Newly Identified Chinese Hamster Ovary Cell Mutants Are Defective in Biogenesis of Peroxisomal Membrane Vesicles (Peroxisomal Ghosts), Representing a Novel Complementation Group in Mammals*

We isolated peroxisome biogenesis-defective mutants from Chinese hamster ovary cells by the 9-(1′-pyrene)nonanol/ultraviolet (P9OH/UV) method. Seven cell mutants, ZP116, ZP119, ZP160, ZP161, ZP162, ZP164, and ZP165, of 11 P9OH/UV-resistant cell clones showed cytosolic localization of catalase, a peroxisomal matrix enzyme, apparently indicating a defect of peroxisome biogenesis. By transfection of PEX cDNAs and cell fusion analysis, mutants ZP119 and ZP165 were found to belong to a novel complementation group (CG), distinct from earlier mutants. CG analysis by cell fusion with fibroblasts from patients with peroxisome biogenesis disorders such as Zellweger syndrome indicated that ZP119 and ZP165 were in the same CG as the most recently identified human CG-J. The peroxisomal matrix proteins examined, including PTS1 proteins as well as a PTS2 protein, 3-ketoacyl-CoA thiolase, were also found in the cytosol in ZP119 and ZP165. Furthermore, these mutants showed typical peroxisome assembly-defective phenotype such as severe loss of resistance to 12-(1′-pyrene)dodecanoic acid/UV treatment. Most strikingly, peroxisomal reminiscent vesicular structures, so-called peroxisomal ghosts noted in all CGs of earlier Chinese hamster ovary cell mutants as well as in eight CGs of patients’ fibroblasts, were not discernible in ZP119 and ZP165, despite normal synthesis of peroxisomal membrane proteins. Accordingly, ZP119 and ZP165 are the first cell mutants defective in import of both soluble and membrane proteins, representing the 14th peroxisome-deficient CG in mammals, including humans.

The hierarchy of the highly organized biogenesis of the dif-ferent subcellular compartments is one of the major characteristics of eukaryotic cells. Peroxisome, a single membranebounded organelle, functions in various metabolic pathways such as ␤-oxidation of very long chain fatty acids and synthesis of ether phospholipids, plasmalogens (1). Significant progress has recently been made in our understanding of the biogenesis of peroxisomes by findings such as those of targeting signals and protein factors, peroxins (2) required for peroxisome assembly, and through disparate studies using yeast and mammalian cells (3)(4)(5). Both peroxisomal membrane and matrix proteins are imported post-translationally into peroxisomes (3). New peroxisomes are then formed by growth and division of preexisting peroxisomes. Peroxisomal targeting signals, type 1 (PTS1) 1 and type 2 (PTS2) independently function in evolutionarily diverse organisms from yeasts to humans (5). Over 15 peroxisome biogenesis factors, peroxins, have been cloned (5)(6)(7)(8).
However, the process of peroxisomal membrane vesicle formation is little understood, despite the development of genetic approaches described above. We report here isolation and characterization of a newly identified CG of CHO cell mutants defective in peroxisome membrane biogenesis, apparently at the initial stage of peroxisome assembly. These CHO mutants apparently represent the most recently classified human CG-J.

Morphological Analysis
Peroxisomes in CHO cells were visualized by indirect immunofluorescence light microscopy using rabbit anti-rat catalase antibody, and those in fused cells of CHO mutants with human fibroblasts were detected with either rabbit anti-human catalase antibody or anti-rat catalase antibody, as described (12). Several CHO mutant cell clones were stained with rabbit antibodies specific for PTS1 peptide comprising the C-terminal, 10 amino acid residues of rat acyl-CoA oxidase (AOx) (18), rat 3-ketoacyl-CoA thiolase (16), C-terminal 15-amino acid peptide of rat peroxisomal 70-kDa integral membrane protein (PMP70) (43), and C-terminal 20-amino acid peptide of Pex12p (35). Antigenantibody complex was detected by fluorescein isothiocyanate-labeled goat anti-rabbit immunoglobulin G antibody (Cappel, Durham, NC) under a Carl Zeiss Axioskop FL microscope (Oberkochen, Germany) using a number 17 filter.

DNA Transfection
cDNA transfection to CHO mutant cells was done by means of liposome-mediated transfection with plasmids of rat PEX2, human PEX5, rat PEX6 expressing vector (17), and rat PEX12 (35). Transfection was carried out for 3 h with 1 g of cDNA, using 12 g of LipofectAMINE (Life Technologies, Inc.), according to the procedure recommended by the manufacturer. The cells were cultured for 2 days and then incubated overnight in 2 ml of serum-free F12 medium before immunostaining.

Expression of GFP Fusion Protein
Plasmid expressing "enhanced" green fluorescent protein (EGFP) (44) fused at the C terminus with human PMP70 (EGFP-PMP70) was constructed as follows. Human PMP70 cDNA (45) was inserted into the NotI site in a mammalian expression vector, pUcD2SR␣MCSHyg (36), containing hyg gene in pUcD2SR␣MCS (32). A SacII-BamHI fragment obtained from this plasmid was inserted into SacII-BamHI site in pEGFP-C1 vector (CLONTECH, Palo Alto, CA). Plasmid expressing a fusion protein of EGFP with rat Pex12p (EGFP-Pex12p) was likewise constructed. The BamHI site was created immediately upstream from the initiator methionine in rat Pex12p cDNA (RnPEX12) (35) by polymerase chain reaction using a forward primer, 5Ј-CGCGGATCCCTAC-TATGGCTGAGCATGG-3Ј (initiation codon is underlined), and a reverse primer, 5Ј-GTATTCAGAAAATGAGGC-3Ј (residues 239 -256, starting from the first nucleotide of the initiator methionine codon of RnPEX12 open reading frame). pBS⅐RnPEX12(B-A) was constructed by replacing a BamHI-SpeI fragment of pBS⅐RnPEX12 (35) with a BamHI-SpeI fragment of the polymerase chain reaction fragment. A BamHI (blunted)-ApaI fragment of pBS⅐RnPEX12(B-A) was replaced into BglII (blunted)/ApaI-digested part of pEGFP-C1, thereby constructing pEGFP⅐RnPEX12. Plasmids for EGFP-PMP70 and EGFP-Pex12p contained deduced sequence at a linker part, SGLRSRALAAA and FGLR-SIP, respectively. EGFP-PMP70 and EGFP-Pex12p were expressed in CHO cells by transient transfection. Transfectants grown on cover glass were fixed and observed under a Carl Zeiss Axioplan 2 microscope using a number 17 filter.

Cell Fusion
Parent CHO cells and cells to be fused were co-cultured for 1 day and then fused with polyethylene glycol as described (16). Selection of fused cells was carried out with 1 mM ouabain or 100 g/ml hygromycin B (Sigma) (16,17). Cell fusion of CHO mutant variants resistant to 6-thioguanine (TG r ) with human fibroblasts was done as described (12).

Subcellular Fractionation
CHO-K1, Z65, and ZP119 cells were homogenized, as described (16), except that protease/inhibitor mixture containing 20 g/ml each of antipain and pepstatin, 20 g/ml of leupeptin, and 50 units/ml aprotinin was added to the homogenization buffer. A postnuclear supernatant fraction, prepared by centrifugation of homogenates at 750 ϫ g for 10 min, was then centrifuged at 100,000 ϫ g for 1 h to separate organelles (heavy and light mitochondrial and microsomal fractions) and cytosol. Each fraction was analyzed by SDS-polyacrylamide gel electrophoresis, electrophoretically transferred to polyvinylidene difluoride membrane (Bio-Rad), and probed with specific antibody. Antigen-antibody complex was visualized with ECL Western blotting detection reagent (Amersham Pharmacia Biotech, Tokyo, Japan).

Radiolabeling of Cells
Continuous Labeling-Metabolic labeling of cells with 10 Ci/ml [ 35 S]methionine plus [ 35 S]cysteine (NEN Life Science Products) for 1 or 24 h in F12 medium and immunoprecipitation of peroxisomal proteins from cell lysates were done as described (16), in the presence of protease/inhibitor mixture.
Pulse-Chase Experiments-Cells growing in a 35-mm dish were pulse-labeled for 1 h with 100 Ci/ml [ 35 S]methionine plus [ 35 S]cysteine (Amersham). The medium was removed, and the cells were washed twice with phosphate-buffered saline and fed 2 ml of F12 plus 10% fetal calf serum medium. At selected intervals, cells were washed three times with phosphate-buffered saline and lysed as described (16), in the presence of protease/inhibitor mixture plus 2 mM phenylmethylsulfonyl fluoride.

Northern Blot Analysis
RNA blot of poly(A) ϩ RNA from wild-type CHO-K1, Z65, and ZP119 cells was hybridized with a 32 P-labeled probe of 0.6-kb NcoI-PvuII fragment of human PMP70 cDNA, under conditions of high stringency. Probe labeling was done with [ 32 P]dCTP using a Megaprime DNA labeling system (Amersham). The membrane was also hybridized with the 32 P-labeled, 1.3-kb fragment of cDNA for human glyceraldehyde-3phosphate dehydrogenase, as a control for load and integrity of the RNA. Washing of the membrane was done twice at room temperature and twice at 65°C with 2ϫ SSPE (1ϫ SSPE, 0.15 M NaCl, 10 mM sodium phosphate, 1 mM EDTA, pH 7.4), 0.5% SDS.

Isolation and Morphological Analysis of CHO Cell Mutants
Eleven CHO mutant cell clones resistant to P9OH/UV treatment were isolated from TKa, after two cycles of the mutant isolation steps. Three mutant clones, ZP119, ZP162, and ZP165, showed catalase in a diffused manner in the cytosol, as in CHO cell mutants previously isolated (12,16,17,19), whereas the wild-type CHO-K1 contained catalase-positive particles, peroxisomes ( Fig. 1, a-d). AOx and 3-ketoacyl-CoA thiolase, PTS1 and PTS2 proteins, respectively, were detected in the cytosol by cell staining with anti-AOx and anti-thiolase antibodies ( Fig. 1, e and f). ZP116, ZP160, ZP161, and ZP164 also contained catalase in the cytosol (data not shown). Taken together, these mutants are defective in peroxisomal import of matrix proteins. A part of cells of several mutant clones, ZP117, ZP118, ZP120, and ZP163, showed catalase-positive particles, presumably peroxisomes, but less in number as compared with CHO-K1 (not shown). The reason for P9OH/UV resistance of the apparently wild-type cells is presently unknown.

Cell Fusion Analysis
Fusion of ZP161 with ZP110 (19) showed catalase in cytosol, suggesting that ZP161 is in the same CG as ZP110 (Table I).
Catalase-positive particles, peroxisomes, were evident in fused cells of ZP119 with all CGs of CHO mutants thus far isolated (12,16,17,19), including Z24 (16) (Fig. 2c), thereby demonstrating that ZP119 belongs to a CG different from seven CGs of our earlier mutants (Table I). Fusion of ZP119 with ZP165 resulted in no appearance of catalase in peroxisomal punctate structure, indicating that ZP119 and ZP165 are in the same CG (Fig. 2d). Other pair-wise cell fusion resulted in restoration of peroxisomes except for the cell hybrids of Z65 with ZP116 and ZP160 (Table I). Furthermore, fused cells, ZP119 with wildtype CHO-K1 and ZP165 with CHO-K1, both showed peroxisomes as numerous as in the wild-type, implying that lesion(s) of allele(s) in the mutants were recessive (Fig. 2, e and f).

Cell Fusion with Patient Fibroblasts
To determine to which human CG the mutants ZP119 and ZP165 belonged, we carried out cell fusion of ZP119 and ZP165, each with several CGs of fibroblasts, including the recently identified CG-H (15) and CG-J, 2 from patients with peroxisome-deficiency disorders in which neither corresponding CGs of CHO cell mutants nor their complementing genes have been identified. Fused cells of ZP119 and ZP165 each with CG-J fibroblasts showed no peroxisomes (Fig. 3, b and d), as assessed by staining using anti-human catalase antibody, whereas numerous peroxisomes were evident in respective hybrids of ZP119 with patient fibroblasts of CG-A in Japan (the same as CG-VIII in the U. S.) and ZP165 with CG-D fibroblasts (Fig. 3, a and c). Peroxisomes were complemented in fused cells of ZP119 and ZP165 each with other CG fibroblasts, including those of CG-B (the same as CG-VII the U. S.), CG-D (the same as CG-IX), CG-G, CG-H, and CG-VI (Table II). Similar results were obtained when fused cells were stained using anti-rat catalase antibody (data not shown). These results strongly suggested that ZP119 and ZP165 are classified to human CG-J and not CGs A, B, D, G, and VI.
Together with the results obtained in CG analyses, we conclude that ZP119 and ZP165 belong to the 14th CG, the same group as CG-J, distinct from previously identified CGs of CHO cell mutants (12, 16 -19, 46) and peroxisome deficiency patients (10 -15) ( Table III).

Latency of Catalase
Intracellular location of catalase was also examined by the digitonin titration method (16). Full activity of catalase was detected at 100 g/ml digitonin, in the mutants ZP119 and ZP165, as in other groups of earlier CHO mutants (12,16,17,19), whereas 65% of the activity was latent in the wild-type cells, CHO-K1 (Fig. 4). Most of catalase activity in the wild-type was released only when digitonin concentration was increased up to 300 g/ml. Accordingly, it is apparent that catalase is localized in the cytosol in both mutants. Cell hybrids of ZP119 with Z24 showed a latency profile of catalase, very similar to that of the wild-type, indicating that peroxisomes were restored in the fused cells and that these mutants are mutually distinct in CG, consistent with the morphological data shown in Fig. 2c.

Biochemical Properties of CHO Cell Mutants
Catalase activity varied in each cell type, wild-type and mutants ZP119 and ZP165, apparently due to the difference in cell size (Table IV; see Fig. 1, a, b, and d), as noted in earlier mutants (16). ZP119 was severely deficient in DHAP-ATase activity, as compared with the level in the wild-type CHO-K1 (Table III). After cell culture in the presence of P9OH followed by short exposure to UV, nearly all ZP119 and ZP165 cells survived, as observed for earlier mutants (12,16,17,19), but the wild-type CHO-K1 was not viable. The mutants ZP119 and ZP165 were highly sensitive to P12/UV treatment, whereas over 60% of the wild-type were resistant, indicating a typical phenotype of peroxisome-deficient CHO cell mutants (12,(17)(18)(19)46).

Biogenesis of Peroxisomal Protein
Synthesis of peroxisomal matrix proteins was investigated by radiolabeling and immunoprecipitation. AOx, the first step enzyme of peroxisomal ␤-oxidation system, is a heterodimer consisting of 75-kDa A, 53-kDa B, and 22-kDa C polypeptide components (16,47). A form is proteolytically converted to B and C forms in peroxisomes (48). Polypeptides A and B of three 35 S-polypeptide components were noted in the wild-type CHO-K1 with barely detectable C which was apparently attributed to the low content of methionine and cysteine (16,47), whereas a reduced amount of 75-kDa 35 S-A form of AOx, but not converted B and C forms, was detected in mutants ZP119 as well as Z24 (Fig. 5, lanes 1-3), consistent with our earlier findings using CHO cell mutants and fibroblasts from patients (12, 16 -19). The third enzyme of the peroxisomal ␤-oxidation system, 3-ketoacyl-CoA thiolase (thiolase), is synthesized as a FIG. 2. Complementation group analysis of CHO cell mutants. CG analysis was done by PEX cDNA transfection and cell fusion method. Cells were stained with rabbit anti-rat catalase antibody. Immunofluorescent microscopy was done as in Fig. 1. a, PEX5-

TABLE II Complementation group analysis by cell fusion
The fibroblasts were from peroxisome-deficient patients belonging to CGs whose complementing genes were not identified. After cell fusion, fused cells of ZP119 and those of ZP165 were stained with antisera to human catalase or rat catalase. ϩ, peroxisomes were complemented; Ϫ, peroxisomes were not complemented; ND, not determined.

Patient fibroblasts
44-kDa precursor and then processed to a 41-kDa mature form in peroxisomes (12,16). Only a 44-kDa [ 35 S]thiolase precursor was seen in ZP119 and Z24, consistent with the observation with earlier mutants (12,16,17,19), although the 41-kDa mature protein was apparent in the wild-type cells (Fig. 5,  lanes 5-7). Thus, peroxisomal proteins are apparently synthesized at a normal level but not processed to mature enzymes in the mutants. AOx and thiolase was properly processed in the hybrids of ZP119 with Z24 (Fig. 5, lanes 4 and 8), as in the wild-type, thereby supporting the notion in cell fusion study that the CHO mutant ZP119 belongs to a complementation group distinct from Z24 (see Fig. 2c).

Morphological Analysis of Peroxisomal Memebrane Protein
Intracellular localization of peroxisomal membrane proteins was investigated. Larger but fewer particles, similar to "peroxisomal ghost" vesicles found in fibroblasts from Zellweger patients (22-25), were detected, when PEX2-defective CHO cell mutants, Z65 (16, 29) (Fig. 6, e) and ZP116 (data not shown), were stained with anti-PMP70 antibody, consistent with findings on earlier mutants such as Z24 and ZP92 (12,17,19). Similar results were obtained using antibody to PMP70 isolated from rat liver peroxisomes (16) (data not shown). In contrast, such particles immunoreactive to anti-PMP70 antibody were not discerned in ZP119 (Fig. 6a) and ZP165 (data not shown). However, unambiguous difference in fluorescence intensity was not obtained between cell staining with anti-PMP70 antibody and that only with fluorescein isothiocyanatelabeled anti-rabbit IgG antibody (data not shown). Thus we investigated expression of an ectopic protein in mutant cells followed by its detection without use of antibody. A chimera protein, EGFP-PMP70, was visualized, rather in an apparently diffused manner, when expressed in ZP119 and assessed by direct fluorescence microscopy (Fig. 6b), whereas this fusion protein was in numerous particles in CHO-K1 (Fig. 6f), as seen for PMP70 (12) (not shown). Z65-and PEX5-defective ZP162, both expressing EGFP-PMP70, showed punctate-staining particles (data not shown), but less in number than those in CHO-K1, like PMP70-positive ghosts in Z65 (Fig. 6e). Moreover, another peroxisomal integral membrane protein, a recently identified rat peroxin Pex12p (35), was expressed in ZP119 and Z65, and then stained with anti-Pex12p antibody. Pex12p was likewise in diffused, cytosolic staining appearance in ZP119, whereas Pex12p-positive ghost-like particles, but less in number as compared with those in CHO-K1 (data not shown), were noted in Z65 cells (Fig. 6, c and g). Large vesicles immunoreactive to anti-Pex12p antibody, similar to those in Z65, were also found in PEX12-transfected mutants, ZP116 and ZP162 (not shown). A cDNA encoding a fusion protein, EGFP-Pex12p, was expressed in ZP119 and CHO-K1 cells. EGFP-Pex12p was present in the cytosol in ZP119, as for Pex12p in PEX12-transfected ZP119, whereas numerous particles, presumably visualizing localization of EGFP-Pex12p in peroxisomal membranes, were detected in CHO-K1 cells (Fig.  6, d and h). ZP165 showed nearly the same cytosolic appearance of EGFP-Pex12p as ZP119, whereas Z65 and ZP116 revealed larger fluorescent particles, possibly visualizing this fusion protein localized in peroxisomal ghosts (data not shown). Collectively, biogenesis and/or transport of peroxisomal membrane proteins, including PMP70 and Pex12p, appear to be impaired in ZP119 and ZP165.  4. Latency of catalase in wild-type, mutant, and hybrid cells. Latency of catalase was determined as described (16). ⅜, wildtype CHO-K1; q, ZP119; OE, ZP165; ‚, hybrid cells of ZP119 with Z24TG r Oua r . The results were represented as an average of duplicate assays.

Biogenesis of Peroxisomal Membrane Protein
Continuous Radiolabeling and Western Blot Analysis-To examine the biogenesis of peroxisomal membrane proteins, metabolic labeling of CHO cells was carried out for 24 h with [ 35 S]methionine and [ 35 S]cysteine. A 35 S-polypeptide of 70 kDa was immunoprecipitated by anti-PMP70 antibody in both wildtype CHO-K1 and mutant Z65 cells, whereas a 70-kDa 35 Sprotein was not obtained from ZP119 (Fig. 7A, upper panel,  lanes 1-3). Immunoprecipitation with anti-3-ketoacyl-CoA thiolase antibody yielded 41-kDa mature [ 35 S]thiolase in the wild-type cells and 44-kDa 35 S-precursor in Z65 and ZP119 (Fig. 7A, lower panel, lanes 1-3), confirming proper labeling of peroxisomal proteins as in Fig. 5.
Immunoblot analysis using anti-PMP70 antibody of subcellular fractions of CHO cells demonstrated that PMP70 was present in the postnuclear fraction of wild-type CHO-K1 and Z65 cells, then recovered in organellar fraction (Fig. 7b, lanes  1-3 and 7-9, respectively), presumably in peroxisomes in CHO-K1 and peroxisomal ghosts in PEX2-defective ZP65 (12,16,49), as noted above. This result confirmed adequate separation of membrane fraction (organelles) from the cytosol. In contrast, polypeptide immunoreactive to anti-PMP70 antibody was barely detected in ZP119, consistent with findings by morphological staining and radiolabeling experiments described above.
The same results were obtained using antibody to purified rat PMP70 (16) in radiolabeling/immunoprecipitation and im-munoblot analyses (data not shown), thereby excluding the possibility that failure in detection of PMP70 in ZP119 was due to the loss by proteolysis of the C-terminal part, the sequence used for raising rabbit antisera. Moreover, the level of PMP70 noted in Fig. 7b, in immunoblots of cell lysates as well as subcellular fractions of ZP119, was not altered by cell culturing in the presence of a proteasome inhibitor MG132 (50) at 50 M for 8 h, or lysosomal inhibitors at 100 M each of leupeptin and pepstatin for 7 h (51), or even at 16°C for 24 h (51) (data not shown).
Taken together, the results were interpreted to mean that biogenesis of peroxisomal membrane proteins, including PMP70, is defective in ZP119.
Pulse-Chase Experiment-We next examined the kinetics of labeling of peroxisomal proteins in pulse-chase radiolabeling experiments. Cells were pulse-labeled with [ 35 S]methionine and [ 35 S]cysteine for 1 h, and the radioactivity was chased for 0.5, 1, 3, 6, 12, and 24 h in normal Ham's F12 medium (Fig. 8,  a and b). The anti-PMP70 antibody immunoprecipitated 35 S-PMP70, at the initial 1-h pulse-labeling, from both CHO-K1 and ZP119 but apparently less in amount in ZP119 than that in the wild-type cells. The PMP70 was stable for at least 24 h during the chase in the wild-type cells (Fig. 8a, lanes 1-7). In contrast, 35 S-PMP70 was hardly detected at 30-min chase, then no more discerned during further chase in ZP119, strongly suggesting a rapid turnover of PMP70, presumably by an increased rate of degradation (Fig. 8a, lanes 8 -14). This is consistent with the notion that PMP70 was apparently absent in ZP119, as tested by 24-h continuous 35 S labeling and immunoblot analyses.
Immunoprecipitation with anti-3-ketoacyl-CoA thiolase antibody yielded little 44-and mostly 41-kDa 35 S-polypeptides in the wild-type cells (Fig. 8b, lanes 1-7; open and solid arrowheads). The larger polypeptide disappeared within 1 h in the chase, with a concomitant increase in the 41-kDa protein. In the mutant ZP119, the 44-kDa 35 S-polypeptide was synthesized and remained in the same form. The radioactivity of the band gradually decreased during the 24-h chase (Fig. 8b, lanes  8 -14). Thereby, 3-ketoacyl-CoA thiolase was synthesized as a larger precursor of 44 kDa, in both wild-type and mutant cells, then processed to 41-kDa mature protein in wild-type cells but not in mutant ZP119, as noted in Z24 and Z65 (16).
Furthermore, 35 S-PMP70 from ZP119 labeled with [ 35 S]methionine and [ 35 S]cysteine for 1 h was found by immunoprecipitation in subcellular fractions, mostly in the cytosolic faction than the organelle fraction, whereas 35 S-PMP70 was in the organelle fraction from CHO-K1 (Fig. 8c, upper panel). Catalase was detected by immunoblotting, predominantly in the cytosolic fraction from ZP119 (Fig. 8c, lower panel), consistent with our earlier observation using other CHO cell mutants such as ZP65 (16), whereas it was in the organelle fraction from the wild-type cells. This result confirmed adequate separation of the subcellular fractions.
Collectively, these results demonstrate that PMP70 is initially synthesized in the cytosol and rapidly translocated to  (16). Peroxisomal DHAP-ATase was measured as described (60). Survival rate of P9OH/UV-and P12/UV-resistant cells was expressed as a percentage of the unselected control (12). CHO   peroxisomes in the wild-type cells but mostly stays in the cytosol, apparently rather degraded more rapidly in the ZP119.

Northern Blot Analysis
On Northern blotting, PMP70 mRNA was detected as a single band of about 3.6 kb in CHO-K1, Z65, and ZP119 cells, but at nearly 2-fold lower level and with apparently the same size as compared with that in rat liver (Fig. 9). The amount and size of PMP70 mRNA were indistinguishable between mutant and wild-type cells, suggesting that the transcription of pmp70 gene in mutants ZP119 as well as Z65 was not affected. The result agrees well with the observation in the pulse-chase experiment described above. DISCUSSION In the present work, we isolated seven peroxisome-deficient CHO cell mutants, ZP116, ZP119, ZP160, ZP161, ZP162, ZP164, and ZP165, by means of a combination of selection using the P9OH/UV method and monitoring intracellular location of catalase. The mutants revealed phenotypic properties, including no latency of catalase, recessive mutation, severe loss of DHAP-ATase, and high sensitivity to the P12/UV treatment, despite normal synthesis of peroxisomal matix proteins, as noted for earlier peroxisome-deficient CHO mutants (12,16,17,19,46,52). Besides these common characteristics, most striking was no morphologically recognizable peroxisomal ghost vesicles found in ZP119 and ZP165. Hence, ZP119 and ZP165 are considered to be the first mammalian cell mutants absent from peroxisomal reminiscent membrane ghosts that have been detected in all seven CGs of CHO mutants (12,(17)(18)(19)46) and eight CGs of the fibroblasts from patients with peroxisome deficiency disease such as Zellweger syndrome (22)(23)(24)(25). Morphological and biochemical studies, including those using EGFP fusion proteins, EGFP-PMP70 and EGFP-Pex12p, demonstrated that ZP119 and ZP165 were impaired in transport of not only peroxisomal matrix proteins but also PMP70 and other peroxisomal membrane proteins such as Pex12p. Rapid degradation of PMP70 in ZP119 and presumably ZP165, despite its normal synthesis, is compatible with a barely detectable level of PMP70 visualized by cell staining using anti-PMP70 antibody. However, we cannot rule out that these two mutants contain peroxisomal membrane vesicles, if any, distinct from the presently known peroxisomal reminiscent ghosts but associated with membrane polypeptide(s) not examined in this study. Moreover, given the several recent reports describing possible involvement of ER in peroxisome biogenesis in yeast (53,54), it is plausible that a gene that targets nascent peroxisomal membrane proteins to the ER, rather than peroxisomes, is impaired in ZP119 and ZP165.
The mutants ZP119 and ZP165 will be more useful for such studies on peroxisome assembly and cloning of a gene encoding a peroxin essential for peroxisome biogenesis, particularly that required for peroxisomal membrane formation, a prerequisite step at early stages of peroxisome assembly. Such factors are less likely to be identified using 12 CGs of earlier CHO mutants as well as fibroblasts from patients in which import of peroxisomal membrane proteins is not affected (5,7). Transfection of human PMP70 cDNA did not complement ZP119 (data not shown), consistent with a normal level of PMP70 mRNA expressed in this mutant, unless mutated, inferring that PMP70 is not required for peroxisome membrane assembly, at least in this CG.
It is noteworthy that two yeast mutants, Hansenula polymorpha pex3 (formerly per9) and Pichia pastoris pex3 (pas2), defective in peroxisome biogenesis were reported to be apparently absent from morphologically recognizable, peroxisomal structures (57,58). Methanol-induced PEX3-null cells of H. polymorpha lacked peroxisomal vesicles, as assessed by immunoelectron microscopy using antibody to Pex10p, a peroxisomal integral membrane protein (57). Similarly, peroxisomal remnants were not detected by immunoelectron microscopy with anti-thiolase antibody, in methanol-and oleate-induced P. pastoris PEX3-null cells, although Pex5p, a PTS1-receptor, sedimented in a membrane-associated form upon density gradient subcellular fractionation (58). In each report, only a single peroxisomal membrane protein was studied in morphological analysis. To assess these mutants that are indeed defective in the early stages of peroxisomal membrane vesicle formation and to delineate the biological function of Pex3p, further investigation, including those at molecular and cellular levels, may be required. It is also of interest to examine whether the yeast PEX3 or its mammalian homolog, if any, complements peroxisome biogenesis in ZP119 and ZP165.
TKa, rat PEX2-transformed CHO-K1, was used in the present mutant isolation work to avoid frequent isolation of PEX2defective Z65-type mutants (12). In our previous reports (17)(18)(19)42), none of the isolated mutants was practically the Z65type mutant of CG-F (the same CG as CG-X in U. S.). However, two mutants isolated in the present study, ZP116 and ZP160, were found to be in the same CG as Z65, implying that introduced PEX2 cDNA (42) was disintegrated during the initial cultivation of TKa, prior to the mutagenesis using N-methyl-NЈ-nitro-N-nitrosoguanidine, thereby resulting in a type of the TKa cells with lesser or only bona fide PEX2 gene. We cannot exclude an alternative possibility that PEX2 gene(s), including the exogenously introduced PEX2 cDNA, were readily mutated in this study. Accordingly, in order to ensure that all TKa cells were indeed a PEX2-transformant of wild-type CHO cells, it will be more practical to culture TKa cells in the presence of neomycin, a maker for the rat PEX2-expressing vector, for certain times, e.g. a week or so, prior to the mutagenesis and mutant isolation.
Several P9OH/UV-resistant mutants isolated here, ZP117, ZP118, ZP120, and ZP163, each represented apparently heterogeneous cell population, with two types of intracellular location of catalase, i.e. cytosolic and particle-associated, despite several cycles of cell cloning by the limited dilution method. These cells behaved like a "leaky" peroxisome-positive mutant containing less number of peroxisomes. It is unclear whether the peroxisome-containing cells reflect rather high frequency of revertant mutation(s). Similar resistance to the P9OH/UV treatment was noted for CHO cells (17,19) and murine macrophage-like cell lines (59) that were isolated by the same method used in the present work. We should await further investigation to clarify this observation.