Pex6p complexes translocate to Pex26p in semi-intact cells
To verify that Pex26p recruits cytosolic Pex1p and Pex6p to peroxisomes, we set up an in vitro targeting assay system using semi-intact cells. pex26 ZP167 cells expressing EGFP-Pex26p or a 22-kDa peroxisomal integral membrane protein fused to EGFP (PMP22-EGFP) were semipermeabilized with digitonin and incubated with the cytosolic fraction from pex26 ZP167 cells that had been transfected with Flag-PEX1, PEX6-HA or Flag-PEX1 plus PEX6-HA. Pex6p-HA or Flag-Pex1p plus Pex6p-HA were detected in a punctate staining pattern and in a manner superimposable on EGFP-Pex26p (Figure 2A, e–l), while Flag-Pex1p expressed alone was not detectable on EGFP-Pex26p-positive particles (Figure 2A, a–d). In mock-transfected and PMP22-EGFP-expressing cells, neither Flag-Pex1p nor Pex6p-HA was detectable in a punctate staining pattern (Figure 2A, m–t). Moreover, this finding was confirmed by immunoblot analysis of the cells assayed above (Figure 2B). We also performed the in vitro targeting assays using other CHO cell mutants, pex1 ZP107 and pex6 ZP164 cells, as performed with pex26 ZP167. Nearly the same results were obtained in these assays, suggesting that endogenous Pex1p and Pex6p less likely affected the overexpressed Pex1p and/or Pex6p (data not shown). Collectively, we conclude that Pex26p recruits Pex1p and Pex6p via Pex26p–Pex6p and Pex6p–Pex1p interactions, consistent with our earlier findings in in vivo expression and cell-free binding assay (2).
Cytosolic factors enhance peroxisomal targeting of Pex1p and Pex6p
As a further step to understanding the molecular mechanisms underlying peroxisomal targeting of Pex1p and Pex6p mediated by Pex26p, we attempted to express and purify recombinant Pex1p and Pex6p using an Sf9/baculovirus expression system. Flag-Pex1p and Flag-Pex6p were purified nearly to the homogeneity (Figure 3A). We performed an in vitro targeting assay using purified Pex1p and Pex6p. After separate incubations of the purified Flag-Pex1p and Flag-Pex6p with the semi-intact ZP167 cells expressing EGFP-Pex26p in the presence of ATP and an ATP-regenerating system (ARS), Flag-Pex6p coincided with EGFP-Pex26p (Figure 3B, a–c), while Pex1p was barely discernible in the cells (Figure 3B, d–f). In contrast, upon incubation in the presence of Pex6p, Flag-Pex1p was detectable in a pattern superimposable on EGFP-Pex26p, not PMP22-EGFP, in semipermeabilized cells expressing EGFP-Pex26p or PMP22-EGFP (Figure 3B, g–l). Thus, Pex26p more likely anchors Pex6p and Pex1p in a Pex6p-dependent manner.
Next, to investigate whether any cytosolic factors are required for efficient targeting of Pex1p and Pex6p to Pex26p, we likewise carried out the transport assay using semi-intact cells expressing EGFP-Pex26p with mixtures of the purified recombinant Flag-Pex1p and/or Flag-Pex6p in the presence or absence of the cytosolic fraction. Flag-Pex6p and Flag-Pex1p plus Flag-Pex6p were specifically transported to Pex26p, peroxisomes, apparently with a moderate increase in Flag-Pex6p and over twofold elevated level of Flag-Pex1p in the presence of cytosolic fraction (Figure 3C, left panel, lanes 6 and 7; right panel), as compared to the assay performed in the absence of the cytosolic fraction (Figure 3C, left panel, lanes 2 and 3; right panel). However, no such enhanced targeting was evident using a heat-treated cytosolic fraction (Figure 3C, left panel, lane 9; right panel), hence suggesting that protein factors in the cytosol are involved in the targeting of Pex1p and Pex6p.
To verify whether molecular chaperones are involved in peroxisomal targeting of Pex1p and Pex6p, HSP70 was depleted from the ZP167-derived cytosolic fraction containing Pex1p-HA and Pex6p-HA by immunodepletion with anti-HSP70 antibody. After the depletion step, HSP70 was barely detectable in the cytosolic fraction where the level of Pex1p-HA and Pex6p-HA was not altered (Figure 3D, left panel, lanes 1 and 2), hence suggesting that HSP70 did not interact with Pex1p and Pex6p. In the targeting assay using the HSP70-depleted cytosol and semi-intact ZP167 cells expressing EGFP-Pex26p, the transport of Pex1p-HA and Pex6p-HA to EGFP-Pex26p was not affected (Figure 3D, left panel, lanes 3 and 4), thereby suggesting that HSP70 is not involved in the localization of Pex1p and Pex6p to Pex26p. Next, we prepared the cytosolic fraction from Pex1p-HA- and Pex6p-HA-expressing ZP167 cells that had been treated at 42°C for 30 min, under which conditions HSP70 and HSP40, but not Pex1p-HA and Pex6p-HA, were induced (Figure 3D, right panel, lanes 1 and 2). In the assay using the heat-shocked cytosolic fraction and semi-intact ZP167 cells expressing EGFP-Pex26p, Pex1p-HA and Pex6p-HA were transported to peroxisomes nearly at the same level as compared to those with normal cytosolic fraction (Figure 3D, right panel, lanes 3 and 4). These results also suggest that heat shock proteins such as HSP70 and HSP40 are not involved in recruiting Pex1p and Pex6p to Pex26p.
Pex26p recruits Pex1p and Pex6p to peroxisomes in an ATP- and temperature-dependent manner
AAA ATPases undergo conformational changes via their nucleotide-binding states (32–36). However, little is known about how nucleotides alter conformations and functions of the AAA peroxins, Pex1p and Pex6p. We first verified whether Pex1p and Pex6p require ATP for targeting to Pex26p on peroxisomes. Semi-intact pex26 ZP167 cells expressing EGFP-Pex26p were incubated with a cytosolic fraction derived from pex26 ZP167 cells expressing Flag-Pex1p and Pex6p-HA, in the presence of ATP or adenylyl imidodiphosphate (AMP-PNP) or in the absence of any external nucleotides. To deplete endogenous ATP, the cytosolic fraction was also pretreated with apyrase and added to the assay in place of the cytosol. In the presence of externally added ATP, the targeting of Flag-Pex1p and Pex6p-HA to peroxisomes in a manner superimposable on EGFP-Pex26p was significantly elevated (Figure 4A, a–h). In contrast, under the conditions where ATP was depleted by treatment with apyrase, Flag-Pex1p and Pex6p-HA were no longer detectable on EGFP-Pex26p-positive peroxisomes (Figure 4A, m–p). Thereby, these results strongly suggested that Pex1p and Pex6p were targeted to peroxisomes in an ATP-dependent manner. Furthermore, in the Flag-Pex1p plus Pex6p-HA transport assay in the presence of AMP-PNP, a non-hydrolyzable ATP analog, Pex6p was targeted to peroxisomes but Pex1p failed to localize to peroxisomes (Figure 4A, i–l). Taken together, we interpreted these findings to mean that Pex1p requires ATP hydrolysis for its localization to peroxisomes and that Pex6p requires ATP, but not its hydrolysis, for targeting to peroxisomes. Immunoblotting analysis of the assayed cells confirmed these findings, where the Flag-Pex1p and Pex6p-HA transport was enhanced twofold to threefold in the presence of ATP but it was severely lowered by ATP depletion (Figure 4B).
To verify whether Pex1p and Pex6p are targeted to Pex26p in a temperature-dependent manner, we performed in vitro targeting assays using semi-intact cells at 4°C as well as 26°C. In the presence of ATP, the targeting of Pex1p-HA and Pex6p-HA to EGFP-Pex26p was significantly elevated at 26°C as compared to that at 4°C (Figure 4C, left panel, lanes 2 and 4; right panel). Such temperature-dependent enhancement of the targeting of Pex1p-HA and Pex6p-HA was also evident in the absence of any externally added ATP (Figure 4C, left panel, lanes 1 and 3; right panel). Thus, it is more likely that endogenous and/or externally added ATP is efficiently hydrolyzed by AAA peroxins at a higher temperature and that not only ATP binding but also ATP hydrolysis is important for stable ternary complex formation.
Mutation of the Walker motif affects peroxisomal targeting of Pex1p and Pex6p
Pex1p and Pex6p contain two AAA cassettes, so-called D1 and D2 domains, including the Walker A and B motifs for ATP binding and ATP hydrolysis, respectively (37,38). To determine whether these cassettes are essential for peroxisomal targeting of the AAA peroxins, we performed in vitro targeting assays using their variants mutated in either the Walker motif A or B of the first (D1) or second (D2) AAA cassette. These Walker mutants were verified for peroxisomal targeting in the semi-intact cell system as in Figure 2B. A Pex6p-HA mutant with substitution of D with N at 803 in the Walker B (D803N), termed Pex6pB2-HA, was transported to peroxisomes nearly at the same level as compared with wild-type Pex6p-HA (Figure 5A, upper panel, lanes 1 and 4; lower panel), while the Walker A mutants each harboring K476E and K750E, named Pex6pA1-HA and Pex6pA2-HA, respectively, showed a reduced level of targeting to EGFP-Pex26p. Moreover, wild-type Pex1p-HA coexpressed with Pex6pA1-HA or Pex6pA2-HA was scarcely transported to EGFP-Pex26p (Figure 5A, upper panel, lanes 2 and 3; lower panel). These results suggested that Pex6p is required to be in an ATP-bound form on its D1 and D2 domains for peroxisomal targeting of Pex1p–Pex6p complexes, consistent with our earlier findings in binding assays including a mammalian two-hybrid system (21).
Targeting to EGFP-Pex26p of Pex1p-HA mutant harboring D940N in the Walker B2, termed Pex1pB2-HA, was normal or slightly increased as compared to wild-type Pex1p, while a Walker B1 mutant with D662N, named Pex1pB1-HA, showed a lowered level of targeting to EGFP-Pex26p (Figure 5C, upper panel, lanes 1, 3 and 5; lower panel). Thus, ATP hydrolysis by Pex1p B1 site, not B2 site, is likely essential for peroxisomal targeting of Pex1p. Walker A mutants of Pex1p-HA each harboring K605E and K887E, termed Pex1pA1-HA and Pex1pA2-HA, respectively, were significantly reduced in translocation to EGFP-Pex26p (Figure 5C, upper panel, lanes 2 and 4; lower panel), thereby suggesting that ATP binding to D1 and D2 domains of Pex1p is indispensable for peroxisomal localization, as noted in Pex6p. Moreover, in morphological analysis, the Walker B2 mutants, Pex6pB2-HA and Flag-Pex1pB2, were detectable in a manner superimposable on EGFP-Pex26p-positive particles, peroxisomes (Figure 5B, i–l; 5D, m–p), as for the wild-type Pex6p-HA and Flag-Pex1p (Figure 2A). The other Walker mutants including Pex6pA1, Pex6pA2, Pex1pA1, Pex1pB1 and Pex1pA2 that showed the reduced level of interaction with Pex26p (Figure 5A,C) were also discernible but lesser in number of puncta that were superimposable on EGFP-Pex26p-positive peroxisomes (Figure 5B, a–h; 5D, a–l). We interpreted these results to mean that these mutants were only partly translocated to peroxisomes and largely remained in the cytosol. In semipermeabilized cells, these mutants in the cytosolic fraction were readily washed out and only those targeted to peroxisomes were visible in a punctate staining pattern. Indeed, we earlier reported that these mutants were mostly diffused in the cytosol in vivo(21). Collectively, these results from the in vitro targeting assays strongly suggested the importance of the AAA cassettes of both Pex1p and Pex6p for the targeting to and stable interaction with Pex26p, in good agreement with the finding by mammalian two-hybrid assays (21).
Nucleotide-dependent binding of the AAA peroxins, Pex1p and Pex6p, to Pex26p
Peroxisomal integral membrane protein, Pex26p recruits Pex1p–Pex6p complex to peroxisomes via direct interaction with Pex6p (2). First, to assess whether ATP modulates the interaction of Pex1p–Pex6p complex with Pex26p, we investigated any difference in the binding of Pex26p to Pex1p–Pex6p complex in the presence and absence of ATP. The cytosolic fraction from pex26 ZP167 cells expressing Pex1p-HA and Pex6p-HA (Figure 6A) and purified recombinant Flag-Pex1p and Flag-Pex6p (Figure 6B) were separately incubated with glutathione-S-transferase (GST)-fused Pex26p, termed GST-Pex26p, in the presence or absence of ATP. Cytosolic Pex1p-HA and Pex6p-HA as well as recombinant Flag-Pex1p and Flag-Pex6p were detected in the fractions each specifically bound to GST-Pex26p (Figure 6A, left panel, lanes 1 and 2; Figure 6B, left panel, lanes 1 and 2), apparently reflecting Pex6p and Pex1p plus Pex6p binding to Pex26p in the transport assays using a semi-intact cell system (Figures 2 and 3). Binding of both cytosolic and recombinant types of Pex1p and Pex6p to GST-Pex26p was increased about twofold to nearly fourfold in the presence of externally added ATP (Figure 6A,B, left panels, lane 3; right panels), as compared to those in the absence of any additional ATP (Figure 6A,B, left panels, lane 2; right panels). In the presence of AMP-PNP, the levels of cytosolic and recombinant Pex1p and Pex6p bound to GST-Pex26p were nearly the same as those in the absence of additional ATP (Figure 6A,B, left panels, lanes 2 and 4; right panels). In contrast, both types of Pex1p and Pex6p were barely discernible in fractions bound to GST-Pex26p using the assay mixtures that had been pretreated with apyrase to deplete endogenous ATP (Figure 6A,B, left panels, lane 5; right panels). These results suggested that the Pex1p–Pex6p complex interacted with Pex26p in an ATP-dependent manner. It is noteworthy that Pex1p binding to Pex26p was not inhibited by AMP-PNP, in contrast to the results from in vitro targeting assays using semi-intact cells, where the targeting of Pex1p, but not Pex6p, to peroxisomes was significantly lowered by AMP-PNP (Figure 4). This result implies that ATP hydrolysis is more likely required for stable localization of Pex1p to peroxisomes, although ATP hydrolysis is not essential for the Pex1p binding to Pex26p. It is also plausible that other factors on peroxisomes regulate the Pex1p–Pex6p–Pex26p complex.
Nucleotide binding protects Pex1p and Pex6p from protease digestion
AAA ATPases change their conformation in response to binding to nucleotides (32–36). Pex1p and Pex6p require ATP binding for assembly of their ternary complex with Pex26p (Figure 6). Therefore, we investigated by the limited proteolysis method whether Pex1p and Pex6p induce their conformational changes upon binding to nucleotides. Purified recombinant Pex1p and Pex6p were treated with various concentrations of trypsin in the presence of ATP or AMP-PNP or in the absence of any externally added nucleotides. In the presence of ATP, Pex1p and Pex6p were more resistant to partial digestion with trypsin than those in the presence of AMP-PNP or in the absence of any externally added nucleotides (Figure 7), hence strongly suggesting that Pex1p and Pex6p are altered to more compact and protease-resistant conformation upon ATP binding. Only slight enhancement of the resistance to trypsin with AMP-PNP is apparent as compared to that in the absence of ATP, possibly implicating a lesser conformational change than the alteration with ATP. Taken together, it is more likely that ATP binding to Pex1p and Pex6p induces their initial conformational change and the following ATP hydrolysis further alters the configurations.
In the present work, we identified the region of Pex26p involved in the peroxisomal targeting of Pex1p and Pex6p. The N-terminal cytoplasmic domain of the integral peroxisomal membrane protein Pex26p is required for peroxisomal targeting of Pex1p–Pex6p complexes and peroxisomal matrix protein import (2,20,22) (Figure 1). Similarly in the yeast Saccharomyces cerevisiae, Pex15p functions as the homolog of mammalian Pex26p (23). Pex6p interacts with the cytoplasmic part of Pex15p and the assembly of Pex1p–Pex6p–Px15p complexes is crucial for matrix protein import (23,24). Importantly, Pex26pΔ33-40 with a deletion of eight amino acids at 33-40 is compromised in the activity in recruiting Pex1p and Pex6p, thereby giving rise to the defect of matrix protein import into peroxisomes. These results raise several possibilities with respect to the relationship between the function and peroxisomal localization of these two AAA peroxins. One possibility is that peroxisomal localization of Pex1p and Pex6p is indispensable for the transport of matrix proteins including catalase and PTS1 and PTS2 proteins. However, several PBD patient-derived Pex26p mutants showing insufficient binding to Pex1p and Pex6p, hence indicative of cytosolic localization of the Pex1p–Pex6p complexes, were partially competent in the import of PTS1, but less efficient for the import of catalase and PTS2 proteins (19,20,22). Therefore, a part of Pex1p and Pex6p in the cytosol might be responsible for PTS1 protein import. Nevertheless, peroxisome-associated Pex1p and Pex6p are required for the efficient import of PTS1 proteins. It is also plausible that N-terminal domain of Pex26p functions with auxiliary factors besides Pex1p–Pex6p complexes in the transport of matrix proteins. PEX26 mutations identified in the PBD patients of CG8 are indeed located mostly at the genome sequence encoding the N-terminal domain of Pex26p (19,20,22), implying that this portion is crucial for the Pex26p function.
By making use of the semi-intact cell assay system successfully established in the present work, we showed that the peroxisomal targeting of Pex1p and Pex6p was enhanced by heat-labile factor(s) from the cytosolic fraction, implying that protein factors are involved in this process. To further characterize such factors, we performed immunodepletion assays with anti-HSP70 antibody. HSP70 depletion did not affect the peroxisomal targeting of Pex1p and Pex6p. We also showed that heat-inducible proteins, including HSP70 and HSP40, did not enhance the peroxisomal targeting of Pex1p and Pex6p (Figure 3). Any potential cytosolic protein(s) remain to be defined.
Limited digestion with trypsin has been used to verify the conformational changes of a number of AAA proteins such as NEM-sensitive factor (NSF) (34). On the basis of the patterns of limited tryptic digests of Pex1p or Pex6p under different nucleotide conditions (Figure 7), we conclude that conformational changes of Pex1p and Pex6p are induced upon binding to ATP and a little with AMP-PNP (Figure 7). In the absence of nucleotides, Pex1p and Pex6p show relatively loose and trypsin-sensitive conformation. Upon binding to nucleotides, Pex1p and Pex6p undergo a conformational change, becoming rather resistant to trypsin. It is also noteworthy that Pex1p fragments with lower molecular masses, apparently below ∼50 kDa, were detected only with the antiserum specific to the N-terminal, not C-terminal, part of Pex1p (21,25), thereby suggesting that the C-terminal part of Pex1p represents a loose and flexible conformation sensitive to trypsin. Intriguingly, ∼30-kDa fragment (Figure 7A, left panels, downward open arrowhead) is more resistant to tryptic cleavage in the presence of ATP than in the absence of any externally added nucleotides or in the presence of AMP-PNP. While the ∼30-kDa fragment of Pex1p appears to be further degraded to a 25-kDa fragment, both detectable with anti-N-terminal domain antibody, it is conceivable that an ATP hydrolysis-driven structural change gives rise to the upper fragment band protected from the tryptic digestion. The N-terminal domain of Pex1p is suggested to bind phosphoinositides and thereby associate with the peroxisomal membrane by recognizing specific lipids (39). Pex1p binding to the peroxisomal membrane may be regulated by an N-terminal conformational change dependent on the nucleotide-bound state.
With the semi-intact cell assay system, we show that the peroxisomal targeting of Pex1p requires ATP hydrolysis (Figure 4). In GST pull-down assays, Pex1p binds to Pex26p in a manner dependent on Pex6p and ATP binding, but not requiring ATP hydrolysis (Figure 6). We interpret this difference in regard to the requirement of ATP hydrolysis to imply that the interactions between Pex1p and phospholipids on the peroxisomal membrane surface are also involved in the regulation of Pex1p targeting. Another possibility is that Pex1p-interacting proteins including membrane-associated proteins and/or soluble peroxins such as Pex5p influence the Pex1p complex formation. On the other hand, Pex6p interacts with Pex26p in a manner dependent on ATP binding, but not requiring ATP hydrolysis, both in semi-intact cell assays and in GST pull-down assays (Figures 4 and 6). Therefore, it is more likely that Pex1p and Pex6p translocate to peroxisomes in a mutually distinct manner. Similar results were obtained in semi-intact cell assays using Walker-motif mutants of Pex1p and Pex6p, where the Walker A mutants influencing the ATP binding were more severely affected in targeting to Pex26p than the Walker B mutants (Figure 5). Pex6p mutated in the ATP-binding site (A1 or A2) shows a reduced level of targeting to Pex26p, while a point mutation in the ATP-hydrolysis site (B2) does not affect the peroxisomal targeting of Pex6p. These results suggest that the conformational changes in both D1 and D2 domains of Pex6p upon ATP binding, not its hydrolysis, are critical for the translocation of Pex6p to peroxisomes. With regard to Pex1p, the Walker mutants, A1, B1, and A2, show weaker interaction with Pex6p–Pex26p complexes on peroxisomes as compared to the wild type, while Pex1pB2 is transported to peroxisomes as efficiently as the wild-type Pex1p (Figure 5). Accordingly, it is more likely that ATP binding in both AAA cassettes and the following ATP hydrolysis in D1 of Pex1p are prerequisites for peroxisomal localization (Figure 8).
Crystal structure of the N-terminal domain of mouse Pex1p resembles valocin-containing protein (VCP), another member of the AAA proteins (39,40). VCP forms a barrel-like homo-hexameric structure (41). Recently, we showed that Pex1p is in a homo-trimer and a homo-hexamer (21). VCP as well as Pex1p possesses a highly protease-sensitive C-terminal domain, which is conversely protected from tryptic digestion upon nucleotide binding in the case of VCP (36). Thus, it is plausible that Pex1p forms a hexameric ring structure upon ATP binding and undergoes a structural change during the ATPase cycle in a manner similar to VCP (Figure 8). Likewise, ATP binding to Pex6p may trigger a conformational change and regulate its homo-oligomerization.
Pex1p and Pex6p are involved in early and late steps of peroxisome biogenesis including membrane fusion (42) and protein dislocation such as Pex5p export from peroxisomes (12,13,43). The yeast Pex1p and Pex6p likewise play pivotal roles in peroxisome biogenesis (13,17,23,24). These functions of Pex1p and Pex6p are reminiscent of those of NSF and VCP. AAA ATPases play various roles including vesicle fusion by NSF (44,45) and endoplasmic reticulum (ER)-associated protein degradation by p97/VCP/CDC48 (46). Despite their main localization in the cytosol, NSF and p97 exert their functions on membranes in vesicle fusion (47,48) and ER-associated protein degradation (49), respectively. AAA ATPase peroxins, Pex1p and Pex6p, also function on peroxisomal membranes to export Pex5p to the cytosol for a next round of protein import. Therefore, it is a prerequisite for AAA ATPases to target organelle membranes where they function. In the present work, we have dissected the mechanisms underlying the targeting of AAA ATPase peroxins to peroxisomes. As shown in Figure 8, our findings strongly suggest that ATP-dependent binding of Pex1p and Pex6p to and possibly their release from Pex26p via concomitant conformational changes are responsible for the regulation of subcellular localization and functions of AAA ATPase peroxins. Furthermore, our newly developed transport system using semi-intact cells readily analyzes the dynamics of Pex1p and Pex6p targeting to and release from their recruiter Pex26p under the physiological conditions. This system would open a way to elucidate the mechanisms underlying peroxisome biogenesis involving matrix protein import.
Materials and Methods
Cell culture and DNA transfection
pex26 CHO cell mutant, ZP167 (2), was cultured at 37°C or 30°C in Ham's F-12 medium (Invitrogen) and supplemented with 10% fetal calf serum under 5% CO2, 95% air. DNA transfection into CHO cells was performed with lipofection method using Lipofectamine (Invitrogen) as described (2,50). Cells transfected with plasmids encoding human Pex1p and/or Pex6p were cultured for 2 days at 30°C to elevate a lower expression level of Pex1p and Pex6p at 37°C (31).
To construct PEX26-HA variants encoding N-terminally truncated human Pex26p-HA mutants, polymerase chain reaction (PCR) was performed as follows. cDNAs coding for Pex26p mutants truncated in amino-acid residues at positions 1-32 termed Δ32 (the same as Pex26p33-305) and 1-40 named Δ40 (the same as Pex26p41-305) were constructed by PCR using pCMVSPORT/PEX26-HA(2) as a template and each set of primers: respective forward primers, SalI-Met-Hs26-33 (5′-ACGCGTCGACATGCCGGCCGTGGACCTTCTGGAGGAG-3′) and SalI-Met-Hs26-41 (5′-ACGCGTCGACATGGCGGCCGACCTCCTGGTGCAC-3′), and a reverse primer T7 (5′-TAATACGACTCACTATAGGG-3′). The SalI-BglII fragment of the PCR product was inserted into the SalI-BglII site of pCMVSPORT/PEX26-HA. Δ33-40 was generated by two-step PCR. We carried out the first PCR using pCMVSPORT/PEX26-HA as a template and a set of forward and reverse primers, SP6 (5′-ATTTAGGTGACACTATAG-3′) and Hs26del33-40Rv (5′-GAGGTCGGCGCCGCCCGGGCCGGGACCGC-3′), and Hs26del33-40Fw (5′-GCCCGGGCGGCGGCCGACCTCCTGCTG-3′) and T7. The resulting fragments were used as a template for the second PCR. Second PCR was performed using a set of forward and reverse primers SP6 and T7. The SalI-BglII fragment of the second PCR product was inserted into the SalI-BglII site of pCMVSPORT/PEX26-HA.
For PEX6-Myc, the XhoI-BamHI fragment of the PCR product generated from pCMVSPORT/PEX6-HA(2) and primers PEX6.F5 (5′-ATGCATGCCGTAGTCAGG-3′) and Hs6-BamHI-3′ (5′-CGCGCATCCGCAG GCAGCAAACTTGCGCTG-3′) was inserted into the XhoI-SphI site of pCMVSPORT/PEX6-HA, together with the BamHI-SphI fragment of pCMVSPORT/6Myc (18). PMP22-EGFP expression plasmid, pUcD2HygPM P22EGFP, was as described (51).
Rabbit antibodies to HA (52), catalase (5), PTS1 peptide (53), thiolase (54), Pex1p (21,25), Pex6p (55) and guinea pig anti-Pex14p antibody (56) were as described. Rabbit antibodies to GST (Sigma) and HSP40 (Stressgen) and mouse antibodies to c-Myc (9E10; Santa Cruz), FlagM2 (Sigma), HA (16B2; Covance), HSP70 (Stressgen) and green fluorescent protein (GFP) (Santa Cruz) were purchased.
Cells were fixed with 4% paraformaldehyde at room temperature for 10 min, permeabilized with methanol on ice for 10 min and were then incubated with PBS containing 1% BSA at room temperature for 30 min. Indirect immunostaining using secondary antibodies labeled with Alexa 488, 568 and 633 (Invitrogen) was as described (50). EGFP fluorescence was directly monitored under a confocal LSM510 microscope (Carl Zeiss).
Preparation of semi-intact cells
Semipermeabilized CHO cells were prepared as described (18). Briefly, cells were plated at a density of 2.5 × 105 cells/mL on an 18-mm glass coverslip (Matsunami). Cells were incubated for 10 min on ice with 5% BSA containing ice-cold semi-intact buffer (SB; 250 mm sucrose, 25 mm HEPES–KOH, pH 7.4, 2.5 mm EGTA, pH 7.4, 2.5 mm magnesium acetate, 25 mm KCl, 1 mm dithiothreitol, 0.5 µm taxol and 0.1% BSA). After incubation, BSA was washed out with ice-cold SB, and then cells were incubated with 50 µg/mL digitonin in SB for 5 min at room temperature to form pores in the plasma membranes. The permeabilized cells were washed 3× with ice-cold SB and then incubated in SB at 4°C for 2 h to remove the residual cytosol from the cells. About 80% of the total cells were permeabilized under this condition as described previously (18).
Preparation of cytosolic fraction from CHO cells
CHO cells expressing epitope-tagged human Pex1p and/or Pex6p were harvested and homogenized in SB. Post-nuclear supernatant fraction was obtained by centrifugation of the cell homogenate at 800 ×g for 10 min. Cytosolic fraction was prepared by centrifugation of the post-nuclear supernatant fraction at 100 000 ×g for 30 min.
Immunodepletion of HSP70 from cytosolic fraction
Cytosolic fractions were incubated with anti-HSP70 antibody on ice for 6 h. The antigen–antibody complexes were removed with Protein G-Sepharose beads (GE Healthcare). HSP70-depleted cytosolic fractions were used for in vitro targeting assay.
In vitro targeting assay
Semipermeabilized CHO cells expressing EGFP-Pex26p or PMP22-EGFP were incubated at 26°C for 1 h unless otherwise described, with cytosolic fraction from another set of CHO cells expressing epitope-tagged Pex1p and/or Pex6p and recombinant Flag-tagged Pex1p and Pex6p (see below). After incubation, cells were washed 3× with ice-cold SB and further washed with ice-cold SB at 10-min intervals for 30 min to remove untargeted proteins. Cells were fixed and subjected to morphological and western blot analyses. To inhibit ATPase, cytosolic fraction was incubated with 5 U/mL of apyrase (Sigma) at 26°C for 45 min or supplemented with 10 mm AMP-PNP (Sigma). In several in vitro targeting assays, ARS containing 10 mm creatine phosphate (Roche Diagnostics) and 50 µg/mL creatine kinase (Roche Diagnostics) were added in addition to 2 mm ATP (Sigma). To assess the targeting of recombinant Pex1p and Pex6p, these recombinant proteins were diluted with SB and incubated with semipermeabilized CHO cells expressing EGFP-Pex26p or PMP22-EGFP.
Preparation of recombinant proteins
For isolation of Flag-Pex1p and Flag-Pex6p, Sf9 cells were infected with baculoviruses each harboring FLAG-PEX1 and FLAG-PEX6 (S. Tamura and Y. Fujiki, unpublished data). At 72 h post-infection, cells were lysed by sonication in L buffer consisting of 50 mm Tris–HCl, pH 7.4, 150 mm KCl, 10% glycerol, 1 mm phenylmethylsulfonyl fluoride, 25 µg/mL each of antipain and leupeptin and 50 units of aprotinin (Peptide Inst.). Flag-Pex1p and Flag-Pex6p were purified from the soluble fraction with anti-FLAG immunoglobulin G M2-agarose (Sigma). Flag-Pex1p and Flag-Pex6p were isolated by elution with FLAG peptide (Sigma) in SB.
Partial proteolysis experiments using 0.5–1 µg/mL of recombinant Flag-Pex1p or Flag-Pex6p with 0–1.0 µg/mL of trypsin (Sigma) were carried out for 30 min at 37°C in the presence or absence of 10 mm nucleotides in reaction buffer: 25 mm HEPES–KOH, pH 7.4, 2.5 mm EGTA, pH 7.4, 2.5 mm magnesium acetate, 25 mm KCl and 1 mm dithiothreitol. The reaction was terminated by the addition of SDS–PAGE sample buffer and boiling. The digests were analyzed by SDS–PAGE and immunoblotting with rabbit antibodies specific to N- and C-terminal parts of Pex1p and Pex6p.
Expression of GST-Pex26p fusion protein in Escherichia coli and GST pull-down assay were performed as described (15,52), using purified recombinant Flag-Pex1p and Flag-Pex6p or cytosolic fraction containing Pex1p and Pex6p under various ATP conditions. Protein bands in SDS–PAGE were detected by immunoblot analysis (56) and quantified using a Fuji-film MultiGauge software (Fuji Film). Statistical significance was examined by Aspin–Welch's t-test (57,58) and represented as *p < 0.05 and **p < 0.01.
We thank M. Nishi for preparing figures and the other members of our laboratory for discussion. This work was supported in part by SORST and CREST grants (to Y. F.) from the Science and Technology Agency of Japan; Grants-in-Aid for Scientific Research (to Y. F.); Grant of National Project on Protein Structural and Functional Analyses (to Y. F.); The 21st Century COE and Global COE Programs from The Ministry of Education, Culture, Sports, Science, and Technology of Japan and a grant (to Y. F.) from Japan Foundation for Applied Enzymology.