The Peroxin Pex14p

Rat cDNA encoding a 376-amino acid peroxin was isolated by functional complementation of a peroxisome-deficient Chinese hamster ovary cell mutant, ZP110, of complementation group 14 (CG14). The primary sequence showed 28 and 24% amino acid identity with the yeast Pex14p from Hansenula polymorpha andSaccharomyces cerevisiae, respectively; therefore, we termed this cDNA rat PEX14 (RnPEX14). Human and Chinese hamster Pex14p showed 96 and 94% identity to rat Pex14p, except that both Pex14p comprised 377 amino acids. Pex14p was characterized as an integral membrane protein of peroxisomes, exposing its N- and C-terminal parts to the cytosol. Pex14p interacts with both Pex5p and Pex7p, the receptors for peroxisome targeting signal type 1 (PTS1) and PTS2, respectively, together with the receptors’ cargoes, PTS1 and PTS2 proteins. Mutation in PEX14 from ZP161, the same CG as ZP110, was determined by reverse transcription-PCR as follows. A 133-base pair deletion at nucleotide residues 37–169 in one allele created a termination codon at 40–42; in addition to this mutation, 103 base pairs were deleted at positions 385–487, resulting in the second termination immediately downstream the second deletion site in the other allele. Neither of these two mutant forms of Pex14p restored peroxisome biogenesis in ZP110 and ZP161, thereby demonstrating PEX14 to be responsible for peroxisome deficiency in CG14.

ture before pouring onto the cells. After incubation for 1 h at 37°C, the DNA-liposome mixture was removed by aspiration, and fresh F-12 medium with fetal calf serum was added. The cells were cultured for 2 days and further incubated overnight in 2 ml of serum-free F-12 medium. Among cDNA pools examined, a positive one (A9) that restored peroxisomes in ZP110 was further divided into subpools and screened. A cDNA fragment (SalI-NotI fragment) of an isolated clone pBK-CMV⅐PAF-4, was blunted and subcloned into the SmaI site of pBluescript II KS(Ϫ) (Stratagene). Nucleotide sequence of both strands was determined by the dideoxy chain termination method using Dye-terminator DNA sequence kit (Applied Biosystems, Foster City, CA). Alignment was done using a CLUSTAL W program (EMBL, Heidelberg, Germany).
Screening of Human and Chinese Hamster cDNA Library-A 32 Plabeled, 0.65-kilobase pair BamHI fragment of open reading frame from rat (Rattus norvegicus; Rn) PEX14, RnPEX14, was used as a probe to screen 2.4 ϫ 10 5 colonies of human liver cDNA library (16) in pCMVS-PORT I (Life Technologies). Three positive clones were isolated; a longer one was sequenced. About 1.5 ϫ 10 5 independent colonies of a cDNA library from wild-type CHO-K1 cells (14) were also screened with a 32 P-labeled RnPEX14 (0.85-kilobase pair SalI-EcoRI) fragment. Two of 10 positive clones were subcloned into pBluescript II SK(Ϫ) (Stratagene) at SalI and NotI sites.
Transfection of PEX14 -An expression plasmid pUcD2Hyg⅐ RnPEX14 was generated by inserting a SalI-ApaI fragment (nucleotide residues at Ϫ16 to 1,315, taking A of the initiation codon as 1) of the original RnPEX14 cDNA clone, pBK-CMV⅐PAF-4, to the pUcD2SR␣MCSHyg vector (20,21). CHO cell mutant ZP110 was transfected with pUcD2Hyg⅐RnPEX14, followed by selection in the presence of 200 g/ml hygromycin B (Sigma). The hygromycin B-resistant colonies formed on the coverslips were examined for complementation of peroxisomes, 4 days after transfection. Peroxisome-restored colonies were counted. Seven other groups of CHO cell mutants, Z24, Z65, ZP92, ZP105, ZP109, ZP114, and ZP119, and fibroblasts derived from peroxisome-deficient patients were similarly transfected with pUcD2Hyg⅐RnPEX14. A stable cell clone of the RnPEX14 transformant of ZP110 was isolated as follows. Three of five hygromycin B-resistant transformant clones were peroxisome-positive, on the basis of immunostaining; one of the three, named 110P14, was cloned by the limiting dilution method. Transfection with human and Chinese hamster PEX14 (HsPEX14 and ClPEX14, respectively) was likewise done, as with RnPEX14.
Morphological Analysis-Peroxisomes in CHO cells and human fibroblasts were visualized by indirect immunofluorescence light microscopy, under a Carl Zeiss Axioskop FL microscope (Oberkochen, Germany) using a number 17 filter, as described by Tateishi et al. (15). Antibodies used were rabbit antibodies to rat liver catalase (11), human catalase (4), PTS1 peptide (14), 70-kDa peroxisomal integral membrane protein (PMP70) (11). Anti-Pex14p antibody was raised in rabbits by immunizing with synthetic peptide comprising the C-terminal, 19-amino acid sequence of rat Pex14p (see Fig. 2, dashed underline) supplemented with cysteine at the N terminus that had been linked to keyhole limpet hemocyanin (17). Antigen-antibody complex was detected using fluorescein isothiocyanate-labeled sheep anti-rabbit immunoglobulin G antibody (Cappel, Durham, NC), under a Carl Zeiss Axioskop FL microscope.
Expression of Epitope-tagged Pex14p-An epitope, myc, tagging to the C terminus of Pex14p was done with a PCR-based technique, using a forward primer (5Ј-CTGGATCCTGGAGTC-3Ј) and a reverse primer (5Ј-CCGGGTACCTCACAAGTCTTCTTCAGAAATCAGCTTTTGTTCG-TCTCGCTCACTCTC-3Ј) containing the myc epitope (underline) and a stop codon. myc-tagged RnPEX14 was cloned into pBluescript II KS(Ϫ) (pBS⅐RnPEX14-myc) by replacing the NheI-KpnI fragment of pBS⅐RnPEX14 with NheI-KpnI fragment of PCR-amplified products using the primers described above. Plasmid pUcD2Hyg⅐RnPEX14-myc was constructed by inserting the XbaI-KpnI fragment of pBS⅐RnPEX14-myc into pUcD2SR␣MCSHyg vector (21). Flag-tagging to the N terminus of RnPEX14 protein was likewise done by PCR using a forward primer, 5Ј-GCCCAGATCTTGGCGTCCTCGGAGC, and a reverse primer, 5Ј-TCAGGGCTGTGCCCGTCCTT-3Ј. Flag-tagged Rn-PEX14 in pUcD2Hyg⅐SR␣ vector (pUcD2Hyg⅐flag-RnPEX14) was constructed as follows. The BamHI fragment of pUcD2Hyg⅐RnPEX12 (21) was replaced with a BglII-BamHI fragment of the PCR products; the plasmid was then digested with ApaI and ligated with the ApaI fragment of pBS⅐RnPEX14. All plasmid constructs were assessed by nucleotide sequence analysis. Pex14p-myc and Flag-Pex14p were detected using mouse monoclonal antibodies to human c-myc (9E10) (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) and Flag (M2) (Scientific Imaging Systems, New Haven, CT), and Texas Red-labeled sheep anti-mouse IgG second antibody (Amersham Pharmacia Biotech, Tokyo, Japan), in cells that had been permeabilized with either 25 g/ml of digitonin or 1% Triton X-100.
Subcellular Fractionation-Liver was excised from a 7-week-old male Wistar rat and homogenized, as described by Miura et al. (24), except that 1 M each of pepstatin and leupeptin was added to a homogenizing buffer. A postnuclear supernatant fraction, prepared by centrifugation of the liver homogenates at 750 ϫ g for 10 min, was then centrifuged at 2,500 ϫ g for 10 min, 25,000 ϫ g for 10 min, and 100,000 ϫ g for 1 h to separate heavy and light mitochondria, microsomes, and cytosol, respectively (25,26). The light mitochondrial fraction (12 mg of protein) was further subjected to sucrose density gradient ultracentrifugation (24). Protein and marker enzymes, catalase for peroxisomes, glutamate dehydrogenase for mitochondria, esterase for microsomes, and N-acetyl-␤-glucosaminidase for lysosomes were assayed, essentially as described (17). Protein in each fraction was separated by SDS-PAGE, and electrophoretically transferred onto polyvinylidene difluoride membrane (Bio-Rad). Pex14p and marker proteins were probed with respective antibodies and then visualized with ECL Western blotting detection reagent (Amersham Pharmacia Biotech). For determination of intraperoxisomal localization, peroxisomal fraction was diluted with 20 mM Hepes-KOH, pH 7.6, and sonicated four cycles for 30 s each in the presence of 1 M NaCl (17). Membrane and soluble fractions were separated by centrifugation for 30 min at 100,000 ϫ g. Sodium carbonate treatment (26) and Triton X-114 extraction (27) were performed as described.
Expression of Fusion Protein-Glutathione S-transferase (GST) fusion protein with rat Pex14p was constructed in Escherichia coli expression vector pGEX4T-1 (Amersham Pharmacia Biotech) as follows. The 3Ј-half of the PvuII fragment of pBS-RnPEX14 was inserted into the SmaI site of pGEX4T-1. This vector expresses the C-terminal region of Pex14p, termed GST-C. The 5Ј-half of RnPEX14 was amplified using PCR primers (forward, 5Ј-GCCAGATCTGCGTCCTCGGAGCAG-3Ј; reverse, 5Ј-TCTTCATCCACGTGGCTCAC-3Ј). GST-Pex14p-expression plasmid was constructed by replacing the BamHI fragment of GST-C with the BglII-BamHI fragment of the amplified 5Ј-half products of RnPEX14. GST fusion protein with a shorter isoform of Pex5p (Pex5pS) and a longer, 37-amino acid-inserted isoform Pex5p (Pex5pL) of Chinese hamster (14) was likewise expressed in E. coli. 2 His 6 -tagged human Pex7p-expression vector was constructed as follows. Blunted SmaI-SalI fragment of human PEX7 in pTZ18R (28) 3 was ligated into the blunted SalI site of E. coli expression vector pQE30, containing a histidine tag sequence (QIAGEN, Hilden, Germany).
Purification of Fusion Protein-Five hundred l each of 16-h culture of E. coli, separately expressing GST and GST fusion protein, were diluted with 10 volumes of YT medium at 2-fold concentration. After 1 h of culture at 37°C, isopropyl-␤-D-thiogalactoside was added to the final concentration of 1 mM. After 2 h at 37°C, the cells were harvested, resuspended in 1 ml of chilled PBS containing 1 mM phenylmethylsulfonyl fluoride, 1% Triton X-100, and then sonicated. Lysates were centrifuged, and supernatant was collected. The supernatant was incubated with 100 l of 50% slurry glutathione-Sepharose (Amersham Pharmacia Biotech) at 4°C for 1 h. Glutathione-Sepharose beads were washed three times and resuspended in 250 l of binding assay buffer: 50 mM HEPES-KOH, pH 7.4, 150 mM NaCl, 0.2% Nonidet P-40, 10% glycerol, 0.25% BSA, 25 g/ml each of leupeptin and antipain, 50 units/ml aprotinin, 1 mM phenylmethylsulfonyl fluoride, 1 mM EDTA, and 1 mM dithiothreitol. GST-Pex5pS and GST-Pex5pL were separately mixed with Glutathione-Sepharose beads at 4°C. After a thorough wash, the protein-bound Sepharose beads were suspended in 10 volumes of a cleavage buffer, 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM dithiothreitol, 0.1% Nonidet P-40. GST-Pex5p was cleaved with PreScission Protease (Amersham Pharmacia Biotech) at 4°C for 5 h, and Pex5p was isolated as an eluate according to the procedure recommended by a manufacturer.
coli cells were sonicated, and inclusion bodies were isolated by centrifugation. After solubilized with 8 M urea, His-Pex7p was purified by affinity chromatography using a nickel-nitrilotriacetic acid-agarose column (QIAGEN), where elution was done with 0.5 M imidazole, 8 M urea, according to the procedure recommended by the manufacturer. His-Pex7p was used at a 10 -20-fold dilution.
In Vitro Binding Assay-Binding reaction mixture contained protein components to be examined, including 20 g of GST or GST-Pex14p, His-Pex7p (5 g), and 35 S-labeled 3-ketoacyl-CoA thiolase (29) (5 l) in the binding assay buffer (400 -500 l) described above. The reaction was done overnight at 4°C. A binding reaction was also done, using GST-Pex14p (3 g), each of the two isoforms of Pex5p (3 g) (i.e. Pex5pS or Pex5pL), 35 S-labeled rat acyl-CoA oxidase (AOx) (30) (5 l), GST-Pex14p (20 g), and 35 S-Pex14p (5 l). Proteins were recovered using glutathione-Sepharose, washed four times with the binding assay buffer minus BSA, and analyzed by SDS-PAGE using 12% or 4 -20% polyacrylamide gel. Radioactive protein bands were detected using a FujiX BAS1000 Bio-Autoimazing Analyzer (Fuji Photo Film, Tokyo, Japan), and His-Pex7p and Pex5p were detected by immunoblot using mouse monoclonal anti-RGS-His tag antibody (QIAGEN) and rabbit anti-Chinese hamster Pex5pS antibody, 4 (30) were performed as described. Immunoprecipitation of proteins using antibodies to rat AOx and 3-ketoacyl-CoA thiolase and a catalase latency assay using digito- . Hyphens represent spaces. Identical amino acids between species, including two or more mammalian ones, are shaded. A putative membrane-spanning segment is underlined; the dashed overline indicates a predicted coiled-coil region, where conserved hydrophobic amino acids are marked by dots. The sequence used for chemical synthesis of Pex14p peptide is shown by a dashed underline. The solid arrowheads indicate the position of mutation in ZP161 (see Fig. 4). The GenBank TM data base accession numbers for rat, human, and Chinese hamster PEX14 genes are AB017544, AB017546, and AB017545, respectively. nin were performed as described (11). For Northern blotting, RNA was isolated from CHO cells and the liver of a rat treated for 7 days with clofibrate and of an untreated rat. The blot was hybridized with the BamHI fragment (nucleotide residues at 232-605) of RnPEX14 labeled with [␣-32 P]dCTP (Amersham Pharmacia Biotech). Western blot analysis was done with primary antibodies, including anti-Pex12p antiserum (20,21), and a second antibody, donkey anti-rabbit IgG antibody conjugated to horseradish peroxidase (Amersham Pharmacia Biotech), using ECL Western blotting detection reagent (Amersham Pharmacia Biotech). 12-(1Ј-pyrene)dodecanoic acid/long wavelength ultraviolet light (P12/UV) and 9-(1Ј-pyrene)nonanol/UV (P9OH/UV) resistance was determined under conditions of 2 M/1.5 min and 6 M/2 min (4), respectively.

Cloning of a Rat PEX14 cDNA
We used a transient expression assay as a cDNA cloning strategy (16, 18, 20, 21) on a CG14 CHO cell mutant, ZP110, defective in peroxisome assembly (15) (Fig. 1a). A cDNA library divided into small pools was transfected to ZP110. Peroxisomerestoring positive cDNA clones were searched for by immunofluorescence staining of catalase, a peroxisomal matrix marker enzyme. One combined pool (A9) yielded several peroxisomerestored cells of ZP110, in a single dish (Fig. 1b, arrows). After a fourth round of screening, one positive clone, named pBK-CMV⅐PAF-4, was isolated, which restored peroxisomal import of catalase in ZP110 (data not shown). The cDNA portion of pBK-CMV⅐PAF-4 was sequenced on both strands, indicating that the cDNA was 1,921 bp in length with an open reading frame encoding a protein consisting of 376 amino acids (Fig. 2). The calculated molecular mass of its deduced amino acid sequence was 40,936 Da. The amino acid sequence showed 28 and 24% identity with those of Pex14p from Hansenula polymorpha (31) and Saccharomyces cerevisiae (32, 33), respectively (Fig. 2). Circles, CHO-K1; triangles, ZP110; squares, 110P14 (a stable rat PEX14 transformant of ZP110); diamonds, lactate dehydrogenase in ZP110. Relative free enzyme activity is expressed as a percentage of the total activity measured in the presence of 1% Triton X-100 (11). The results represent a mean of duplicate assays. B, biogenesis of peroxisomal proteins. Cells (1 ϫ 10 5 ) were labeled for 24 h with [ 35 S]methionine and [ 35 S]cysteine. Cell types are indicated at the top. Immunoprecipitation was performed with rabbit anti-rat AOx and 3-ketoacyl-CoA thiolase (TH) antibodies. Immunoprecipitates were analyzed by SDS-PAGE (12% polyacrylamide gel). Radioactive polypeptide bands were detected with Hyperfilm (Amersham Pharmacia Biotech), using a FujiX BAS1000 Bio-Imaging Analyzer (exposure, 20 days). The open arrowheads show the positions of AOx components A, B, and C; protein bands are indicated by dots; one with a slightly greater mobility than AOx-B and one nearly indistinguishable from AOx-C are nonspecific (14). The stippled and solid arrowheads indicate a larger precursor (P) and mature protein (M) of 3-ketoacyl-CoA thiolase, respectively. C, cell lysates (ϳ1.8 ϫ 10 5 cells) were subjected to SDS-PAGE and transferred to polyvinylidene difluoride membrane. Cell types are indicated at the top. Immunoblot analysis was done using anti-AOx and anti-thiolase (TH) antibodies. AOx components and a thiolase precursor (P) and mature protein (M) are as in B. Protein bands larger than AOx-A are nonspecific (16).
Thus, we termed this cDNA rat PEX14, RnPEX14. RnPEX14 complemented peroxisomal import of catalase in ZP110 (Fig.  1c). Human (HsPEX14) and Chinese hamster (ClPEX14) cDNA were also cloned by colony hybridization from human and Chinese hamster cDNA libraries, using RnPEX14 as probe. Both HsPEX14 and ClPEX14 encoded 377-amino acid Pex14p, with 96 and 94% identity to rat Pex14p at a deduced amino acid sequence level, while rat Pex14p was shorter by one amino acid, at alignment position 329 (Fig. 2). Pex14p apparently contained at least one hydrophobic segment and a coiled-coil region (Fig. 2, underline and broken overline).

PEX14 Restored Peroxisome Biogenesis in ZP110
Several phenotypic abnormalities due to peroxisome deficiency, such as impaired import of PTS1 and PTS2 proteins, were found in ZP110 (15). To determine whether RnPEX14 could correct these phenotypes, a stable RnPEX14 transformant of ZP110, named 110P14, was isolated. PTS1 proteins were noted in numerous vesicular structures, presumably peroxisomes, when stained with antibodies to PTS1 (Fig. 1d), catalase and 3-ketoacyl-CoA thiolase (data not shown). Numerous PMP70-positive particles were detected in 110P14 cells (Fig. 1f), whereas peroxisomal remnants, larger PMP70-positive vesicles but fewer in number, were discernible in ZP110 (15) (Fig. 1e). These results strongly suggested that 110P14 cells had morphologically normal peroxisomes, as seen in the wild-type CHO-K1 cells. When other CHO cell mutants, ZP111 (15) and ZP161 (7), of the same CG as ZP110, were transfected with human PEX14 (HsPEX14), PTS1 proteins were likewise localized in peroxisomes, demonstrating HsPEX14 to be functional in CHO cells (see Table II).
In peroxisome-deficient cells, peroxisomal proteins are mislocalized to the cytosol, rapidly degraded, or not converted to mature forms, despite normal synthesis (4,11,13). In the digitonin titration assay, nearly 60% of the catalase activity was latent at the digitonin concentration of 100 g/ml in the wild-type cells (Fig. 3A). In ZP110 cells, full activity of catalase was detected at 100 g/ml digitonin, with the same latency profile as lactate dehydrogenase, a cytosolic enzyme; hence, catalase was present in the cytosol (Fig. 3A). This was consistent with our earlier observation (4,11,16,21). In 110P14 cells, catalase showed almost the same latency as in the wild-type CHO-K1 cells, demonstrating restoration of peroxisome biogenesis.
AOx, the first enzyme of the peroxisomal fatty acid ␤-oxidation system, is synthesized as a 75-kDa polypeptide (A component) and is proteolytically converted into 53-and 22-kDa polypeptides (B and C components, respectively) in peroxisomes (11,30,34). All three polypeptide components were evident in CHO-K1 cells, upon continuous cell labeling with [ 35 S]methionine and [ 35 S]cysteine followed by immunoprecipitation (Fig. 3B, lane 1), but only the A component was seen in the ZP110, in a much smaller amount, probably due to rapid degradation (11) (Fig. 3B, lane 2). The three components of AOx were found in 110P14, as in CHO-K1 cells (Fig. 3B, lane 3), indicative of proper import and proteolytic conversion of AOx. Peroxisomal 3-ketoacyl-CoA thiolase, the third enzyme of the peroxisomal ␤-oxidation system, is synthesized as a larger precursor with an amino-terminal presequence that contains PTS2 (35,36) and is converted to a mature form in peroxisomes (11,29,37,38). In wild-type CHO-K1 cells, only the matured thiolase was detected (Fig. 3B, lane 4), thereby reflecting rapid processing of the precursor form. In ZP110 cells, only the larger precursor was found (lane 5), implying the absence of processing activity. 110P14 cells showed only the mature form of thiolase, demonstrating the complementation of PTS2 protein import and processing (Fig. 3B, lane 6). Proper processing of AOx and thiolase was also evident, as assessed by immunoblotting, in 161P14, i.e. ZP161 cells stably transfected with Rn-PEX14, confirming the restored biogenesis of peroxisomes (Fig.  3C). These results demonstrated that RnPEX14 restored peroxisome biogenesis in ZP110 and ZP161.
Treatment with P12/UV specifically kills peroxisome-deficient mutants, due to almost no synthesis of plasmalogen (4,39), whereas P9OH/UV selection kills wild-type cells, which incorporate this fatty alcohol analogue into plasmalogen molecules (4,40). 110P14 cells showed the resistance to P12/UV treatment and P9OH/UV-sensitive phenotype, as was the case for CHO-K1 cells (Table I). Conversely, ZP110 was sensitive to the P12/UV but resistant to the P9OH/UV, consistent with our earlier observation (15). Taken together, RnPEX14 could fully complement the abnormality of CG14 ZP110.

Expression of PEX14
Peroxisomes and peroxisomal proteins are induced in rat liver by administration of hypolipidemic agents, such as clofibrate (41,42). RnPEX14 mRNA was not induced by clofibrate, as assessed by Northern blotting (Fig. 4A, lanes 1 and 2, upper  panel), under which AOx mRNA was induced by about 50-fold (middle panel), findings consistent with previous observations (21,42,43). PEX14 mRNA was detected as a single band of ϳ2 kilobase pairs in CHO-K1 and ZP161 (Fig. 4A, lanes 3 and 4).
The amount and size of PEX14 mRNA were indistinguishable between the mutant and wild-type CHO cells, suggesting that alteration in the PEX14 gene in ZP161 was not at a transcriptional level. PEX14 mRNA in CHO cells was similar in size but at much lower levels, as compared with those in rat liver ( Fig.  4A; note that poly(A) ϩ RNA was loaded on lanes [3][4][5]. It is noteworthy that PEX14 mRNA in ZP110 was nearly undetectable, implying alteration at a transcriptional level or rather rapid degradation of PEX14 mRNA (Fig. 4A, lane 5). The undetectable level of PEX14 mRNA in ZP110 was confirmed by reverse transcription-PCR, using up to 2-5 g of poly(A) ϩ RNA (data not shown).

Dysfunction of Pex14p in CHO Mutants
PEX14 in Mutants-To investigate dysfunction of Pex14p in ZP161, we determined the nucleotide sequence of Pex14p cDNA isolated from ZP161 by reverse transcription-PCR. In 12 independent cDNA clones isolated, six clones showed a 133-bp deletion from nucleotide C at position 37 of a codon for Pro 13 to G at position 169 of a codon for Gly 57 , termed ClPEX14del133, resulting in a frameshift that created a mutation at Pro 13 to Gly immediately followed by a termination codon at amino acid alignment position 14 (Fig. 4B, ZP161⌬1, upper panel). In six remaining cDNA clones, a 103-bp deletion from nucleotide A at position 385 of a codon for Lys 129 to G at position 487 of a codon for Val 163 was likewise identified, termed ClPEX14del133/103, in addition to the 133-bp deletion described above, thereby resulting in another frameshift with a termination codon at position 129 in the deduced sequence (Fig. 4B, ZP161⌬2, lower  panel). Thus, dysfunction of the Pex14p caused by a deletion is most likely to be the primary defect in the mutant ZP161. No reverse transcription-PCR product was obtained from ZP110derived mRNA, consistent with the finding by Northern analysis (see Fig. 4A) (data not shown).
Complementation of Protein Transport by ClPEX14 -When mutant ZP161 cells were transfected with ClPEX14 cDNA from the wild-type CHO-K1, PTS1 proteins were found in punctate staining, thereby indicating complementation of peroxisomal protein import (Fig. 4C, a), as was the case by HsPEX14 (data not shown). Transfection of a mock vector did not alter intracellular location of PTS1 proteins in ZP161 (data not shown). To assess the impaired function of Pex14p in ZP161, two types of ZP161-derived PEX14 cDNAs, ClPEX14del133 with a 133-bp deletion and ClPEX14del133/103 with two-site, 133-and 103-bp deletions, were separately transfected back to the mutant cells. PTS1 proteins were present in the cytosol, in a diffuse manner, in both ZP110 transfected with ClPEX14del133 and ClPEX14del133/103-transfected ZP161 (Fig. 4C, b and c,  respectively), hence demonstrating dysfunction of the mutated forms of Pex14p. Similar results were likewise obtained for ClPEX14del133-transfected ZP161 and ClPEX14del133/103transfected ZP110 cells (data not shown). Import of catalase was not restored in these transfectants as assessed by cell staining with anti-catalase antibody (data not shown). Moreover, ZP111, another mutant of CG14 (15), showed cytosolic staining of PTS1 proteins in transfectants with ZP161-derived PEX14 (data not shown), confirming the impaired function of ClPEX14del133 and ClPEX14del133/103.
Collectively, we conclude that dysfunction of Pex14p caused by truncation is the primary defect in impaired peroxisome biogenesis in CG14 CHO mutant ZP161.

Intracellular Localization of Pex14p
The C-terminal peptide of rat Pex14p (residues 357-376) was used to raise rabbit antibody. This antibody reacted with a protein of an apparent mass of ϳ56 kDa, larger than the predicted size, in immunoblot of rat liver homogenates (Fig. 5A,  lane 4). The mobility of Pex14p in SDS-PAGE was indistinguishable from that of Pex14p synthesized in vitro by coupled transcription/translation of RnPEX14 that was specifically immunoprecipitated by anti-Pex14p antibody but not by preimmune serum (lanes 1-3), thereby indicating that a cloned Rn-PEX14 encodes bona fide Pex14p. This result implies the synthesis of Pex14p at its final size, consistent with a general rule for peroxisomal proteins (8) (see below). The reason for the reduced mobility of Pex14p in SDS-PAGE is presently unknown.
Intracellular localization of Pex14p was investigated by subcellular fractionation of rat liver. Pex14p was detected in a postnuclear supernatant fraction and then exclusively present in the organellar fraction, not in cytosol (Fig. 5A, lanes 5-7). Pex14p was apparently enriched in isolated peroxisomes (lane 8). Upon further fractionation of the light mitochondrial fraction by sucrose density gradient centrifugation, Pex14p was detected as a single band and co-sedimented with a peroxisomal marker enzyme, catalase, and AOx as well as peroxisomal integral membrane proteins, such as Pex12p, although less in amount (20,21) and PMP70 (data not shown), thereby indicating that Pex14p is a peroxisomal protein (Fig. 5B). This was consistent with morphological observation (see below). The apparently slight difference in the distribution of Pex12p, as compared with that of Pex14p and AOx, is presently unknown. The distribution of Pex14p on the gradient was distinct from those of the marker enzymes, glutamate dehydrogenase for mitochondria, esterase for microsomes, and N-acetyl-␤-glucosaminidase for lysosomes, confirming the peroxisomal location of Pex14p.
The subcellular localization of Pex14p was also determined by immunofluorescent microscopy using Pex14p tagged at its N terminus with an epitope Flag. In the wild-type CHO-K1 cells transfected with Flag-RnPEX14, Pex14p was detected in a punctate staining pattern with anti-Flag antibody (Fig. 6A, a). The pattern was superimposable on that obtained using anti-PTS1 antibody, thereby demonstrating that Flag-Pex14p was targeted to peroxisomes (Fig. 6A, b and c). Similar results were obtained when RnPEX14 was expressed in CHO-K1 cells and stained using anti-Pex14p peptide antibody (data not shown). The flag-Pex14p fully restored peroxisome assembly in ZP110, as efficiently as Pex14p and was colocalized with PTS1 protein, indicating that the N-terminal tagging did not interfere with Pex14p function (Fig. 6A, d and e). These results were interpreted to mean that Flag-Pex14p and Pex14p were translocated to peroxisomes. Flag-Pex14p was colocalized with PMP70 in peroxisomal ghosts in other complementation groups of CHO cell mutants, such as PEX2-defective Z65 (17) (Fig. 6A, f and g), as was the case for Pex12p (21). Thus, translocation of Pex14p does not appear to be impaired in these mutant cells. It is noteworthy that endogenous Pex14p was also detectable by anti-Pex14p antibody, in CHO-K1 and PEX6-deficient ZP92 (4, 18) (Fig. 6A, h and i). Collectively, the data demonstrate peroxisomal localization of Pex14p.
Hydropathy analysis of Pex14p suggested that Pex14p contains at least one hydrophobic segment (see Fig. 2, solid line). Pex14p was not extracted with 50 mM Hepes-KOH (pH 7.6), from freshly isolated rat liver peroxisomes (data not shown). The integrity of Pex14p was verified by sonication in 50 mM Hepes-KOH (pH 7.6), 1 M NaCl; treatment with 1% Triton X-114 (27); and extraction with 0.1 M sodium carbonate (pH 11.5) (26, 44) (Fig. 6B). Pex14p was not extractable with either 1 M NaCl or sodium carbonate, as was the case for peroxisomal integral membrane proteins, PMP70 (25) and Pex12p (20,21), but in contrast to a matrix enzyme, AOx, thereby strongly suggesting that Pex14p is an integral membrane protein (Fig.  6B, lanes 2, 3, 6, and 7). Upon treatment with Triton X-114, Pex14p, but not PMP70 and Pex12p, was recovered more in an aqueous phase than in a detergent phase (lanes 4 and 5), probably due to the presence of a highly charged region at the N-terminal part of Pex14p. A similar observation for highly charged membrane glycoproteins was reported by Bodier (27).

Membrane Topology of Pex14p
Topology of Pex14p in peroxisomal membranes was investigated by the differential cell permeabilization procedure, where detection of Pex14p was performed using anti-epitope tag antibody. When the wild-type CHO-K1 cells transfected with RnPEX14-myc encoding C-terminally myc-tagged Pex14p were permeabilized with 25 g/ml of digitonin, under which plasma membranes are selectively permeabilized and intraperoxisomal proteins are inaccessible to exogenous antibodies (20,21). Pex14p-myc was observed in a punctate staining pattern, whereas almost no staining of cells was noted with anti-PTS1 antibody (Fig. 7A, a and b). When the Pex14p-myc-expressing CHO-K1 cells were treated with Triton X-100, which solubilizes all cellular membranes, both Pex14p-myc and PTS1 proteins were detected in particulates similar in number, presumably peroxisomes (Fig. 7A, c and d). The data demonstrating exactly the same topology were also obtained with a set of antibodies to Pex14p C-terminal peptide and catalase (not shown). Similar results were likewise obtained in studies of RnPEX14-myc-transfected ZP110 cells (data not shown). CHO-K1 cells expressing Pex14p-myc were pemeabilized as above and then incubated with Proteinase K. In an immunoblot of mock-treated cells, Pex14p-myc was detected using anti-myc antibody, where AOx was used as a matrix protein marker (Fig.  7B, lane 1). Upon cell permeabilization with digitonin, Pex14pmyc became sensitive to Proteinase K and was no more detectable, while AOx was not accessible to the protease (lane 2). Both Pex14p and AOx were no more discernible after protease treatment in the presence of Triton X-100 (lane 3). The data together strongly suggest that the C-terminal part of Pex14pmyc is exposed to the cytosol.
To determine the membrane topology of N-terminal part of Pex14p, N-terminally Flag-tagged Pex14p was expressed in CHO-K1 cells. Flag-Pex14p was detected in an immunoblot, using antibodies to Flag (Fig. 7B, lanes 1) and Pex14p Cterminal peptide (data not shown). After cell permeabilization using digitonin, Flag-Pex14p became accessible to Proteinase K digestion (Fig. 7B, lane 2), as was the case for Pex14p-myc, thereby indicating the Flag N-terminal part of Pex14p to be facing the cytosol. This conclusion was confirmed by using anti-Pex14p C-terminal peptide antibody (data not shown). It is noteworthy that Flag-RnPEX14-transfected CHO-K1 was not readily stained using anti-Flag antibody (data not shown). It is plausible that the Flag epitope was inaccessible to the antibody, presumably due to steric hindrance by other peroxisomal proteins.
Taken together, the results demonstrate that N-and Cterminal parts of Pex14p are exposed to the cytosol.

Expression and Purification of Fusion Proteins
To investigate whether Pex14p interacts with other peroxins such as a PTS2 receptor Pex7p and Pex5p, a PTS1 receptor, fusion protein of Pex14p with GST, GST-Pex5pS, GST-Pex5pL, and His-tagged human Pex7p were expressed in E. coli. Each protein was purified by affinity chromatography of E. coli lysates using their respective ligand columns. Pex5pS and Pex5pL were isolated from the GST fusion protein. The purity of all proteins was assessed by SDS-PAGE.

Pex14p Interacts with Pex5p and Pex7p
In an in vitro Pex14p-binding assay, 35 S-labeled 3-ketoacyl-CoA thiolase, a PTS2 protein (35,38), was detected at 20% recovery of the input, as determined by quantitation of the autoradiogram, by incubation with GST-Pex14p plus His-Pex7p (Fig. 8A, lane 5), while 35 S-labeled thiolase was found at a nearly background level with His-Pex7p plus GST, GST-Pex14p, or GST (lanes 2-4), thereby demonstrating Pex7pmediated interaction of Pex14p with PTS2 protein to be specific (see also Fig. 8A, lower panel). 35 S-labeled AOx, a PTS1 protein (24,30), was found in an unbound fraction in the assay, thus indicating Pex7p to be specific for PTS2 (data not shown). Moreover, 35 S-labeled Pex7p was recovered by incubation with GST-Pex14p, indicating direct binding of Pex7p to Pex14p (data not shown).
We identified two isoforms of Pex5p, Pex5pS and Pex5pL, that restored peroxisome biogenesis in CG-II CHO cell mutants, ZP105 and ZP139 (14). Pex5pL is involved in peroxiso- mal import of not only PTS1 proteins but also PTS2 proteins, while Pex5pS apparently transports only PTS1 proteins (14,45). Both Pex5pS and Pex5pL bound to GST-Pex14p, as assessed by immunoblot, together with 35 S-AOx in the Pex14pbinding assay (Fig. 8B, lanes 3 and 4). GST-Pex14p showed almost no direct binding to 35 S-AOx (lane 2), while both GST-Pex5pS and GST-Pex5pL bound to 35 S-AOx at a similar level (data not shown), thereby implying that both Pex5p isoforms mediate the interaction of Pex14p with PTS1. Specific Pex5pmediated PTS1 protein binding to Pex14p was envisaged by quantitation of 35 S-AOx, where nearly 40% of the input AOx was recovered in the case of both Pex14p/Pex5pS and Pex14p/ Pex5pL, evidently indicative of a complex comprising these three components (Fig. 8B, lower panel). In contrast, no binding of 35 S-labeled thiolase was observed in this assay, confirming Pex5p to specifically recognize PTS1, not PTS2 (data not shown). Similar results showing interaction of Pex14p with Pex5p/ 35 S-AOx were obtained using His-tagged protein such as His-Pex5p (data not shown). Moreover, 35 S-labeled Pex5pS and Pex5pL bound to GST-Pex14p (data not shown), confirming the direct interaction of Pex14p with Pex5p.
Taken together, Pex14p was evidently demonstrated to interact with Pex7p/PTS2 protein and Pex5p/PTS1 protein, thereby implying Pex14p-mediated import of PTS1 and PTS2 proteins into peroxisomes.

Binding to Pex14p
To determine whether Pex14p interacts with itself, GST-Pex14p was incubated with 35 S-labeled Pex14p. 35 S-Pex14p was recovered with GST-Pex14p at a nearly 4-fold amount as compared with the amount with GST, suggesting that Pex14p proteins bind each other in a homomeric manner (Fig. 9, A and  B). The forms of homomeric interaction, i.e. dimer or oligomer, are presently unknown. DISCUSSION In the present work, we isolated rat PEX14 cDNA encoding 41-kDa integral membrane protein of peroxisomes, by functional complementation of peroxisome-deficient CHO cell mutant, ZP110. CG14 mutants, ZP110, ZP111, and ZP161, were restored in peroxisome assembly, for both morphological and biochemical criteria. None of the fibroblasts from patients with peroxisome deficiency disorders of 11 CGs was complemented, indicating that the PEX14 gene is not the causal gene of any CG of human peroxisome-deficient disorders thus far classified. PEX14 is novel; the eighth gene required for peroxisome assembly in mammals, including humans; and responsible for the defects in peroxisome assembly of CG14. We have shown that a  35 S-labeled thiolase, used for assay was loaded in lane 1. Fusion protein components added in the assay mixture are indicated at the top. After thoroughly washing, proteins bound to glutathione-Sepharose were analyzed by SDS-PAGE. 35 S-labeled thiolase was detected using a FujiX BAS1000 Bio-Imaging Analyzer (Fuji Photo Film) at an exposure for 6 h; His-Pex7p, by immunoblot using anti-His tag antibody. Lower part, 35 S-labeled thiolase of each lane in the autoradiogram was quantitated by a FujiX BAS1000 Bio-Imaging Analyzer and is represented as a percentage of the input. B, binding of Pex14p to Pex5p/PTS1 protein. Upper part, in vitro binding assay was likewise done using two isoforms of Pex5p, Pex5pS and Pex5pL, and 35 S-AOx. A half aliquot of the input, 35 S-AOx, was loaded in lane 1. Protein components used are indicated at the top. 35 S-AOx was detected at an exposure for 14 h; Pex5pS and Pex5pL were detected by immunoblot using anti-Pex5p antibody. Lower part, 35 S-AOx of each lane was quantitated and is represented as a percentage of the input. cloning strategy using mammalian expression vector is highly efficient in delineating peroxisome biogenesis factors in mammals, i.e. PEX1 (16), PEX2 (17,22), PEX5 (14), PEX6 (18,19), and PEX12 (20,21), including the present work. While several human orthologues of yeast genes have recently been cloned by expressed sequence tag search, such as PEX1 (46,47), PEX5 (48), PEX7 (28,49,50), PEX12 (51), and more recently identified PEX10 (52,53) and PEX16 (54), direct cDNA cloning by functional complementation is an independent, parallel approach that often complements the cloning of human cDNAs of interest using expressed sequence tags. It is noteworthy that Fransen et al. (55) very recently identified Pex14p, using a combination of a peroxisomal protein import system in semipermeabilized CHO cells and antisera to peroxisomal membrane proteins. Human Pex14p was shown to bind Pex5pS in ligand blots, in good agreement with our findings.
The peroxin Pex14p from three mammalian species (i.e. rat, human, and Chinese hamster, each comprising 376, 377, and 377 amino acids) all showed nearly 30% identity in the amino acid sequence to Pex14p of H. polymorpha (31) and S. cerevisiae PEX14 (32,33), indicating a high degree of conservation of Pex14p in eukaryotic evolution. At present, however, we do not know if the yeast PEX14 gene can complement ZP110 and the mammalian PEX14 restores peroxisomal protein import in the yeast pex14 mutants (31)(32)(33). This type of complementation may be less likely, since no direct complementation by yeast genes of mammalian peroxisome assembly-defective mutant cells, including human patient fibroblasts, has been reported to date (3,7). In the present study, dysfunction of Pex14p in ZP161 cells was evidently demonstrated to be due to truncation in PEX14 gene transcripts. Furthermore, both alleles of PEX14 were found to be expressed in CHO cells, at least in ZP161, although the type of gene expression in CHO cells has been postulated to be hemizygotic. The level of PEX14 mRNA derived from each allele remains to be determined.
The present finding that Pex14p is located in the peroxisomal membranes as an integral PMP, with topology of the Nand C-terminal parts exposed to the cytosol, agrees well with the observation by Wendland and Subramani (56) demonstrating in patient fibroblasts that the molecular defects are mostly associated with the organelles, while cytoplasmic factors seem to be functionally normal. It is noteworthy that S. cerevisiae Pex14p has been shown to be peripheral (32) or integral (33) membrane protein of peroxisomes, H. polymorpha Pex14p being integral (31). While peroxisomal transport of membrane proteins, including PMP70, appears to be normal in the CG14 mutants, ZP110, ZP111, and ZP161, import of soluble proteins such as PTS1 and PTS2 proteins is defective. We showed that Pex14p interacts with a WD motif protein Pex7p (PTS2 receptor) and PTS2 protein, as was the case for S. cerevisiae Pex14p (32,33). Pex14p likewise binds to a tetratricopeptide repeat (TPR) protein Pex5p (PTS1 receptor), both isoforms Pex5pS and Pex5pL (14), carrying their cargo PTS1 protein. This is compatible with the finding in S. cerevisiae by in vivo studies using the yeast two-hybrid system (32,33), although only one type of Pex5p is present in yeast.
Given the fact that Pex5p and Pex7p bind to PTS1 and PTS2 proteins, respectively, it is more likely that several types of the receptor-cargo complex, such as Pex5p-PTS1 protein and Pex7p-PTS2 protein, translocate to a peroxisomal import machinery, presumably comprising Pex14p as well as other peroxin(s) and mediating the transport of polypeptides into the peroxisomal matrix. Thereby, Pex14p possibly plays a pivotal role in the peroxisome assembly process as a factor, e.g. a convergent component of the import machinery of matrix proteins but not membrane polypeptides. The interacting site(s) of Pex14p with Pex5p and Pex7p, both carrying their cargoes, are presumably in the cytosolic phase of Pex14p. Precise domains of Pex14p, each responsible for the interaction with Pex5p and Pex7p as well as for the homomeric binding, remain to be investigated. It would be also intriguing to study whether Pex14p interacts with other mammalian peroxisome biogenesis factors so far identified, including RING peroxins Pex2p (17,22,43), Pex10p (52,53), and Pex12p (20,21,51); AAA family peroxins Pex1p (16,46,47) and Pex6p (18,19,57); a Src homology 3 peroxisomal membrane protein Pex13p (58); and Pex16p (54). Such studies may be done by in vitro binding assay developed in this report and using semipermeabilized CHO cells, including wild-type CHO-K1 as well as PEX14 mutants ZP110, ZP111, and ZP161.