Peroxisomal Targeting Signal Receptor Pex5p Interacts with Cargoes and Import Machinery Components in a Spatiotemporally Differentiated Manner: Conserved Pex5p WXXXF/Y Motifs Are Critical for Matrix Protein Import

Abstract

Two isoforms of the peroxisomal targeting signal type 1 (PTS1) receptor, termed Pex5pS and (37-amino-acid-longer) Pex5pL, are expressed in mammals. Pex5pL transports PTS1 proteins and Pex7p-PTS2 cargo complexes to the initial Pex5p-docking site, Pex14p, on peroxisome membranes, while Pex5pS translocates only PTS1 cargoes. Here we report functional Pex5p domains responsible for interaction with peroxins Pex7p, Pex13p, and Pex14p. An N-terminal half, such as Pex5pL(1-243), comprising amino acid residues 1 to 243, bound to Pex7p, Pex13p, and Pex14p and was sufficient for restoring the impaired PTS2 import of pex5 cell mutants, while the C-terminal tetratricopeptide repeat motifs were required for PTS1 binding. N-terminal Pex5p possessed multiple Pex14p-binding sites. Alanine-scanning analysis of the highly conserved seven (six in Pex5pS) pentapeptide WXXXF/Y motifs residing at the N-terminal region indicated that these motifs were essential for the interaction of Pex5p with Pex14p and Pex13p. Moreover, mutation of several WXXXF/Y motifs did not affect the PTS import-restoring activity of Pex5p, implying that the binding of Pex14p to all of the WXXXF/Y sites was not a prerequisite for the translocation of Pex5p-cargo complexes. Pex5p bound to Pex13p at the N-terminal part, not to the C-terminal SH3 region, via WXXXF/Y motifs 2 to 4. PTS1 and PTS2 import required the interaction of Pex5p with Pex14p but not with Pex13p, while Pex5p binding to Pex13p was essential for import of catalase with PTS1-like signal KANL. Pex5p recruited PTS1 proteins to Pex14p but not to Pex13p. Pex14p and Pex13p formed a complex with PTS1-loaded Pex5p but dissociated in the presence of cargo-unloaded Pex5p, implying that PTS cargoes are released from Pex5p at a step downstream of Pex14p and upstream of Pex13p. Thus, Pex14p and Pex13p very likely form mutually and temporally distinct subcomplexes involved in peroxisomal matrix protein import.

Most organellar proteins are synthesized on cytoplasmic polyribosomes and are then directed to their destined compartments. Peroxisomal matrix and membrane proteins are also synthesized on free polyribosomes and posttranslationally imported into peroxisomes (25). Two distinct topogenic signals, peroxisomal targeting signal type 1 (PTS1) and PTS2 (16, 40), direct proteins to the peroxisomal matrix. Genetic analyses of peroxisome-deficient mutants isolated from yeast and mammalian cells led to the identification of a number of proteins, known as peroxins, which are essential for peroxisome biogenesis (11, 17, 23, 41). PEX5 and PEX7 encode the receptors for PTS1 and PTS2, respectively. Of note, catalase with the C-terminal sequence KANL is also imported into peroxisomes by the Pex5p-dependent pathway (29, 32). However, the underlying molecular mechanism for catalase import remains unclear. Dysfunction of Pex5p and Pex7p causes human peroxisome biogenesis disorders (PBD), such as Zellweger syndrome of complementation group 2 (CG2) and rhizomelic chondrodysplasia punctata of CG11, respectively (17, 23, 41). Pex5p and Pex7p have been proposed to be mobile receptors that bind cargo proteins in the cytoplasm, traverse and dock at the peroxisomal membrane, release their cargoes, and then recycle to the cytoplasm (17, 41). Whether cargo proteins are released at the inner surface and/or inside of peroxisomes is not well understood. Very recently, Pex5p, in part if not as a whole, was shown to be a mobile signal receptor shuttling between the cytosol and the peroxisomal matrix (10).

A fundamental issue as to how matrix proteins are imported into peroxisomes has been extensively investigated by making use of peroxisome-deficient mutant yeast and mammalian cells, including CHO cells and PBD patient-derived fibroblasts. Lessons from yeast and mammalian systems led to the conclusion that the import mechanisms for these two evolutionarily distinct systems are basically similar. However, several aspects are distinct. (i) Mammalian cells synthesize two isoforms of Pex5p, Pex5pS and Pex5pL, which has an internal 37-amino-acid insertion (6, 31) and which plays a pivotal role in PTS2 import by interacting with Pex7p, in addition to PTS1 cargo transport (26, 29). (ii) In yeasts such as Saccharomyces cerevisiae, Pex14p and Pex13p interact with Pex5p, implying that both function as docking receptors for Pex5p-PTS1 cargo complexes. (iii) In contrast, mammalian Pex14p is suggested to be the initial docking site of Pex5p-cargo complexes Pex5p-PTS1 and Pex5pL-Pex7p-PTS2 in matrix protein import. The biochemical functions of other potential members of this import machinery such as Pex13p, Pex12p, Pex10p, and Pex2p are not well defined yet. Hence, the molecular mechanisms of a number of peroxins involved in peroxisomal protein import need to be determined for us to understand the protein translocation processes that are essential to peroxisome biogenesis.

Pex5p is composed of two distinct parts, a highly conserved C-terminal half comprising seven tetratricopeptide repeat (TPR) motifs and an N-terminal half in which only a few amino acids are strictly conserved, typically in the multiple pentapeptide WXXXF/Y repeats (5, 36). While the TPR region was shown to mediate the binding to PTS1-containing proteins (18), specific functions were not fully assigned to the N-terminal half. Mammalian Pex5pL contains seven WXXXF/Y motifs, while two and three motifs are identified in Pex5p from S. cerevisiae and Pichia pastoris, respectively. The N-terminal part of P. pastoris Pex5p has recently been shown to be required for interaction with the Src homology 3 (SH3) domain of Pex13p (45). In S. cerevisiae, the pentapeptide motifs have been demonstrated to be involved in the interaction of Pex5p with Pex13p (5). However, mutation of the motifs did not affect the binding of Pex5p with Pex14p, implying that the WXXXF/Y motifs are not involved in such an interaction (5). Whether or not the WXXXF/Y motifs are involved in the interaction of mammalian Pex5p with Pex13p has not been reported, but seven pentapeptide motifs in human Pex5pL have been shown to bind Pex14p in vitro with different affinities (35). The physiological significance of such multiple Pex14p-binding sites is not understood.

As a further step toward a better understanding of the underlying molecular mechanisms of Pex5p in matrix protein import into peroxisomes in mammals, we have searched for functional regions of Pex5pL that are responsible for interaction with Pex7p, Pex13p, and Pex14p. These members of the peroxin family may all be involved in early stages of peroxisomal matrix protein import. We have employed a combination of an in vitro binding assay and a functional complementation assay of a large number of Pex5p variants using pex5 cell mutants such as CHO ZP105 defective in PTS1 and PTS2 import (29, 31). We show here that the WXXXF/Y motifs of Pex5p are essential for the interaction with Pex13p and Pex14p. We have found that Pex7p bound to Pex5pL at a short sequence of amino acid residues (residues 190 to 233), including the N-terminal 18 amino acids of the Pex5pL-specific 37-amino-acid insertion. Several aspects such as the Pex5p-binding region of Pex13p are also distinct from the findings for yeast. Furthermore, we discuss the interaction between Pex5p and both Pex14p and Pex13p as well as their molecular dynamics.

MATERIALS AND METHODS

Biochemicals.

Restriction enzymes and DNA-modifying enzymes were purchased from Nippon Gene (Tokyo, Japan) and Takara (Kyoto, Japan). Fetal calf serum and Ham's F-12 medium were from Life Technologies Inc. We used rabbit antibodies to rat catalase (44), PTS1 peptide (31), acyl-coenzyme A (CoA) oxidase (AOx) (44), 3-ketoacyl-CoA thiolase (thiolase) (44), sterol carrier protein X (SCPx) (30), human Pex7p (S. Mukai and Y. Fujiki, unpublished data), Pex13p (43), and Pex14p (38). A rabbit antibody to green fluorescent protein (GFP) (Clontech, Tokyo, Japan) was purchased.

Cell culture, PEX cDNA transfection, and morphological analysis.

CHO cells, including pex5 mutants ZP105 (31) and ZPG231 (26), were cultured in Ham's F-12 medium supplemented with 10% fetal calf serum under 5% CO₂-95% air. DNA transfection to cells was done with Lipofectamine (Life Technologies, Gaithersburg, Md.) as recommended by the manufacturer. After 3 days of culture, peroxisomes in CHO cells were visualized by indirect immunofluorescence light microscopy with the monospecific rabbit antibodies described above. Antigen-antibody complexes were detected with a fluorescein isothiocyanate-labeled goat anti-rabbit immunoglobulin G antibody (Cappel, Durham, N.C.) under an Akioskop FL microscope (Carl Zeiss, Oberkochen, Germany) (29). Cells were fixed with a fixative containing 4% paraformaldehyde. Permeabilization of cells was done by treatment with 1% Triton X-100 (29).

Construction of Pex5p variants.

Expression plasmids for cDNAs encoding partially truncated Pex5pL mutants were constructed, basically by two steps. To construct cDNAs coding for Pex5pL mutants comprising amino acid residues 1 to 243 [Pex5pL(1-243)], 1 to 184 [Pex5pL(1-184)], 1 to 163 [Pex5pL(1-163)], 1 to 158 [Pex5pL(1-158)], 1 to 144 [Pex5pL(1-144)], and 1 to 139 [Pex5pL(1-139)], PCR was done using as a template ClPEX5L cDNA (29), with a primer set consisting of forward primer Sse8387I.f and reverse primer 243.r, 184.r, 163.r, 158.r, 144.r, or 139.r, respectively (Table 1). The Sse8387I-NotI regions of ClPEX5L-HA in mammalian expression vector pUcD2SRαMCSHyg (29) and ClPEX5L in Escherichia coli expression vector pGEX6P-1 (Amersham Pharmacia Biotech, Tokyo, Japan) (29) were replaced by the respective Sse8387I-NotI fragments of the PCR products. For cDNAs encoding Pex5pL(145-243), Pex5pL(185-243), and Pex5pL(190-243), EcoRI-NotI fragments of PCR products amplified with forward primer 145.f, 185.f, or 190.f, respectively, and reverse primer 243.r were separately subcloned into the EcoRI-NotI sites of pUcD2SRαMCSHyg and pGEX6P-1. Expression plasmids for Pex5pL(190-233) and Pex5pL(190-223) were likewise constructed by PCR using forward primer 190.f and reverse primers 233.r and 223.r. EcoRI-SalI fragments of the PCR products were separately subcloned into the EcoRI-SalI sites in pUcD2SRαMCSHyg and pGEX6P-1 vectors. To construct cDNA coding for Pex5pL(190-233) with missense mutation S214F identified in pex5 CHO mutant ZPG231 (26), named Pex5pL(190-233)S214F, PCR was done with forward primer 190.f and reverse primer 233.r, by using as a template ClPEX5L cDNA derived from ZPG231 (26). EcoRI-SalI fragments of the PCR products were subcloned into the EcoRI-SalI site in pGEX6P-1. For cDNAs coding for Pex5pL(118-243) and Pex5pL(140-243), BamHI-NotI fragments of PCR products amplified with forward primers 118.f and 140.f, respectively, and reverse primer 243.r were separately inserted into the BamHI-NotI sites of pUcD2SRαMCSHyg and pGEX6P-1. The expression plasmid for Pex5pL(140-184) was similarly made by PCR using forward primer 140.f and reverse primer 184.r; BamHI-NotI fragments were subcloned into the BamHI-NotI sites in pUcD2SRαMCSHyg and pGEX6P-1. To construct cDNAs encoding Pex5pL(145-189) and Pex5pL(206-262), EcoRI-SalI fragments of PCR products obtained by using primer pairs 145.f and 189.r and 206.f and 262.r, respectively, were separately inserted into the EcoRI-SalI sites in pUcD2SRαMCSHyg and pGEX6P-1. A cDNA construct encoding Pex5pL(145-184) was made by introducing the EcoRI-NotI fragment of the PCR-amplified product obtained by using primers 145.f and 184.r into the EcoRI-NotI sites in pUcD2SRαMCSHyg and pGEX6P-1. An expression plasmid for Pex5pL(306-632) was likewise constructed by using forward primer 306.f and reverse primer 632.r. The BamHI-NotI fragment of the PCR product was subcloned into the BamHI-NotI sites in pUcD2SRαMCSHyg and pGEX6P-1.

TABLE 1.

Synthetic oligonucleotide primers used in this study

Primera	Sequence (5′-3′)	Significance of underlined sequence
Sse8387I.f	CCTGCAGGACCAGAATGC
139.r	GCGGCCGCTCAGTCAGTCTCATTATAATC	NotI site, termination codon
144.r	GCGGCCGCTCAGAATTCTTGGGACCAGTC	NotI site, termination codon
158.r	GCGGCCGCTCATCGGGCAGGGGACACAGA	NotI site, termination codon
163.r	GCGGCCGCTCAATACTCCTCAGCCCATCG	NotI site, termination codon
184.r	GCGGCCGCTCATCGATCAGTGGTGGATGA	NotI site, termination codon
189.r	AACGGTCAATGCGATGTCGACTCAATAGTCATCATACCATCGATC	SalI site, termination codon
223.r	AACGGTCAATGCGATGTCGACTCAAATCTGTCGCACGAATTTCAG	SalI site, termination codon
233.r	AACGGTCAATGCGATGTCGACTCAAGCAGACTCCAGGGACACCTG	SalI site, termination codon
243.r	GCGGCCGCTCACTGTTCTGCCTGAGCTCG	NotI site, termination codon
248.r	AACGGTCAATGCGATGTCGACTCAAAACTCTGCTGCCCACTGTTC	SalI site, termination codon
257.r	AACGGTCAATGCGATGTCGACTCAGGCCTCTGATGTGCCCTGCTG	SalI site, termination codon
262.r	AACGGTCAATGCGATGTCGACTCAGAACTGATCGACCCAGGC	SalI site, termination codon
300.r	AACGGTCAATGCGATGTCGACTCAGGGGTGCGCCTCAGCATCCCG	SalI site, termination codon
305.r	AACGGTCAATGCGATGTCGACTCAATAGTCAGAAAGCCAGGGGTG	SalI site, termination codon
328.r	AACGGTCAATGCGATGTCGACTCAGTGGTCACGCAGGGGGTTCTC	SalI site, termination codon
632.r	AGTCGACTCACTGGGGCAGGCCAAACATAGC	SalI site, termination codon
118.f	TCGAATCGTAGGGATCCATGTGGGCCCAGGAGTTTCTTGCG	BamHI site, initiation codon
123.f	TCGAATCGTAGGGATCCATGCTTGCGGCTGGAGATGCTGTG	EcoRI site, initiation codon
140.f	TCGAATCGTAGGGATCCATGTGGTCCCAAGAATTCATCGTC	BamHI site, initiation codon
145.f	TCGAATCGACTGAATTCATGATCGCTGAAGTCACAGACCCA	EcoRI site, initiation codon
159.f	TCGAATCGACTGAATTCATGTGGGCTGAGGAGTATCTGGAG	EcoRI site, initiation codon
164.f	TCGAATCGACTGAATTCATGCTGGAGCAGTCTGAGGAGAAG	EcoRI site, initiation codon
185.f	TCGAATCGACTGAATTCATGTGGTATGATGACTATCATCCC	EcoRI site, initiation codon
190.f	TCGAATCGACTGAATTCATGCATCCCGAGGAGGATCTGCAG	EcoRI site, initiation codon
206.f	TCGAATCGACTGAATTCATGAAGGTGGACGACCCCAAATTG	EcoRI site, initiation codon
244.f	TCGAATCGACTGAATTCATGTGGGCAGCAGAGTTTATACAG	EcoRI site, initiation codon
249.f	TCGAATCGACTGAATTCATGATACAGCAGCAGGGCACATCA	EcoRI site, initiation codon
258.f	TCGAATCGACTGAATTCATGTGGGTCGATCAGTTCACAAGG	EcoRI site, initiation codon
263.f	TCGAATCGACTGAATTCATGACAAGGTCAGGAAACACGTCT	EcoRI site, initiation codon
306.f	AGAATTCATGGATGACCTCACATCTGCTTCC	EcoRI site, initiation codon
Sse8387I.f-2	TAGGTTCTGAAGATGAGT
W118A/F122A.f	TCTGAGAATGCGGCCCAGGAGGCTCTTGCGGCTGGA	Trp118 and Phe122 mutated to Ala
W118A/F122A.r	TCCAGCCGCAAGAGCCTCCTGGGCCGCATTCTCAGA	Trp118 and Phe122 mutated to Ala
W140A/F144A.f	GAGACTGACGCGTCCCAAGAAGCCATCGCTGAAGTC	Trp140 and Phe144 mutated to Ala
W140A/F144A.r	GACTTCAGCGATGGCTTCTTGGGACGCGTCAGTCTC	Trp140 and Phe144 mutated to Ala
W159A/Y163A.f	CCTGCCCGAGCGGCTGAGGAGGCTCTGGAGCAGTCT	Trp159 and Tyr163 mutated to Ala
W159A/Y163A.r	AGACTGCTCCAGAGCCTCCTCAGCCGCTCGGGCAGG	Trp159 and Tyr163 mutated to Ala
W185A/Y189A.f	ACTGATCGAGCGTCTGATGACGCTCATCCCGAGGAG	Trp185 and Tyr189 mutated to Ala
W185A/Y189A.r	CTCCTCGGGATGAGCGTCATCATACGCTCGATCAGT	Trp185 and Tyr189 mutated to Ala
Mut-RV-HA	AGTTCAGATAACGCGGCCGCTCACAATGATGCGTAATCTGGTACGTCGTATGGATAGCACTGTTCTGCCTGAGCTCG	NotI site, termination codon, HA epitope
KpnI.r	GAGTAGCGCAGCCAGTC
PEX13.f	GCGGATCCTCTAGAGCCATGGCGTCCCAGCCGCCACC	BamHI site
PEX13SH3.f	GCGAATTCACTCACAGTGATGAAG	EcoRI site
PEX13N.r	ATGCGGCCGCTCACTTTCAATGGACTGAAATGC	NotI site, termination codon
M13.r	GTACCAGTATCGACAAAGG
S593W.f	GGCGCCATGTGGGAGAACATCTGGAGT	Ser593 mutated to Trp
S593W.r	ACTCCAGATGTTCTCCCACATGGCGCC	Ser593 mutated to Trp
1183.f	AACGAGTCCCTGCAGCG
PEX5stopHA.r	AGTTCAGATAACGCGGCCGCTCACAATGATGCGTAATCTGGTACGTCGTATGGATAGCACTGGGGCAGGCCAAACATAGC	NotI site, termination codon, HA epitope
HsCatalase.f	GCGCGTCGACGCTGACAGCCGGGATC	Sal I site
HsCatalase.r	GCGCGTCGACTAGTCACAGATTTGCCTTC	SalI site, termination codon
1148.f	GACACTCACCGCCATCGCCTG
ΔKANL.r	GCGCCTCGAGTCACTCCCTTGCCGCCAAGTGAGA	XhoI site, termination codon
GFP-SKL.f	CCGACCATGGTGAGCAAGGGC	NcoI site
GFP-SKL.r	GGCGTC GACTAGAGCTTGGACTGCAGGTGATGGTGATGGTGATGGAGTCCGGACTTGTA	SalI site, LQSKLTer, His6

^af and r, sense and antisense primers, respectively.

Expression plasmids for glutathione S-transferase (GST) fusion proteins containing single or double WXXXF/Y motifs in short stretches of Pex5pL were constructed as follows. To construct cDNAs coding for Pex5pL(123-158) and Pex5pL(164-243), PCR was done with primer pairs 123.f and 158.r and 164.f and 243.r, respectively, by using ClPEX5L as a template. EcoRI-NotI fragments of the PCR products were separately inserted into the EcoRI-NotI site in pGEX6P-1. To construct cDNAs encoding Pex5pL(190-257), Pex5pL(249-300), Pex5pL(263-328), Pex5pL(159-189), Pex5pL(185-248), Pex5pL(244-262), and Pex5pL(258-305), EcoRI-SalI fragments of PCR products, amplified by using primer sets 190.f and 257.r, 249.f and 300.r, 263.f and 328.r, 159.f and 189.r, 185.f and 248.r, 244.f and 262.r, and 258.f and 305.r, respectively, were separately inserted into the EcoRI-SalI site in pGEX6P-1. To construct cDNAs coding for Pex5pL(118-144) and Pex5pL(140-163), BamHI-SalI fragments of PCR-amplified products obtained by using primer pairs 118.f and 144.r and 140.f and 163.r, respectively, were separately inserted into the BamHI-SalI site in pGEX6P-1.

cDNA encoding ClPex5pL with mutation S593W, named Pex5pL-S593W, was constructed as follows. PCR was initially performed with template ClPEX5L and a set consisting of forward primer 1183.f and reverse primer S593W.r and one consisting of forward primer S593W.f and reverse primer PEX5StopHA.r. A second PCR was done using the initial PCR products as the template and primers 1183.f and PEX5StopHA.r. A BamHI-NotI fragment of the PCR products was substituted for the BamHI-NotI region of pBlueScript SK(−)·ClPEX5L. The Sse8387I-NotI fragment of the resultant pBlueScript SK(−) · ClPEX5L(S593W) was placed into the Sse8387I-NotI sites of pUcD2SRαMCSHyg · ClPEX5L-HA and pGEX6P-1 · ClPEX5L.

Construction of an expression plasmid encoding the N-terminal 40-amino-acid sequence of rat Pex3p fused with enhanced GFP (EGFP), termed PEX3(1-40)-EGFP, in pUcD2SRαMCSHyg was as described previously (20). All constructs were confirmed by nucleotide sequencing and used for assays.

Alanine substitution in the WXXXF/Y motifs.

Substitution of Ala for the conserved Trp and Phe/Tyr of the WXXXF/Y motifs was performed by degenerate oligonucleotide mutagenesis. PCR primers used were forward primer W118A/F122A.f and reverse primer W118A/F122A.r, introducing point mutations W118A and F122A in the first WXXXF/Y motif; W140A/F144A.f and W140A/F144A.r, incorporating W140A and F144A into the second WXXXF/Y motif; W159A/Y163A.f and W159A/Y163A.r, introducing W159A and Y163A into the third WXXXF/Y motif; and W185A/Y189A.f and W185A/Y189A.r, incorporating W185A and Y189A into the fourth WXXXF/Y motif. The point mutations were generated by two steps of PCR. The first PCR was done with ClPEX5L cDNA as a template and with a primer set consisting of reverse primer Mut-RV-HA and forward primer W118A/F122A.f, W140A/F144A.f, W159A/Y163A.f, or W185A/Y189A.f and another set consisting of forward primer Sse8387I.f-2 and reverse primer W118A/F122A.r, W140A/F144A.r, W159A/Y163A.r, or W185A/Y189A.r. The second PCR for the point mutations was done with primers Sse8387I.f-2 and Mut-RV-HA by using as a template each pair of the first PCR products containing the same-site point mutations. After digestion with Sse8387I and NotI, the amplified fragments were separately subcloned into mammalian expression plasmid pUcD2SRαMCSHyg and E. coli expression plasmid pGEX6P-1 by replacing the Sse8387I-NotI fragment of PEX5LHA. Resulting mutated forms of Pex5pL(1-243)HA were named Mut1 for the mutant with A¹¹⁸XXXA¹²², Mut2 for one with A¹⁴⁰XXXA¹⁴⁴, Mut3 for one with A¹⁵⁹XXXA¹⁶³, and Mut4 for one with A¹⁸⁵XXXA¹⁸⁹.

To introduce a combination of multiple pentapeptide mutations (W118A and F122A, W140A and F144A, W159A and Y163A, and W185A and Y189A) into Pex5pL(1-243)HA, Mut1 plasmid pGEX6P-1, encoding Pex5pL(1-243)HA with a single mutated WXXXF/Y motif, was used as a template. Mut12, Mut 13, and Mut14 were created by PCR with Mut1 as a template by using primer pairs W140A/F144A.f (forward) and W140A/F144A.r (reverse), W159A/Y163A.f and W159A/Y163A.r, and W185A/Y189A.f and W185A/Y189A.r, respectively. Other Mut variants were likewise constructed: Mut2 was used as a template for constructing Mut23 and Mut24, and Mut3 was used for Mut34. To construct triple-motif mutants Mut123, Mut124, and Mut134, Mut12 was used as a template for the first two and Mut13 was used as a template for the third. Mut234 was made by using Mut23 as a template. Four-motif-mutated form Mut1234 was constructed with Mut123 as a template. Hemagglutinin (HA) tagging of the C terminus of Pex5pL(1-243) was done with a PCR-based technique using forward primer Sse8387I.f-2 and reverse primer Mut-RV-HA and template ClPEX5L cDNA. HA-tagged wild-type Pex5pL(1-243) was cloned into pUcD2SRαMCSHyg and pGEX6P-1 by replacing the Sse8387I-NotI fragment of PEX5LHA with the respective Sse8387I-NotI fragments of the PCR-amplified products. Point mutation S214F was introduced by PCR by using as a template ClPEX5L cDNA derived from pex5 CHO mutant ZPG231, with forward primer Sse8387I.f-2 and reverse primer Mut-RV-HA. After digestion with Sse8387I and NotI, the amplified fragment was subcloned into pUcD2SRαMCSHyg and pGEX6P-1, respectively, by replacing the Sse8387I-NotI fragment of PEX5L(1-243)HA. The point mutations in the WXXXF/Y motifs of full-length Pex5pL-HA were done by PCR using ClPEX5L cDNA as a template, with forward primer Sse8387I.f-2, reverse primer KpnI.r, and the degenerate oligonucleotide primers described above. The multiple mutations of full-length Pex5pL-HA(W140A/F144A, W159A/Y163A, W185A/Y189A) were created using a single-motif PEX5LHA mutant as a template. After two steps of PCR, the resulting PCR fragment was introduced into pUcD2SRαMCSHyg and pGEX6P-1 by replacing the Sse8387I-KpnI fragment of PEX5LHA. All plasmid constructs were assessed by nucleotide sequence analysis.

cDNAs encoding several oligopeptides, each containing a single pentapeptide motif with mutation of W and F/Y to A as well as its flanking sequence and containing the N-terminal part encompassing WXXXF/Y motifs 1 to 4, including (1-139)Mut1, (123-158)Mut2, (145-184)Mut3, and (164-243)Mut4, were constructed by PCR by using Mut1234 as a template, with pairs of primers Sse8387I.f and 139.r, 123.f and 158.r, 145.f and 184.r, and 164.f and 243.r, respectively. An Sse8387I-NotI fragment of the PCR products for (1-139)Mut1 was introduced into the Sse8387I-NotI site of ClPEX5L in pGEX6P-1. EcoRI-NotI fragments of the PCR products for (123-158)Mut2, (145-184)Mut3, and (164-243)Mut4 were likewise inserted into the EcoRI-NotI site of pGEX6P-1.

Construction of GST-PEX13, GST-PEX13N, and GST-PEX13C.

Expression plasmids encoding human Pex13p fused to GST (GST-Pex13p) were constructed as follows. PCR was done using pBlueScript SK(−) · HsPEX13 (43) as a template with forward primer PEX13.f and reverse primer M13.r. A BamHI-NotI fragment of PCR products was subcloned into the BamHI-NotI site in the pGEX6P-1 vector. The expression plasmid for the GST fusion with the N-terminal part comprising amino acid residues 1 to 135 of HsPex13p (see Fig. 3; GST-Pex13pN) carrying the BamHI-NotI fragments of PCR products amplified using forward primer PEX13.f and reverse primer PEX13N.r was subcloned into the BamHI-NotI site of pGEX6P-1. The expression plasmid for GST fused to the C-terminal part consisting of residue 255 to the C terminus of HsPex13p (GST-Pex13pC) carrying the EcoRI-NotI fragment of products of PCR with forward primer PEX13SH3.f and reverse M13.r was inserted into the EcoRI-NotI site of pGEX6P-1.

FIG. 3.

Domain mapping of Pex13p involved in interaction with Pex5p, Pex7p, and Pex14p. (A) Schematic representation of human Pex13p constructs. Numbers, positions of amino acid residues of Pex13p; solid boxes, transmembrane segments; hatched boxes, SH3 region. (B) GST fused with full-length (FL) and N- and C-terminal parts (N and C) of Pex13p was expressed in E. coli as in Fig. 1. Purified GST-Pex13p variants (1 μg each) were assessed by SDS-PAGE; proteins were stained with Coomassie blue. (C) Binding assay of GST-Pex13p fusions and variants. GST fusion proteins (2 μg each) were incubated with recombinant Pex5pL, His₆-Pex7p, and His₆-Pex14p (1 μg each). After a thorough washing, proteins bound to Sepharose beads were analyzed by SDS-PAGE. Pex5p, Pex7p, and Pex14p were detected by immunoblotting using the respective antibodies.

Expression plasmids for GST-catalase and GST-catalase lacking C-terminal KANL.

Expression plasmids for the GST fusion with human catalase and catalase with C-terminal tetrapeptide KANL deleted (GST-catalase and GST-catalaseΔKANL, respectively) were constructed as follows. Human catalase cDNA, termed HsCAT, was cloned by a PCR-based procedure using as a template a human liver cDNA library with forward primer HsCatalase.f and reverse primer HsCatalase.r. SalI fragments of PCR products were subcloned into the SalI site in the pGEX6P-3 vector (Amersham Pharmacia Biotech). To construct an expression plasmid for GST-catalaseΔKANL, an XhoI fragment of PCR products likewise amplified by using template HsCAT with forward primer 1148.f and reverse primer ΔKANL.r was placed into the XhoI fragment of HsCAT in pGEX6P-3. These constructs were assessed by nucleotide sequence analysis.

Purification of recombinant proteins.

E. coli DH5α cells transformed with cDNAs coding for GST and GST fusion proteins, including GST-ClPex5pS (29) and GST-Pex14p (38), in pGEX6P-1 were cultured overnight. Five hundred microliters of each of the cultures of E. coli expressing GST and GST fusion proteins was diluted in 5 ml of yeast extract-tryptone medium. After being cultured for 1 h at 37°C, cells were further grown for 2 h at 37°C in the presence of 0.1 mM isopropyl-β-d-thiogalactopyranoside (IPTG) to induce the expression of fusion proteins at adequate levels. For GST-Pex13p, E. coli cells were cultured for 1 h at 37°C and then the cells (the number of cells was approximately equal to the number used for the previous culture) were further grown for 8 h at 18°C in the presence of 0.1 mM IPTG. Harvested cells were resuspended in 400 μl of ice-cold suspension buffer consisting of 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% Triton X-100, 10% glycerol, 1 mM phenylmethylsulfonyl fluoride, 1 mM EDTA, and 1 mM dithiothreitol, sonicated, and centrifuged to remove cell debris. The resulting supernatant was mixed with 60 μl of a 50% slurry of glutathione-Sepharose beads (Amersham Pharmacia Biotech), and the beads were washed three times with the suspension buffer. After being incubated for 2 h at 4°C, Sepharose beads were washed and used for an in vitro binding assay of the 50% slurry. Pex5p, Pex13p, catalase, and catalaseΔKANL were isolated from GST-Pex5p, GST-Pex13p, GST-catalase, and GST-catalaseΔKANL, respectively, by cleaving with PreScission protease (Amersham Pharmacia Biotech) according to the manufacturer's protocol. His₆-Pex14p and His₆-Pex7p were expressed in E. coli and purified as described previously (38). An E. coli expression plasmid coding for C-terminally His₆-PTS1(SKL)-tagged EGFP, termed His₆-GFP-SKL, was generated by PCR using pEGFP (Clontech) as the template and primer pair GFP-SKL.f and GFP-SKL.r. A SalI-NcoI fragment of the PCR product was placed into the SalI-NcoI site of pEGFP. Expressed His₆-GFP-SKL was purified as described above.

In vitro binding assay.

Two types of GST pull-down assays were performed. One used GST fusion proteins and Pex5p, His₆-Pex7p, His₆-Pex14p, Pex13p, His₆-GFP-SKL, catalase, or catalaseΔKANL. GST or GST fusion proteins (typically 5 μg each), were loaded to glutathione-Sepharose beads and then incubated with Pex5p (1 μg), His₆-Pex7p (1 μg), His₆-Pex14p (1 μg), Pex13p (0.1 μg), His₆-GFP-SKL (1 μg), catalase (1 μg), or catalaseΔKANL (1 μg), respectively, by rotation for 2 h at 4°C in an in vitro binding buffer consisting of 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% Triton X-100, 10% glycerol, 1 mM phenylmethylsulfonyl fluoride, 1 mM EDTA, and 1 mM dithiothreitol. Glutathione-Sepharose beads were collected by centrifugation and washed three times with the binding assay buffer minus glycerol. Bound fractions were eluted with 50 μl of sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) sample buffer. An in vitro binding assay was also performed using cell lysates of 207P7, a stable human PEX7 transfectant of pex7 CHO mutant ZPG207 (19) expressing a higher level of Pex7p (29). 207P7 cells (10⁶) were lysed on ice with the binding assay buffer for 1 h and centrifuged at 100,000 × g for 40 min at 4°C. The supernatant fraction (typically 400 μl) was incubated with GST- or GST fusion protein (5 μg)-bound glutathione-Sepharose beads (30 μl) in 700 μl of buffer by rotation for 2 h at 4°C. The Sepharose beads were washed three times with the binding assay buffer minus glycerol. Bound proteins were analyzed by SDS-PAGE and immunoblotting.

Other methods.

Western blot analysis was performed on samples transferred electrophoretically to a polyvinylidene difluoride membrane (Bio-Rad, Hercules, Calif.) with primary antibodies and a second antibody, donkey anti-rabbit immunoglobulin G antibody conjugated with horseradish peroxidase (Amersham Pharmacia Biotech). The antigen-antibody complexes were revealed with enhanced chemiluminescence Western blotting detection reagent ECL (Amersham Pharmacia Biotech).

RESULTS

Purification and binding assays of Pex5p deletion mutants.

To investigate the function and interacting partners of Pex5pL involved in PTS1 and PTS2 import, we first prepared various constructs encoding a number of truncated forms of Pex5pL (Fig. 1A). To obtain a large quantity of homogeneous proteins, we created GST-fused versions of various Pex5pL deletion mutants and expressed them in E. coli. With all recombinant deletion mutants high levels of expression were noted. The amount and purity of GST and GST fusion proteins bound to glutathione-Sepharose were assessed by staining with Coomassie blue (Fig. 1C), where several GST fusion proteins such as GST-Pex5pL(118-243) showed anomalous migration during SDS-PAGE (lanes 9 to 13).

FIG. 1.

Functional domain mapping of Pex5pL. (A) Schematic representation of Chinese hamster ClPEX5L product constructs. The truncated products of ClPEX5L (the longer isoform of ClPEX5) were expressed in E. coli as fusion proteins placed at the C terminus of GST. Numbers, amino acid residues of ClPex5pL; vertical bars, highly conserved pentapeptide WXXXF/Y motifs; hatched bars, Pex5pL-specific 37-amino-acid sequence. (B) Interaction of Chinese hamster Pex5pL variants with other peroxins and PTS1. Binding assays were performed using recombinant fusions of GST to Pex5pL and its variants. GST fusion proteins (5 μg each) were incubated with lysates of 207P7 cells (10⁶) expressing a high level of Pex7p. Purified Pex13p (0.1 μg) and His₆-GFP-SKL (5 μg) were separately incubated with glutathione-Sepharose beads conjugated to GST fusion proteins (5 μg each). After a thorough washing, proteins bound to Sepharose beads were analyzed by SDS-PAGE on 12% gels and immunoblotting using antibodies specific for Pex7p, Pex13p, Pex14p, and GFP. (C) Heterologous expression and purification of recombinant PEX5L proteins. GST-Pex5pL variants were expressed in E. coli and purified as described in Materials and Methods. GST fusion proteins (1 μg each) were analyzed by SDS-PAGE and stained with Coomassie blue. (D) GST-Pex5pL(190-233) and GST-Pex5pL(190-233)S214F were likewise incubated with the lysates of 207P7 cells (10⁶). Pex7p in bound fractions was probed with an anti-Pex7p antibody (top). GST fusion proteins were stained with Coomassie blue (bottom).

To map the functional region of Pex5pL, we performed GST pull-down assays using full-length and various truncated forms of GST-Pex5pL and cell lysates of 207P7, a stable human PEX7 transfectant of pex7 mutant ZPG207 expressing a higher level of Pex7p. Pex13p was expressed at a rather lower level in normal CHO-K1 and mutant cells (data not shown). We then expressed GST-Pex13p in E. coli and purified Pex13p as described in Materials and Methods. To determine whether Pex5p interacts with Pex13p, we used recombinant Pex13p (Fig. 1B). His₆-GFP-SKL was likewise assessed for purity. The fractions bound to various GST-Pex5pL truncation mutants were probed with rabbit antibodies to Pex7p, Pex13p, Pex14p, and GFP, respectively. Pex7p, Pex13p, Pex14p, and His₆-GFP-SKL (PTS1 protein) did not bind to GST (Fig. 1B, lane 1). Deletion from Pex5pL of the complete C-terminal half containing the whole TPR domain eliminated PTS1 binding activity but did not exert any detectable effects on Pex5p interaction with Pex7p, Pex13p, and Pex14p (Fig. 1B, lanes 2 and 3). These results indicated that the N-terminal half of Pex5pL, containing a Pex5pL-specific 37-amino-acid insertion, is sufficient for the interaction with Pex7p, Pex13p, and Pex14p.

Binding to interacting partners Pex7p, Pex13p, and Pex14p. (i) Interaction with Pex7p.

In higher eukaryotes including humans and Chinese hamsters two isoforms of Pex5p which differ in an insertion of 37 amino acids as a result of alternative splicing have been reported. We previously demonstrated that Pex5pL directly interacts with the PTS2 receptor Pex7p, suggesting that this interaction plays an important role in mammal-specific PTS2 transport. GST(206-262) did not interact with Pex7p, although this region contains the entire 37-amino-acid insertion (Fig. 1B, lane 19), indicating that an upstream region of the 37-amino-acid insertion is also crucial for the interaction with Pex7p. Pex7p was in fractions bound to GST(1-243), lacking the sequence from amino acid 244 to the C terminus, but not in those bound to GST(1-184), thus demonstrating that the 58-amino-acid sequence between amino acids 185 and 243 is required for binding to Pex7p (Fig. 1B, lanes 3 and 4). Next, to determine the minimum region of Pex5pL responsible for Pex7p binding, we constructed four more GST-fused truncation mutants (with fusion on one or both sides of the N and C termini): GST(185-243), GST(190-243), GST(190-233), and GST(190-223). Pex7p was in fractions bound to truncation forms GST(185-243), GST(190-243), and GST(190-233) but not to GST(190-223), indicating that the 43-amino-acid region between positions 190 and 233 with the N-terminal 18 residues of the Pex5pL-specific 37-amino-acid insertion is required for binding to Pex7p (Fig. 1B, lanes 15 to 18). GST(190-215) and GST(140-215), both devoid of the 37-amino-acid sequence, did not bind to Pex7p (data not shown), indicating that the Pex5pL-specific insertion is indeed required for the interaction with Pex7p. Moreover, GST(190-233) containing the S214F mutation did not bind Pex7p (Fig. 1D); this is in line with our recent observation using the full-length Pex5pL-S214F (26). Therefore, the N-terminal part of the Pex5pL-specific 37-amino-acid insertion, possibly shorter than 18 amino acids but longer than 8 residues, plus its upstream region comprising less than 26 residues are most likely to be involved in the Pex5pL-Pex7p interaction.

(ii) Interaction with Pex13p.

It was shown in several yeast species that Pex13p interacts directly with Pex5p, whereas a physical interaction between mammalian Pex13p and Pex5p remained unclear. We next purified Pex13p from GST-Pex13p by cleaving with PreScission protease and investigated whether mammalian Pex5p interacts with Pex13p by a similar GST pull-down assay. Pex13p was detected in the bound fraction of GST-Pex5pL, whereas Pex13p was not detectable in the fraction recovered with GST (Fig. 1B, lanes 1 and 2), indicative of specific binding of Pex5p to Pex13p. Moreover, Pex13p was detected in fractions bound to GST(1-243) at levels similar to that noted by using the full-length Pex13p but not to that noted by using GST(306-632) containing the C-terminal TPR domain (Fig. 1B, lanes 2, 3, and 20), demonstrating that the Pex13p-binding site resides at the N-terminal half of Pex5pL. GST(118-243) and GST(140-243) showed the same level of binding to Pex13p as GST(1-243), suggesting that the region of residues 140 to 243 is involved in binding to Pex13p (Fig. 1B, lanes 9 and 10). It is noteworthy that this part of Pex5pL includes pentapeptide WXXXF/Y repeats 2 to 4 (Fig. 1A). It has recently been reported that the WXXXF/Y motif of the yeast S. cerevisiae Pex5p is important for the interaction with Pex13p (5). Several more deletion variants containing two of WXXXF/Y motifs 2 to 4, such as GST(1-184), GST(1-163), GST(145-243), and GST(145-189), showed a reduced level of Pex13p-binding activity compared to GST(1-243) (Fig. 1B, lanes 4, 5, 11, and 13). Other truncation mutants apparently had no binding activity (Fig. 1B). Collectively, these results suggested that pentapeptide repeats 2 to 4 form a site for the binding of Pex5p to Pex13p.

(iii) Interaction with Pex14p.

In yeast and higher eukaryotes the interaction between Pex5p and Pex14p has been demonstrated by the yeast two-hybrid system and in vitro binding assay. To determine the Pex14p-binding site(s) of Pex5p, we performed GST pull-down assays using GST-Pex5pL variants and 207P7 cell lysates. Pex14p was detected in fractions bound to GST(1-243) but not to GST(306-632) of the C-terminal TPR domains, indicating that the N-terminal half of Pex5pL was required for binding to Pex14p (Fig. 1B, lanes 3 and 20). This result was in line with a recent observation using human Pex5p by Schliebs et al. (36). All of the Pex5pL truncation forms containing WXXXF/Y motifs, except for GST(185-243), bound to Pex14p in our in vitro binding assays (Fig. 1B, lanes 4 to 15 and 19). In contrast, three truncation forms lacking the WXXXF/Y motif, GST(190-243), GST(190-233), and GST(190-223), did not interact with Pex14p (lanes 16 to 18). Accordingly, these results strongly suggested that Pex5p (both Pex5pL and Pex5pS) possesses multiple binding sites for Pex14p, which appear to be distributed throughout the WXXXF/Y motifs located in the N-terminal half of Pex5p. A very similar finding for human Pex5p has recently been reported (35, 36).

His₆-GFP-SKL was detected in the bound fraction of GST(306-632) containing the domain comprising TPR motifs 1 to 7. (Fig. 1B, lane 20), whereas His₆-GFP-SKL was undetectable in the fractions recovered with other truncation forms (lanes 3 to 19). In view of these findings, it is more likely that the N-terminal half of Pex5p is involved in targeting Pex5p-cargo complexes to peroxisomes, while the C-terminal TPR domain binds PTS1 proteins.

Complementing activity of various truncation forms of Pex5pL.

Coding sequences for various truncated forms of Pex5pL were separately subcloned downstream of the SRα promoter in mammalian expression vector pUcD2SRαMCSHyg. ZP105 cells defective in PTS1 and PTS2 import (31) were transfected with each of the PEX5L variants and were stained with the antibody to 3-ketoacyl-CoA thiolase for assessing complementing activity of PTS2 import (Table 2). When ZP105 cells were transfected with cDNA encoding full-length Pex5pL, but not with the vector only, thiolase was found in a punctate staining pattern, indicating complementation of PTS2 protein import (Fig. 2A, a and c). In ZP105 cells expressing Pex5pL(1-243), larger but fewer PTS2-positive particles were observed (Fig. 2A, b-1). We earlier reported that the N-terminal 40-amino-acid sequence of Pex3p fused with EGFP, termed Pex3p(1-40)-EGFP, was targeted not only to peroxisomes in the wild-type CHO-K1 cells but also to “peroxisome ghost” vesicles in CHO mutants defective in PTS1 and PTS2 import (20). To investigate whether these thiolase-positive larger particles are peroxisome ghost-like vesicles, ZP105 cells were cotransfected with PEX5L(1-243) and PEX3(1-40)-EGFP. In cells expressing Pex5pL(1-243), thiolase-positive particles with superimposable EGFP-positive punctate structures were observed, indicating that thiolase was translocated into peroxisome ghost-like vesicles (Fig. 2A, b-1 and b-2). These results were interpreted to mean that Pex5pL(1-243) is biologically active in PTS2 transport. A similar complementing assay was done using solely PTS2 import-defective pex5 mutant ZPG231 cells that had been transformed with PTS2-GFP (Fig. 2A, e to h) (26). In cells expressing Pex5pL(1-243), PTS2-GFP was localized in apparently normal peroxisomes as observed in cells carrying full-length PEX5L (Fig. 2A, f and g). Furthermore, to determine whether PTS1 protein import is required for normal peroxisome biogenesis, PEX5S encoding Pex5pS possessing only PTS1 import activity and PEX5L(1-243) were cotransfected into ZP105 cells. Thiolase was detected in peroxisomes, which were apparently normal in size and numbers, in Pex5pL(1-243)- and Pex5pS-coexpressing ZP105 cells (Fig. 2B, c), whereas it was discernible in peroxisome ghost-like particles solely in Pex5pL(1-243)-expressing cells (Fig. 2B, b). These results strongly suggested that PTS1 import is required for normal peroxisome biogenesis. We previously demonstrated that the cellular phenotype of PEX5-defective ZP105 is temperature sensitive (29). Pex5p was hardly detectable in ZP105 cells when they were cultured for 3 days at 37°C. After a shift to 30°C, Pex5p was clearly discernible after 3 days of culture. To examine whether PEX5L(1-243)-transfected ZP105 cells show a temperature-sensitive phenotype, the level of expression of endogenous Pex5p was assessed by immunoblotting with an anti-Pex5p antibody using the lysates of mock- and PEX5L(1-243)-transfected ZP105 cells. Endogenous Pex5p expression was not elevated by transfection of PEX5L(1-243) or by mock transfection (data not shown). Together, these results confirmed that the restoration of impaired PTS2 import found in these complementing assays relies on the biological activities per se of the respective constructs. In contrast, Pex5pL(1-184), possessing interacting activities for Pex13p and Pex14p but not Pex7p, showed no complementing activity (Table 2). Likewise, none of the C-terminally truncated variants of Pex5pL(1-184), including Pex5pL(1-163), Pex5pL(1-158), Pex5pL(1-144), and Pex5pL(1-139), restored PTS2 protein import in ZP105 cells (Table 2), consistent with the in vitro binding studies described above (Fig. 1). Collectively, the N-terminal region of Pex5pL, at least from residue 1 to 243, was required for PTS2 protein transport. Thus, it is more likely that the N-terminal part, comprising 243 amino acid residues and having multiple biological activities, is involved in Pex5p targeting to and translocation through the peroxisomal membrane machinery.

TABLE 2.

Functional domain mapping of Pex5pL^a

^aVarious constructs of PEX5L shown as in Fig. 1A were transiently expressed in pex5 mutants ZP105 and ZPG231. ZP105 cells were stained using antibodies to thiolase (PTS2 protein) and PTS1. ZPG231 cells were assessed for import of PTS2-GFP by GFP fluorescence microscopy.

^b+, complemented; −, not complemented.

^cThe in vitro binding assay was performed as described in Materials and Methods. +, bound; −, not bound.

FIG. 2.

N-terminal part of Pex5pL, Pex5pL(1-243), is sufficient for PTS2 import. (A) Expression of Pex5pL(1-243) lacking the C-terminal TPR region in pex5 CHO cell mutants ZP105 and ZPG231. ZP105 cells were transfected with the vector only (a),ClPEX5L(1-243) plus PEX3(40AA)-EGFP encoding N-terminal rat Pex3p comprising 40 amino acid residues fused to EGFP (20) (b-1 and b-2), full-length ClPEX5L (c), and ClPEX5L(1-243) (d). ZPG231 cells were also transfected with the vector only (e and h-1),ClPEX5L(1-243) (f and h-2), and full-length ClPEX5L (g). Thiolase, a PTS2 protein, was visualized with an antithiolase antibody (a, b-1, and c), and Pex3p(40AA)-EGFP was visualized by fluorescence microscopy (b-2). PTS1 proteins were stained with an anti-PTS1 antibody (d, h-1, and h-2). PTS2-GFP was visualized by fluorescence microscopy (e to g). Magnification, ×650; bar, 20 μm. Note that thiolase was seen in larger but fewer particles, in a manner superimposable with that for Pex3p(40AA)-EGFP (b-1 and b-2). (B) Expression of Pex5p reduces PTS2-containing large peroxisomes to normal size. ZP105 cells were transfected with full-length ClPEX5S (a), ClPEX5L(1-243) plus vector (b), and ClPEX5L(1-243) plus ClPEX5S (c). Cells were stained with an antithiolase antibody. Bar, 20 μm.

Moreover, Pex5pL(118-243) did not restore PTS2 protein import in ZP105 and ZPG231 cells, although this region is sufficient for binding to Pex7p, Pex13p, and Pex14p, implying that the N-terminal 117-amino-acid region is important for peroxisomal protein import activity of Pex5p (Table 2). We recently demonstrated that Pex5p forms homo- or heterodimers (29). The N-terminal half has recently been reported as a region required for oligomerization of Pex5p (36). Hence, it is tempting to speculate that the N terminus 117 amino acids function as a dimerization domain and that such dimer formation is essential for PTS protein import activity of Pex5p. The highly conserved C-terminal half comprising seven TPR motifs has been shown to mediate the binding to PTS1-containing proteins. Pex5pL(306-632), containing the whole C-terminal TPR domain, showed no complementing activity in pex5 ZP105 cells (Table 2). The same results with respect to the complementation of PTS2 import by Pex5pL variants were obtained with another PEX5-deficient mutant, ZPG231, defective solely in PTS2 import (Table 2).

Pex5p and Pex7p bind to the N-terminal domain of Pex13p.

Pex13p is a peroxisomal integral membrane protein with two transmembrane segments, exposing both the N-terminal part and the C-terminal SH3 domain to the cytoplasm (43) (Fig. 3A). The SH3 domains are small noncatalytic protein modules capable of mediating protein-protein interaction by binding to a Pro-X-X-Pro (PXXP) motif (9). In S. cerevisiae, both Pex5p and Pex14p were shown to bind to the SH3 domain of Pex13p (5). Pex14p contains the classical SH3 binding motif PXXP, whereas this motif is absent from yeast Pex5p. Recently, it has been reported that Pex14p is the classical SH3 domain ligand and that Pex5p binds the SH3 domain in an alternative way (4). Similarly, in mammals, classical SH3 binding motif PXXP is not found in the minimal Pex13p-binding region (amino acids 140 to 243) of Pex5pL. Instead, three WXXXF/Y motifs exist in this region (Fig. 1). To search for a mammalian Pex5p- and Pex14p-binding region(s) in Pex13p, in vitro binding assays using bacterially expressed recombinant proteins were performed (Fig. 3B). GST-fused full-length Pex13p and N-and C-terminal parts of Pex13p, termed GST-Pex13pN and GST-Pex13pC, as well as GST were incubated with Pex5pL and His₆-tagged Pex14p. The amount and purity of GST and GST fusion proteins bound to glutathione-Sepharose were assessed by staining with Coomassie blue (Fig. 3B). His₆-Pex14p was detected in the bound fractions of GST-Pex13p and GST-Pex13pC (Fig. 3C, bottom, lanes 2 and 3). This interaction was not detectable with GST-Pex13pN and GST alone (Fig. 3C, bottom, lanes 1 and 4), suggesting that the binding is dependent on the presence of the Pex13p-SH3 domain. In contrast, Pex5p was recovered with GST-Pex13p and GST-Pex13pN, not with GST-Pex13pC and GST alone (Fig. 3C, top). We obtained essentially the same results by using 207P7 cell lysates in place of recombinant peroxins (data not shown). It is noteworthy that Pex5p binds to the SH3-containing C-terminal portion of Pex13p in yeast (5). Accordingly, our finding implies that a different manner of Pex5p-Pex13p interaction exists in mammalian cells.

We next investigated whether Pex7p binds to Pex13p by an in vitro binding assay. His₆-Pex7p was detected in the bound fractions of GST-Pex13p and GST-Pex13pN, but not in those of GST and GST-Pex13pC (Fig. 3C, middle). Pex7p was likewise detectable in fractions bound to GST-Pex13p and Pex13pN when 207P7 cell lysates were used (data not shown). This mode of interaction was consistent with that in S. cerevisiae, where Pex7p directly or indirectly interacted with an N-terminal region of Pex13p (21).

WXXXF/Y motifs of Pex5p are essential for the interaction with Pex13p and Pex14p.

We showed that mammalian Pex5p possesses multiple binding sites for Pex14p (Fig. 1). Pex5pL contains seven WXXXF/Y motifs in the N-terminal half which are conserved among PEX5 proteins of different species, from yeast to humans. There are six motifs in mammalian Pex5pS. In S. cerevisiae, the W²⁰⁴XXQF²⁰⁸ motif in 24 amino acid residues (Pro²⁰³ to Lys²²⁷) of the N-terminal Pex5p was shown to be essential for binding to Pex13p (4). We showed that the minimal Pex13p-binding stretch of mammalian Pex5p contains three pentapeptide repeats, WXXXF/Y motifs 2 to 4 (Fig. 1). We further investigated whether the WXXXF/Y motifs of mammalian Pex5p are essential for interaction with Pex14p and Pex13p. To determine the structural requirement for the binding of the WXXXF/Y motifs to Pex14p and Pex13p, alanine substitution was carried out on the conserved amino acids at positions 1 (W) and 5 (F/Y) in each of the four WXXXF/Y motifs in GST fusion protein Pex5pL(1-243)HA, a part sufficient for transport of the PTS2 protein (Fig. 4, left). We first verified that these Ala mutants bound to recombinant Pex13p. Mut1, the Ala-substituted form of the first WXXXF/Y (motif 1), located upstream the minimal Pex13p-binding region (amino acids 140 to 243), interacted with Pex13p as efficiently as the wild type, while negative results were obtained with GST (Fig. 4, lanes 1 to 3). In contrast, the binding of Mut2, Mut3, Mut4, Mut12, and Mut13 to Pex13p was mildly reduced (lanes 4 to 7 and 10). Mut23 and Mut14 were significantly affected in the interaction with Pex13p (Fig. 4, lanes 8 and 11). Pex13p was not detectable in Mut34, Mut24, Mut123, Mut234, Mut124, Mut134, and Mut1234 (lanes 9 and 12 to 17). These results strongly suggested that WXXXF/Y motifs 2 to 4 of the minimal Pex13p-binding segment of Pex5p were essential for binding activity, consistent with recent observations for two yeast species (4, 5, 45), such as observations of residues W¹⁰⁰ to E²¹³ of P. pastoris Pex5p. Thus, it is most likely that mammalian Pex5p interacts with the N-terminal part, not with the C-terminal SH3 domain, of Pex13p via WXXXF/Y motifs 2 to 4.

FIG. 4.

Mutation of the WXXXF/Y pentapeptide motifs affects binding of Pex5p(1-243) to Pex14p and Pex13p, but not to Pex7p. The pull-down assay was done as for Fig. 1, with fusions of GST to Pex5pL(1-243) variants (5 μg each) and 207P7 cell lysates (10⁶ cells) or recombinant human Pex13p (0.1 μg). Mut1 represents GST-Pex5pL(1-243) with mutations W118A and F122A in the pentapeptide motif 1 (Fig. 1A); Mut2, Mut3, and Mut4, are variants with the same double mutations in motifs 2, 3, and 4, respectively. Other variants with mutated, multiple pentapeptide motifs were likewise constructed by Ala replacement of Trp and Phe/Tyr in motifs 1 to 4. S214F(1-243) represents GST-Pex5pL(1-243) with mutation S214F identified in pex5 mutant ZPG231 (26). Bound peroxins were detected by immunoblotting using specific antibodies. CBB, Commassie blue-stained GST and GST-fused Pex5pL(1-243) variants (1 μg each). GST (lane 1) is shown irrespective of its molecular size compared with those of the others (lanes 2 to 18).

Next, we investigated whether or not mutations in the WXXXF/Y motifs of Pex5p affect the interaction with Pex14p. Pentapeptide variants with a single motif mutated, Mut1, Mut2, Mut3, and Mut4, bound to Pex14p similarly to the wild type, whereas results for GST were negative (Fig. 4, lanes 1 to 6). In contrast, replacing with alanine eight aromatic residues of all four WXXXF/Y motifs (Mut1234) completely abolished the activity of binding Pex14p (Fig. 4, lane 17). Multiple-WXXXF/Y motif mutants containing mutated motif 1 were associated with severely affected binding to Pex14p (Fig. 4, lanes 7, 10, 11, 15, and 16), whereas all of other mutants showed a moderately reduced level of interaction with Pex14p (lanes 8, 9, 12, and 14). These results were interpreted to mean that the WXXXF/Y motifs act as Pex14p-binding sites with different affinities. It is also possible that the first motif creates the highest-affinity site for Pex14p, whereas the fourth WXXXF/Y motif creates a lower-affinity site or does not bind at all, as shown by Mut123, which showed only barely detectable binding to Pex14p (Fig. 4, lane 13, and 1B, lanes 8 and 15).

Pex7p was detected in the bound fractions of all WXXXF/Y motif mutants, implying that the structural requirements of the WXXXF/Y motifs are specific for binding to Pex13p and Pex14p. Pex5pL(1-243)S214F showed normal binding to Pex13p and Pex14p, while binding to Pex7p was completely eliminated (Fig. 4, lane 18). Together, in vitro binding assays demonstrated that the WXXXF/Y motifs were essential for interaction with both Pex13p and Pex14p.

WXXXF/Y motifs are essential for PTS2 import in vivo.

We and others demonstrated several lines of evidence that mammalian Pex5p possesses multiple binding sites for Pex14p and proposed that WXXXF/Y motifs might provide the structural basis for these interactions. However, the physiological significance of such multiple Pex14p-binding sites in vivo is not understood. We next investigated whether mutation of the WXXXF/Y motif in Pex5p affects PTS import activity in vivo. Wild-type Pex5pL(1-243)HA and Pex5pL(1-243)HA WXXXF/Y mutants were expressed in pex5 mutant ZP105 cells, and PTS2 protein import activity was verified by staining cells with an antithiolase antibody. Wild-type Pex5pL(1-243)HA restored PTS2 import (Fig. 5a and b), whereas in cells expressing Mut123, Mut1234, or Mut(S214F) thiolase-positive particles were not observed, indicative of a failure in PTS2 protein import (Fig. 5m, q, and r). Triple-WXXXF/Y-motif mutants Mut124 and Mut134 showed an apparent decrease in PTS2-positive particles (Fig. 5o and p). Expression of other types of mutants, including all combinations of mutants with a single motif as well as with multiple-motif mutations, restored PTS2 import as did wild-type Pex5pL(1-243)HA (Fig. 5c to l and n). These results are in good agreement with the findings of the in vitro binding studies, where only Mut123 and Mut1234 did not bind to Pex14p, implying that the interaction of Pex5pL(1-243) with Pex14p is essential for PTS2 import. Collectively, Pex5pL(1-243) mutants with at least one wild-type WXXXF/Y sequence of motifs 1 to 3, irrespective of the status of motif 4, are potent in binding to Pex14p as well as in complementing the impaired PTS2 import. Multiple sites for Pex14p binding are not a primary prerequisite for the function of Pex5p in PTS protein import, although the Pex5p-Pex14p interaction plays a pivotal role in the PTS2 protein import. Surprisingly, Mut34, Mut24, Mut234, Mut124, and Mut134, despite the completely abolished Pex5p-Pex13p interaction (Fig. 4), were functional in PTS2 protein import in vivo (Fig. 5l and n to p), suggesting that the Pex5p-Pex13p interaction is not required for PTS2 protein import. Essentially the same results were obtained using another pex5 mutant, ZPG231, solely defective in PTS2 import and expressing Pex5pL-S214F (data not shown). We earlier demonstrated that Pex7p-PTS2 complexes are transported by Pex5pL-mediated docking to Pex14p (29). The physiological relevance of the Pex7p-Pex13p interaction in PTS2 import remains to be defined.

FIG. 5.

Mutation of the pentapeptide motifs in Pex5pL(1-243) affects PTS2 import. Ala-substituted WXXXF/Y pentapeptide motif variants of Pex5p(1-243) were expressed in pex5 ZP105 cells. PTS2 import was assessed by staining with an antithiolase antibody. Magnification, ×650; bar, 20 μm.

WXXXF/Y motifs 1 and 5 are most essential for Pex14p binding.

To further characterize the multiple WXXXF/Y motifs with respect to a potential difference in the requirement for binding to Pex14p, we verified the activity of each motif for binding to Pex14p. We constructed a fusion protein consisting of GST fused to various truncated forms of Pex5pL, each containing one pentapeptide motif with N- and C-terminal flanking regions, and expressed them in E. coli (Fig. 6). The interaction between these Pex5pL fragments and Pex14p was examined as for Fig. 1 and 4. The Pex5pL fragment comprising residues 1 to 139, containing WXXXF/Y motif 1, strongly interacted with Pex14p (Fig. 6A, lane 2), as seen in Fig. 1A. The WXXXF/Y motif 5-containing fragment likewise bound to Pex14p (Fig. 6A, lane 10). Pex14p was detected at a low level on incubation with a fragment of either motif 2 or 3 (lanes 4 and 6), whereas Pex14p was barely detectable when a fragment containing motif 4, 6, or 7 was used (lanes 8, 11, and 12), as was the case with GST (lane 1). No activity of WXXXF/Y motif 4 binding to Pex14p was consistent with the in vitro finding that WXXXF/Y motif 4 was not essential for Pex14p binding (Fig. 1 and 4). Moreover, replacement by A of W and F/Y in WXXXF/Y motifs 1 to 3 completely abolished the binding to Pex14p (Fig. 6A, lanes 3, 5, and 7), whereas mutant motif 4 showed negative binding, as did the wild-type form (lanes 8 and 9).

FIG. 6.

Pex14p binds Pex5p at multiple sites. (A) Fusions of GST with Pex5pL fragments, each containing one of WXXXF/Y motifs 1 to 7, were verified to bind to Pex14p using 207P7 cell lysates (10⁶ cells). Numbers, positions of amino acid residues of Pex5pL. Bound Pex14p was detected as for Fig. 1B. (B) Pex14p binding assays were performed as for Fig. 1B by using GST-Pex5pL fragments, each encompassing two tandem pentapeptide motifs. Symbols are as for Fig. 1A.

Pex14p binding to GST-fused Pex5pL fragments, each possessing two WXXXF/Y motifs on the N- and C-terminal sides, was likewise verified. Fragments, each containing N-terminally located WXXXF/Y motif 1, 3, or 5, strongly interacted with Pex14p (Fig. 6B, lanes 2, 4, and 6). Pex14p was detected at a lower level with the fragment containing motifs 6 and 7, while a barely detectable level of Pex14p was found by using other fragments with N-terminally placed motifs 2 and 4 or GST (Fig. 6B, lanes 1, 3, 5, and 7). Taken together, WXXXF/Y motifs 1, 3, and 5 are most likely to be the highest-affinity sites for Pex14p binding.

Pex5p-Pex13p interaction is essential for import of catalase in vivo, but not for PTS1 and PTS2 transport.

To investigate in vivo the consequence of the interaction between Pex5p and Pex13p, we constructed a PEX5L mutant encoding Pex5pL variant Pex5pL(Mut234), defective for the interaction with Pex13p, in pUcD2SRαMCSHyg and pGEX6P-1 (Fig. 7A). The purity of GST fusion proteins bound to glutathione-Sepharose was assessed by SDS-PAGE (Fig. 7B). The interaction of wild-type and mutant Pex5pL with other proteins, including several peroxins, was verified. In GST pull-down assays, fractions bound to GST-Pex5pL using recombinant Pex13p and 207P7 cell lysates were analyzed by SDS-PAGE and immunoblotting using specific antibodies. Pex13p was detected in fractions bound to wild-type Pex5pL but not in those bound to Pex5pL(Mut234) (Fig. 7C, lanes 2 and 3), indicating that Pex5pL(Mut234) was severely impaired in binding to Pex13p, consistent with the findings using GST-fused Pex5pL(1-243)Mut234 (Fig. 4). Unlike Pex13p, Pex14p and Pex7p interacted with both wild-type and mutant Pex5pL but not with GST (Fig. 7C, lanes 1 to 3). PTS1 protein AOx apparently bound to both wild-type and mutant Pex5pL but not GST. Binding to an oligomeric AOx, as assessed with the AOx-B component, was in good agreement with our earlier observation using Pex5pS and Pex5pL (29). Furthermore, catalase possessing PTS1-like C-terminal sequence KANL also bound to wild-type and mutant Pex5pL, but not to GST (Fig. 7C, bottom). Thus, Pex5pL(Mut234) was indistinguishable from the wild type in the interaction with peroxisomal proteins, including cargo proteins such as PTS1, catalase, and interacting peroxins Pex7p and Pex14p, except for Pex13p.

FIG. 7.

Mutation in pentapeptide motifs 2 to 4 of Pex5p inhibits Pex13p binding and catalase import. (A) Schematic representation of Pex5pL with Ala substitutions in WXXXF/Y motifs 2 to 4, (Pex5pL-Mut234). The pentapeptide motifs, 37-amino-acid insertion, and TPR motif region are shown as in Fig. 1A. (B) GST and GST fused to Pex5pL and Pex5pL-Mut234 were expressed in E. coli, purified, and analyzed by SDS-PAGE. Lanes: 1, GST (1 μg); 2, GST-Pex5pL (1 μg); 3, GST-Pex5p-Mut234 (1 μg). WT, wild type. (C) Interaction of GST, GST-Pex5pL, and GST-Pex5p-Mut234 (5 μg each) with peroxins and cargo proteins was verified with 207P7 cell lysates (10⁶ cells) or recombinant human Pex13p (0.1 μg), as in Fig. 1B. Bound protein fractions were assessed by using antibodies specific for Pex13p, Pex14p, Pex7p, AOx, and catalase. AOx-B represents the B component of the AOx hetero-oligomer consisting of 75-kDa A, 52-kDa B, and 23-kDa C polypeptide components (27). (D) Morphological assessment. pex5 ZP105 cells were transfected with a mock vector (a, d, and g) and cDNAs encoding Pex5pL (b, e, and h) or Pex5pL-Mut234 (c, f, and i). Cells were stained using antibodies to catalase (a to c), PTS1 (d to f), and PTS2 protein thiolase (g to i), respectively. Magnification, ×570; bar, 20 μm.

We next transfected PEX5L and PEX5L(Mut234) into pex5 mutant ZP105 cells. Expression of wild-type PEX5L restored the import of catalase (Fig. 7D, b), whereas PEX5L(Mut234) did not complement the cytosolically mislocalized catalase, similar to what was found for mock-transfected cells (c and a), indicating that Pex5pL(Mut234) is not functional in vivo in catalase import. In contrast, a number of PTS1- and PTS2-positive particles were discernible in the PEX5L(Mut234) transfectants (Fig. 7D, f and i) and in wild-type Pex5pL-expressing cells (e and h) but not in mock transfectants (d and g), thereby demonstrating that Pex5pL(Mut234) is functionally active in transport of PTS1 and PTS2 proteins. These results suggest that the Pex5pL-Pex13p interaction is dispensable for PTS1 and PTS2 protein transport, while such an interaction is essential for catalase import. Hence, it is plausible that catalase is translocated by a mechanism distinct, at least at some step(s) if not all, from PTS1 and PTS2 protein import.

Catalase import is dependent on Pex5p TPR and C-terminal PTS more stringently than PTS1.

To further investigate molecular mechanisms of catalase transport, we first assessed the binding region of Pex5p by GST pull-down assays. Catalase was recovered in fractions bound to full-length Pex5pL and the C-terminal TPR-containing part but not to the N-terminal Pex5p and GST (Fig. 8A, lanes 2 to 5). In contrast, catalase with C-terminal tetrapeptide KANL deleted was not detectable in any of the fractions (lanes 8 to 10). These results strongly suggested that catalase bound to Pex5p via C-terminal KANL, consistent with the report that KANL is the PTS of human catalase (32). Thus, a mode of recognition of catalase PTS by Pex5p resembled that of PTS1 but the import mechanisms of these cargoes were apparently distinguishable in the requirement for Pex5p-Pex13p interaction (Fig. 7).

FIG. 8.

Catalase is imported in a Pex5p-dependent manner, but partly distinct from PTS1 proteins. (A) GST pull-down assay. Interaction of GST, GST-Pex5pL, GST-Pex5pL(1-262) [GST(1-262)], and GST-Pex5pL(306-632) [GST(306-632)] (5 μg each) with recombinant catalase (1 μg) (lanes 2 to 5) and a catalase mutant (ΔKANL) with truncation of the C-terminal SKL-like sequence KANL (1 μg) (lanes 7 to 10) was verified. One-tenth aliquots of the input, wild-type, and KANL-truncated catalase were loaded in lanes 1 and 6. Bound fractions were assessed with an anticatalase antibody. (B) Pex5pL mutant Pex5pL-S593W is defective only in catalase transport. Pex5pS-S563W was identified in a CG2 PBD patient (39). Binding of ClPex5pL-S593W (5 μg), equivalent to human Pex5pS-S563W, to PTS1 cargoes (AOx and SCPx with C-terminal tripeptides SKL and AKL, respectively), catalase, and peroxins Pex14p and Pex7p was verified with 207P7 cell lysates (10⁶ cells). Binding to recombinant Pex13p (0.1 μg) was also verified. GST and GST fusion proteins used are indicated at the top. Bound proteins were detected by immunoblotting using antibodies specific for the respective proteins. wt, wild type. (C) pex5 mutant ZP105 cells were transfected with a mock vector (a, d, g, and j), ClPEX5L (b, e, h, and k), and ClPEX5L-S593W (c, f, i, and l). Cells were stained with antibodies to catalase (a to c), AOx (d to f), SCPx (g to i), and thiolase (j to l). Magnification, ×530; bar, 20 μm. Note that mutation S593W in Pex5pL affected only catalase import (c).

The missense mutation of Pex5pS-S563W was shown to be responsible for CG2 PBD, conferring a cell phenotype of partial import of PTS1 and PTS2 proteins but not of catalase (39). It was verified that ClPex5pL-S593W, equivalent to HsPex5pS-S563W in amino acid alignment (31), bound to cargoes and several peroxins. Pex5pL-S593W bound to Pex14p, Pex13p, and Pex7p as efficiently as wild-type Pex5pL (Fig. 8B, lanes 2 and 3). However, the binding of Pex5pL-S593W to typical PTS1-SKL protein AOx and AKL-type SCPx was significantly reduced compared to wild-type Pex5pL binding (lane 3). Binding to catalase was completely eliminated. GST did not interact with any of the cargo proteins and peroxins examined (lane 1). Furthermore, ClPex5pL-S593W was expressed in pex5 ZP105 cells defective in peroxisomal protein import (31) as in mock-transfected cells (Fig. 8C, a, d, g, and j). PTS1 SKL protein AOx, AKL-type SCPx, and PTS2 protein, thiolase were discernible in a punctate staining pattern in PEX5L-trasnfected ZP105 cells (Fig. 8C, e, f, h, i, k, and l), indicating that Pex5pL-S593W was functional in PTS1 import as well as Pex7p-mediated PTS2 transport. A higher level of Pex5pL-S593W expression apparently compensated for the lowered level of PTS1 binding activity, thereby showing the complementation of impaired import of PTS1-type proteins. In contrast, catalase remained in the cytoplasm in PEX5L-S593W-expressing ZP105 cells, while catalase was imported to peroxisomes in normal Pex5pL-expressing cells (Fig. 8C, b and c). Collectively, it is evident that the S593W mutation occurring downstream of the TPR region affects most severely the recognition by Pex5p of the PTS of catalase, KANL, and, less severely, SKL and AKL. It is also possible that the difference between the affinity of Pex5p binding to typical PTS1 SKL and the affinity of its binding to atypical KANL results in such a distinct phenotype.

Pex5p affects Pex13p-Pex14p interaction in a PTS1-dependent manner.

The important issue of whether Pex13p or Pex14p functions as the initial docking receptor for Pex5p-cargo complexes remained to be unequivocally resolved. We recently proposed that Pex14p functions as the initial docking receptor of Pex5p in mammalian cells by making use of CHO pex14 and pex13 mutants (29). In the in vitro binding assays, His₆-GFP-SKL used as a PTS1 cargo protein was detected in the bound fraction upon incubation with GST-Pex14p in the presence of Pex5pL, whereas Pex5pL also bound to GST-Pex14p (Fig. 9A, lane 5). In contrast, His₆-GFP-SKL was not detectable with Pex5pL plus GST-Pex13p, whereas Pex5pL bound to GST-Pex13p (lane 3). Incubation of His₆-GFP-SKL and Pex5pL with GST resulted in a negative response (Fig. 9A, lane 1). Pex5pL directly bound equally to Pex14p and Pex13p in the absence of cargoes (Fig. 9A, lanes 6 and 7), while His₆-GFP-SKL did not bind directly to GST-Pex13p and GST-Pex14p (lanes 2 and 4). These results, taken together, imply that Pex5p-mediated interaction of Pex14p with PTS1 protein was specific. Essentially the same results were obtained with catalase (Fig. 9A, bottom). Accordingly, we conclude that Pex14p functions as the first docking receptor for cargo-loaded Pex5p on peroxisomal membranes. Pex13p binds to the cargo-unloaded Pex5p, implying that it functions at the step(s) after the cargoes are unloaded.

FIG. 9.

Interaction of Pex14p and Pex13p with cargo-loaded or unloaded Pex5p. (A) In vitro binding assays were performed using fusion proteins GST-Pex14p (2 μg), GST-Pex13p (2 μg), Pex5pL (2 μg), and His₆-GFP-SKL (4 μg) (top) or recombinant catalase (4 μg) (bottom). GST pull-down assays were likewise done in the absence of cargoes (top, lanes 6 and 7). Components added to the assay mixtures, including GST in place of GST fusion proteins, are indicated at the top. Pex5pL, His₆-GFP-SKL, and catalase in fractions bound to GST-Pex14p- and GST-Pex13p-linked Sepharose were detected by immunoblotting using antibodies specific for the respective proteins. (B) Formation of a hetero-oligomeric complex comprising Pex14p, Pex13p, Pex5p, and PTS1 cargo protein. Binding assays were done as for panel A using GST-Pex14p (2 μg), purified recombinant proteins, Pex13p (0.1 μg), Pex5pS (2 μg), Pex5pL (2 μg), and His₆-GFP-SKL (4 μg). One-tenth aliquots of the input, Pex5pS, Pex13p, and His₆-GFP-SKL were loaded in lane 10; Pex5pL was in lane 11. Components added to the assay mixtures are indicated at the top. Pex5p, Pex13p, and His₆-GFP-SKL in fractions bound to GST-Pex14p were detected as for panel A.

We further investigated whether or not the Pex13p-Pex14p interaction is affected by cargo-loaded or unloaded Pex5p. GST-Pex14p directly bound to Pex13p (Fig. 9B, lane 4) at the SH3-containing C-terminal part (Fig. 3C). The binding was not observed with GST (lane 1), indicating that the interaction was specific. The Pex13p-Pex14p interaction was apparently inhibited when GST-Pex14p and Pex13p were incubated with Pex5pS or Pex5pL, under which conditions Pex5pS and Pex5pL bound instead to GST-Pex14p (Fig. 9B, lanes 5 and 6). Such preferential Pex5p binding to Pex14p was also noted using Pex5pL-Mut234 solely lacking Pex13p-binding activity (Fig. 7), implying that dissociation of Pex13p-Pex14p complexes did not require the interaction of Pex13p with Pex5p (data not shown). It is more likely that the affinity of the binding of Pex5p to Pex14p is higher than that to Pex13p and of the Pex14p-Pex13p interaction. To our surprise, such a blockade by Pex5p of Pex13p-Pex14p binding was relieved in the presence of PTS1 protein His₆-GFP-SKL (lanes 7 and 8). Pex13p plus Pex5pL and Pex13p plus Pex5pL with His₆-GFP-SKL were not pulled down with GST (Fig. 9B, lanes 2 and 3). His₆-GFP-SKL showed no apparent effect on the Pex13p-Pex14p interaction (lane 9). Using catalase in place of His₆-GFP-SKL, we reproduced essentially the same results as those described above (data not shown). Taken together, these results demonstrated that Pex5p affects the Pex13p-Pex14p interaction in a cargo-dependent manner. Therefore, it is most likely that Pex14p and Pex13p form functionally distinct subcomplexes involved in peroxisomal matrix protein import.

DISCUSSION

Many peroxins are involved in PTS protein targeting and membrane translocation processes, either in a soluble phase (targeting) or as components of potential protein translocation machinery. Pex5p functions in the transport of PTS1 proteins to peroxisomes, a common feature from yeast to humans. However, a distinct functional difference between Pex5pL and Pex5pS in mammalian cells is the pivotal role of Pex5pL in PTS2 transport (26, 29). In the present work, we have provided several lines of evidence that the N-terminal half, residues 1 to 243, of Pex5pL containing 26 amino acid residues of a Pex5pL-specific 37-amino-acid insertion encoded by exon 8 (6) can perform all of Pex5pL functions except for PTS1 binding, i.e., PTS2 transport activity involving the interaction with Pex7p, Pex13p, and Pex14p. These findings imply that Pex5pL(1-243) can transport PTS2-Pex7p complexes to the initial docking site and translocate PTS2 into the matrix as efficiently as full-length Pex5pL (26, 29). However, Pex5pL(118-243) was not competent in PTS2 import, despite being as potent as Pex5pL(1-243) in binding to Pex7p, Pex13p, and Pex14p, thereby indicating that the N-terminal part of residues 1 to 117 plays an essential role in protein import to peroxisomes, including Pex5p targeting to peroxisomes. It is also possible that this region interacts with factors other than Pex14p and Pex13p and/or is responsible for its homomeric interaction (24, 26, 29, 36). In spite of several types of experiments conducted, we were unsuccessful in addressing the issue regarding the function of the sequence comprising residues 1 to 117. The physiological role of this region remains to be defined.

Pex5pL(190-233), containing the N-terminal 18 residues of the 37-amino-acid sequence inserted between positions 215 and 216, interacted with Pex7p in vitro. It is noteworthy that the ClPex5pL(190-233) contains amino acid residues K²¹⁰LXXSXFLXFVXXIXXGXVXL²³⁰ (underline, specific for Pex5pL), which are highly conserved between mammalian Pex5pL and S. cerevisiae Pex18p (residues 216 to 246) as well as Pex21p (residues 230 to 250) (33); both Pex18p and Pex21p are required for efficient translocation of Pex7p-PTS2 cargoes to peroxisomes. However, it is less likely that this Pex5pL sequence of residues 210 to 230 is sufficient for binding to Pex7p, as inferred from the result using Pex5pL(206-262), which is incapable of binding to Pex7p. It is notable that the conserved Ser²¹⁴ mutation to Phe abolished the interaction of Pex5pL with Pex7p, producing pex5 mutant ZPG231, solely defective in PTS2 import (26). Hence, S. cerevisiae Pex18p and Pex21p are likely to be functional orthologues of mammalian Pex5pL, as predicted (29).

Tandemly repeated, characteristic pentapeptide WXXXF/Y motifs have been identified in the N-terminal part of Pex5p of organisms from yeast to humans; there are six and seven such motifs in ClPex5pS and Pex5pL, respectively. It was suggested that the WXXXF/Y motifs in human Pex5p are involved in interaction with Pex14p (36). Various deletion constructs encompassing the N-terminal part of ClPex5pL and containing the WXXXF/Y motifs interacted with Pex14p, suggesting that the motifs are required for binding to Pex14p. Pex5pL variants containing either one or two tandem motifs with mutation of highly conserved amino acids W and F/Y to A were all eliminated in the interaction with Pex14p. Furthermore, the WXXXF/Y motifs interacted with Pex14p with apparently different affinities; motifs 1 and 5 bound most strongly. Pentapeptide motifs 1, 3, and 5 were predicted to form α-helices, while the others do not. This suggests that the amphiphilic structure creates a deep cleft using bulky side chains of aromatic amino acids W and F/Y, which in turn is most likely to be responsible for the interaction with Pex14p. Taking these findings together with a systematic assessment in vivo of the in vitro findings with respect to the consequence of the WXXXF/Y motif for the interaction with Pex14p and Pex13p, by means of alanine scanning of the conserved W and F/Y in the motifs of Pex5pL(1-243), we conclude that the pentapeptide motifs form multiple Pex14p-binding sites, in an mutually independent manner.

In yeast, Pex5p binds not only Pex14p but also the SH3 domain of Pex13p, implying that both serve as a docking receptor for Pex5p (3, 5, 7, 13, 14, 21, 22). On the other hand, the interaction between Pex5p and Pex13p in mammalian cells was not established. It is noteworthy that mammalian Pex5p possesses SH3-binding PXXP motif-like sequence PWPP at positions 49 to 52, while a typical PXXP motif is absent from yeast Pex5p. However, PWPP is less likely to be involved in the interaction with Pex13p, because the N-terminal part of ClPex5p, as in Pex5pL(1-139), containing this PWPP sequence did not bind to Pex13p (Fig. 1). Thus, a key issue is how Pex5p interacts with Pex13p. We demonstrated in this study that ClPex5p interacts with the N-terminal region of HsPex13p, but not with the C-terminal SH3 domain. In the yeast S. cerevisiae, it has recently been shown that Pex13p-SH3 binds to a site distinct from a typical PXXP motif (4). Therefore, it is more likely that Pex5p interacts with Pex13p differently in mammalian and yeast systems. We also showed that mammalian Pex14p interacts with the SH3 region of Pex13p as it does in yeast, in line with the finding from an overlay assay (15). Several PXXP motifs in Pex14p (38) are likely to be responsible for the interaction, as reported for yeast (21).

The activity of Pex5pL(1-243) for complementing the impaired PTS2 import in pex5 ZP105 cells decreased as the Pex14p-binding potency of the WXXXF/Y motif mutants decreased. Loss of binding to Pex14p completely impaired PTS2 import activity, thereby demonstrating that Pex5pL-Pex14p interaction is essential for PTS2 import, in good agreement with our earlier conclusion that Pex5pL mediates translocation of PTS2-Pex7p complexes to peroxisomes by interacting with Pex14p (26, 29). However, Pex5pL-mediated PTS2 transport does not require all of the Pex14p-binding sites to be occupied. It is plausible that the rather precise targeting of the PTS receptor-cargo complexes to peroxisomes is facilitated by the use of multiple sites. This is also inferred from the findings that all types of mutations in Pex5p from CHO cell mutants (31) and human CG2 patients (12, 46) belong to a group of mutations in the TPR, not to those that affect binding to Pex14p (29). In a series of Pex5pL(1-243) variants with mutations at the pentapeptide motifs involved in binding to Pex13p, it is of interest that Mut234 retained PTS2 transport activity despite the loss of binding to Pex13p. We deduced from this finding that Pex5p-Pex13p interaction is not a prerequisite for PTS2 import and that Pex5pL transports Pex7p-PTS2 complexes to docking site Pex14p. Following this step PTS2 import is accomplished by Pex7p and presumably includes a release of the cargo PTS2 protein.

Mammalian catalase binds to the TPR region of Pex5p, while deletion of the C-terminal KANL inhibited the binding to Pex5p, implying that Pex5p interacts with catalase at the C-terminal part including KANL. A full-length Pex5pL mutant defective only in Pex13p binding showed in vivo that catalase import was completely abolished despite normal import of PTS1 and PTS2 proteins (Fig. 7). Several possibilities can be deduced from this observation: (i) Pex5p can bypass the binding to Pex13p during PTS1 and PTS2 import, (ii) Pex5p-Pex13p interaction is required after import of cargoes, and (iii) Pex5p-catalase complexes dissociate upon interacting with Pex13p. A similar notion was reported for S. cerevisiae: Pex5p with mutation F208L, which affected the interaction with Pex13p, was potent in import of PTS1 SKL proteins, such as 3-hydroxyacyl-CoA dehydrogenase and malate dehydrogenase 3, but defective in transport of catalase A with C-terminal PTS tripeptide SKF (5). Moreover, catalase A tagged with SKL in place of SKF was not imported to peroxisomes by Pex5p-F208L. These findings, along with ours, imply that import of cargo proteins such as catalase is regulated not only by a C-terminal PTS signal but also by other transport information internally residing in a distinct part. Catalase more likely carries additional PTS-like information besides the C-terminal KANL motif.

It has been suggested that Pex14p and Pex13p are convergent members of the putative import machinery (15, 23, 29, 38). However, the underlying molecular mechanisms of matrix protein import directed by such machinery remain to be delineated. In the present study, we demonstrated several novel findings. (i) Pex14p interacts with the SH3 domain of Pex13p and dissociates from Pex13p in the presence of Pex5p. (ii) In contrast, the Pex14p-Pex13p complex remains intact in the presence of PTS1 cargo-loaded Pex5p, in a hetero-oligomeric form of Pex13p-Pex14p-Pex5p-PTS1 complexes. (iii) Pex14p can bind to cargo-loaded Pex5p, while Pex13p binds only to unloaded Pex5p, consistent with the notion that Pex14p most likely functions as the initial docking site of Pex5p (29). We herein propose a working model for peroxisomal import of matrix proteins (Fig. 10). In step 1, Pex5p transports cargo proteins including PTS1 and PTS2 to import machinery comprising at least Pex14p-Pex13p complexes; in step 2, the unloading of cargoes from Pex5p initiates dissociation of Pex14p and Pex13p and then Pex5p binds to Pex13p and starts shuttling back to the cytoplasm, possibly through a potential translocation complex consisting of RING peroxins (8, 28) Pex12p, Pex10p, and Pex2p. Note that Pex13p may function as a docking receptor for catalase-Pex5p complexes. However, we were unsuccessful in detecting the Pex13p-Pex5p-catalase complexes in vitro. Using semi-intact human fibroblasts Dammai and Subramani (10) have recently shown that human Pex5p, in part or as a whole, shuttles between the peroxisomal matrix and the cytosol. Pex5p and cargoes are likely to translocate into the matrix as complexes rather than in an independent manner, consistent with our observation (26, 29). Interaction of Pex5p with the cytoplasmically faced, N-terminal part of Pex13p may play a role in transport of Pex5p back to the cytosol. It is also noteworthy that Pex14p and Pex13p possibly form mutually distinct subcomplexes during protein import (T. Harano and Y. Fujiki, unpublished observation) (34). Pex14p can be a convergent component of a subcomplex in the initial step of matrix protein import, while Pex13p plays a major role in a subcomplex for translocation of Pex5p destined for recycling back to the cytosol. These two types of subcomplexes may associate and/or dissociate in a temporal and spatial manner. The underlying molecular mechanisms of such multiple steps involved in protein import may be defined, partly if not completely, by determining the three-dimensional structures of the respective peroxins as well as of a whole complex.

FIG. 10.

A model for peroxisomal protein import mediated by a shuttle translocator Pex5p and import machinery comprising the initial Pex5p-anchoring site, Pex14p, and Pex13p. Pex5p-PTS1 complexes formed in the cytosol traverse to the potential import machinery and dock onto Pex14p interacting with Pex13p (step 1). PTS1 proteins are released at the inner surface and/or inside of peroxisomes (step 2), and then Pex14p and Pex13p dissociate. PTS1-unloaded Pex5p translocates to Pex13p (step 3) and then shuttles back to the cytosol, possibly through RING peroxins Pex12p, Pex10p, and Pex2p (not shown) (step 4).

Peroxisome biogenesis comprises two major distinct steps: the membrane assembly step is initiated by at least three peroxins, Pex3p, Pex16p, and Pex19p, and the following step involves matrix protein import and the growth and division of the assembled organelles (17, 23, 41, 42). Expression of Pex5pL(1-243) rescued only PTS2 import in pex5 ZP105 cells deficient in PTS1 and PTS2 import, resulting in PTS2-positive peroxisomes which were larger in size but smaller in number than normal cells (Fig. 2). In contrast, such larger peroxisomes were not discernible by expression of Pex5pL(1-243) in pex5 mutant ZPG231, solely defective in PTS2 import, thereby strongly implying that formation of normal peroxisomes, including maturation processes presumably following the division of peroxisomes, requires import of PTS1 proteins. Coexpression of Pex5pL(1-243) with Pex5pS that is competent only in PTS1 transport (29) indeed gives rise to assembly of peroxisomes with normal size. Two interpretations can be made for these observations. First, the PTS1 protein is one of the factors involved in the maturation step (or steps) of peroxisomes; second, the maturation and division of peroxisomes are regulated by one or more of the factors that sense the amount of PTS1 proteins in the matrix. Pex11p may be a potential candidate for such a factor (1, 2, 37).