PAY4, a Gene Required for Peroxisome Assembly in the Yeast Yarrowia ZipoZytica, Encodes a Novel Member of a Family of Putative ATPases*

PAY genes are required for peroxisome assembly in the yeast Yarrowia lipolytica. Here we characterize one mutant, pay4, and describe the cloning and sequencing of the PAY4 gene. The pay4 mutant shows no identifiable peroxisomes by biochemical and morphological criteria. The complementing PAY4 gene encodes a polypeptide, Pay4p, 1025 amino acids in length and having a pre- dicted molecular mass of 112,258 Da. The predicted Pay4p sequence contains two putative ATP-binding do- mains and shows structural relationships to other potential ATP-binding proteins involved in biological processes as diverse as peroxisome biogenesis, vesicle- mediated protein transport, cell cycle control, and transcriptional regulation. These proteins all share a highly conserved stretch of approximately 175 amino acids that contains a consensus sequence for ATP binding. Pay4p shows sequence conservation with Paslp and PasSp, putative ATPases required for peroxisomal as- sembly in the yeasts Saccharomyces cerevisiae and Pichia pasforis, respectively. Pay4p, Paslp, and PasSp are presumably related members of a family of putative ATPases involved in peroxisome biogenesis. Pay4p is synthesized in low amounts in K lipolytica cells grown in glucose,

techniques, thereby permitting the dissection of the mechanism of peroxisome biogenesis by a genetic approach. Second, Y Eipolytica grows well on oleic acid relative to a yeast like S. cerevisiae, with a concomitant strong induction of peroxisomes and peroxisomal proteins. This feature facilitates experimental procedures such as the identification of peroxisomal assembly mutants and the isolation of affected genes by functional complementation that involve the observation of growth on oleic acid, as well as the analysis of proteins required for peroxisomal assembly.
Herein we report the detailed morphological and biochemical characterization of one pay mutant, pay4, and the cloning and analysis of the functionally complementing PAY4 gene. The mutant pay4 strain fails to assemble normal peroxisomes and mislocalizes peroxisomal matrix enzymes to the cytosol. The PAY4 gene encodes a polypeptide, Pay4p, of -112 kDa. Pay4p is a novel member of a superfamily of putative ATPases and shows similarity to proteins involved in processes as diverse as vesicle-mediated transport, control of cell cycle, and gene expression in human immunodeficiency virus. Pay4p also shows similarity to Paslp and Pas5p, putative ATPases shown to be essential for peroxisome assembly in S. cerevisiae (Erdmann et al., 1991) and I! pastoris,2 respectively. Together these proteins form a family of putative ATPases involved in peroxisome biogenesis in yeast.
Isolation of Pay4 Mutants-Pay4 mutants were isolated after mutagenesis of E122 cells with 1-methyl-3-nitro-1-nitrosoguanidine (Gleeson and Sudbery, 1988). The screening protocol included selection for an inability to utilize oleic acid as a carbon source, fractionation into 2OkgP (primarily peroxisomes and mitochondria) and 2OkgS (primarily cytosol) fractions of yeast cells (Aitchison et al., 1991), and immunofluorescence microscopy (Aitchison et al., 1992b) with anti-SKL antibodies. Fixation and preparation for electron microscopy was performed as described previously (Waterham et al., 1992) Mutants were characterized by standard genetic techniques for E: lipolytica (Gaillardin et al., 1973).
Cloning and Characterization of the PAY4 Gene-To isolate the PAY4 gene, the strain pay4 was transformed by electroporation with a ge-' S. Subramani, personal communication.
nomic DNA library of E: lipolytica contained in the Escherichia coli shuttle vector pINA445 (Nuttley et al., 1993). Leu' transformants were screened on YNO-agar plates for their ability to utilize oleic acid as a sole carbon source. Complementing plasmids were recovered by transformation of E. coli.
Standard recombinant DNA methodology including enzymatic modification of DNA, DNA fragment purification, and plasmid isolation was performed essentially as described in Ausubel et al. (1989).
DNA Sequencing-Various restriction endonuclease fragments of the PAY4 gene were inserted into the vector pGEM"IZfl+) (Promega, Madison, WI) for dideoxynucleotide sequencing of both strands from doublestranded templates (Sanger et al., 1977;Zhang et al., 1988). The deduced Pay4p sequence was used to search the SWISS-PROT Protein Sequence Data Bank release of August 23, 1992, for similarities with other known protein sequences using the FSTPSCAN program of PC/ GENE (IntelliGenetics, Mountain View, CA).
Integrative Disruption of the PAY4 Gene-The LEU2 gene of E: lipo-Zytica was isolated as an Eco47IIYEglII fragment from pINA445 and inserted into the PAY4 gene digested with StuI and BglII. This construction replaced 809 base pairs (bp) of the PAY4 gene open reading frame with an approximately 2.1-kbp fragment containing the entire LEU2 gene. This construct was digested with SphI and ScaI to liberate the LEU2 gene flanked by 720 and 280 bp of the PAY4 gene at its 5' and 3' ends, respectively. This linear molecule was used to transform E: lipolytica E122 to leucine prototrophy. Leu' transformants were screened for the ole-phenotype and mated to 22301-3. Diploids were sporulated, and random spore analysis performed. The segregation of the Leu+ and ole-phenotypes was analyzed by replica plating. pay4::LEUa segregants were mated to pay4 and the resultant diploids checked for complementation.
Antisera-For the production of antibodies to Pay4p, a 1302-bp fragment of the PAY4 gene open reading frame encoding amino acids 24-458 of Paylp was excised with Ecl 13611 and EgZII and inserted into the pMALc2 vector (New England Biolabs) in-frame and downstream of the open reading frame encoding the maltose binding protein. Expression of the maltose binding protein gene is under the control of the "tac" promoter, which is induced in the presence of isopropyl-P-D-thiogalactopyranoside. A lysate of E. coli synthesizing the maltose binding protein-Pay4p fusion was prepared essentially as described by Ausubel et al. (1989), and the fusion protein was purified further by SDS-PAGE (Laemmli, 1970;Fujiki et al., 1984) on a 10% preparative gel. The fusion protein was electroeluted into dialysis tubing, dialyzed against 50 rn ammonium bicarbonate, lyophilized, and dissolved in a minimal amount of distilled, deionized water. Antibodies were raised by primary injections of 100 pg of fusion protein into New Zealand White rabbits at multiple subcutaneous sites. m e r 5 weeks, booster injections of 50 pg were administered subcutaneously at multiple sites. Six weeks after the primary injections, the rabbits were sacrificed, their blood collected, and sera prepared. The specificities of antisera were determined by western blotting (Burnette, 1981) of yeast cell lysates (Needleman and Tzagoloff, 1975;Nuttley et al., 1990). Antigen-antibody complexes were detected by enhanced chemiluminescence (Amersham Corp.) and quantitated with a GS-300 scanning densitometer and the GS-350 data analysis system (Hoefer Scientific Instruments, San Francisco, CA).
Anti-SKL serum was a gift of Dr. S. Subramani, University of California, San Diego. Antiserum to S. cerevisiae peroxisomal thiolase was a gift of Dr. W.-H. Kunau, Ruhr-University.

RESULTS
Characterization of the Pay4 Mutant-The Y lipolytica mutant strain pay4 has been shown to be incapable of growth on oleic acid (Fig. 1, pay4; Nuttley et al., 1993), Pay4 has been shown to be a mutant in peroxisome assembly by two criteria: the lack of characteristic punctate peroxisomal staining with anti-SKL antibodies in double-labeling immunofluorescence microscopy and the mislocalization of peroxisomal marker enzymes t o the cytosol (Nuttley et al., 1993). Analysis of the parental and mutant strains by electron microscopy supports the classification of the pay4 mutant as peroxisome-deficient. Parental cells grown in the presence of oleic acid, show many peroxisomes (microbodies) scattered through out the cell ( peroxisomal staining in the mutant cells. Peroxisomal ghosts, aberrant membrane structures containing peroxisomal membrane proteins, have been described in a number of peroxisomedeficient mutants from various sources (Santos et al., 1988;Zoeller et al., 1989;Gould et al., 1992). Analysis of micrographs of oleic acid-induced pay4 cells has revealed no evidence of such structures in this strain.
A PTS-2-containing Protein Is Mislocalized to the Cytoplasm in the pay4 Mutant-We have previously demonstrated that peroxisomal proteins containing PTS-1 motifs are mislocalized to the cytoplasm in pay mutants (Nuttley et al., 1993). We investigated whether a peroxisomal protein containing a PTS-2 motif, i.e. thiolase, is also mislocalized to the cytoplasm in the mutant pay4. Antibodies to peroxisomal thiolase of S. cerevisiae recognize thiolase from E: lipolytica. Western blot analysis of 2OkgP (primarily peroxisomes and mitochondria) and 2OkgS (primarily cytosol; Aitchison et al., 1991) fractions of the parental E122 strain showed a single polypeptide band of molecular weight -43,000 exclusively in the 2OkgP fraction (Fig. 3, lane P ) , reflecting the peroxisomal localization of thiolase. Analysis of the pay4 mutant showed that thiolase is not localized to peroxisomes (2OkgP, lane P ) in this mutant but mislocalized to the cytoplasm (20kgS, lane S). Moreover, the thiolase in the 2OkgS (lane S ) of the pay4 mutant showed reduced electrophoretic mobility relative to that in the 2OkgP fraction of the parental E122 strain. Combining the E122 20kgP and pay4 2OkgS fractions showed that this difference in molecular weight was not an artifact of electrophoresis and is most likely evidence of proteolytic cleavage of an amino-terminal PTS-2 sequence upon import of thiolase into the peroxisomal matrix. Cleavage of an amino-terminal PTS-2 has been shown for rat liver peroxisomal thiolase (Swinkels et al., 1991), and our results represent the first demonstration, to our knowledge, that such a process may also occur in yeast peroxisomes.
Isolation of the PM4 Gene-Transformation of the pay4 mutant strain with a library of E: lipolytica genomic DNA yielded 42 transformants capable of restored growth on oleic acid.
Three independent recombinant plasmids, designated p o l , p02, and p03, were rescued into E. coli. A combination of restriction mapping and subcloning analysis showed that the complementing activity of the inserts was localized to a common 4.5-kbp segment (data not shown). Retransformation of pay4 with any one of pol, p02, or p03 confirmed that the recombinant plasmids were responsible for the complementing activity. A transformant, hereafter called PAY4 (Fig. 11, harboring the plasmid p o l was used for further study. The putative PAY4 gene was used for a gene disruption experiment with the E: lipolytica LEU2 gene, as described under "Materials and Methods." A leu+/ole-transformant, designated P4-K0, was isolated, and integration of the LEU2 gene into the PAY4 locus was confirmed by Southern blot analysis (data not shown). The recessive nature of the ole-phenotype was demonstrated by the ability of the P4-KO x 22301-3 diploid to grow on oleic acid ( Fig. 1,04-22). Sporulation of the diploid D4-22 showed cosegregation of the ole-and leu+ phenotypes. When an ole-, matB isolate from sporulation of D4-22 was back-crossed to the original pay4 mutant, the resulting diploids were incapable of growth on oleic acid, thereby confirming that the authentic PAY4 gene had been cloned.
Antibodies raised to the carboxyl-terminal PTS-1 sequence Ser-Lys-Leu (SKL) were used to investigate the targeting of F'TS-1 containing proteins by double-labeling immunofluorescence microscopy. These antibodies reveal a punctate pattern characteristic of peroxisomes in parental E122 cells (Aitchison et al., 1992b;Fig. 4, top panel ). This pattern was absent in the pay4 mutant strain (middle panel) but was restored in the pay4 strain transformed with p o l to give the PAY4 strain (bottom panel). The presence of peroxisomes was also demonstrated by examination of the PAY4 transformant by electron microscopy. As shown in Fig. 5, Panels A and B, transformed cells display a morphology indistinguishable from that of the parental strain, including the presence of numerous peroxisomes (Mb). Examination of the P4-KO strain (Fig. 5, Panel C ) , in which the PAY4 gene has been disrupted by insertion of the LEU2 gene, revealed that this strain lacked any morphologically detectable peroxisomes.
Transformation of pay4 with p o l to yield PAY4 also restored the correct localization of peroxisomal marker enzyme activities. In the parental strain E122, approximately 30% of the catalase activity and 60% of the p-hydroxybutyryl-CoA dehydrogenase activity were found in the 2OkgP fraction upon subcellular fractionation (Fig. 61, reflecting the peroxisomal location of these enzymes. The activities of these enzymes recovered in the 2OkgS fraction were due, at least partially, to leakage from peroxisomes broken during the fractionation procedure (Aitchison et al., 1991). The relatively larger amounts of catalase in the 2OkgS may be due to preferential leakage of catalase from the peroxisome (Alexson et al., 1985). There may also be cytosolic and peroxisomal isoforms of catalase in I: lipolytica, as there are in S. cereuisiue (Hartig and Ruis, 1986;Cohen et al., 1988); however, this has not been investigated further. In the pay4 mutant strain (as well as in the disrupted strain P4-K0), less than 3% of catalase activity and approximately 10% of dehydrogenase activity were recovered in the 2OkgP. Transformation of pay4 with p o l corrected this defect, resulting in the recovery of catalase and dehydrogenase activities in the 2OkgP fraction at levels comparable to those found in the parental strain E122. The localization of the mitochondrial marker enzyme cytochrome c oxidase to the 2OkgP fraction was not affected by the pay4 mutation, as comparable levels of cytochrome c oxidase activity were found in the 20kgP fractions of E122, pay4, PAY4, and P4-KO.
Nucleotide Sequence of the PW4 Gene and the Deduced Amino Acid Sequence of Pay4pSequencing of the 4567-bp BamHI fragment common to the three complementing plasmids pol, p02, and p03 revealed an open reading frame encoding a protein of 1025 amino acids and having a predicted molecular weight of 112,258 (Fig. 7). This predicted molecular weight is in agreement with the relative molecular weight of Pay4p determined by SDS-PAGE and western blot analysis (see Figs. 11 and 12). The nucleotides surrounding the proposed initiating codon conform to the consensus sequence for translation initiation in yeast, with a conserved A at position -3 and a conserved C a t position +5 relative to the A of the initiation codon (Cigan and Donahue, 1987).
A putative TATA element (TATATA'ITA) in the 5'-untranslated region of the PAY4 gene is found between positions -461 and -453 relative to the A of the proposed initiation codon. This relatively large distance between the putative TATA element and the initiator codon makes it unlikely that this element plays a role in the transcriptional regulation of the PAY4 gene (Xuan et al., 1990;Heslot and Gaillardin, 1992). It is possible t h a t t h e expression of the PAY4 gene is controlled by a "TATAless" promoter. Such a scenario probably exists for the Y fipoiytica LYS5 gene, which does not display an obvious TATA box. Transcription of this gene was shown to initiate at a 13-bp CA-rich region (position -50 relative to the A of the initiator codon) immediately downstream of a n 11-bp CT-rich region. This CT-rich region is preceded by a second 11-bp CT-rich region located approximately 70 bp upstream of the first CT-rich region (Xuan et al.. 1990). A strikingly similar situation exists in the region upstream of the initiator codon of the PAY4 gene. A 14-bp CA-rich region between -20 and -7 is immediately preceded by a 12-bp CT-rich region between -32 and -21. A second CT-rich region of 11 bp occurs between -77 and -67.
In Y lipolytica, the TAG . . . TGAT . . . TTT transcription termination motif (Zaret and Sherman, 1982;Sutton and Broach, 1985) is found at the 3' ends of four genes whose sequences have been published (Davidow et al., 1985(Davidow et al., , 1987Xuan et al., 1990;Kattig et al., 1991). This motif is also found in the PAY4 gene starting at position +3152. 73 nucleotides downstream of the stop codon. A " T A T A transcription termination motif (Henikoff and Cohen. 1984) is also found in the PAY4 gene starting at nucleotide +3264. Analysis of the Pay4p Sequence-A number of ATP-binding proteins have been shown by comparative studies using primary sequence and crystallographic data to share certain consewed motifs. The sequence motif (A or G)-X-X-X-X-G-KG or T). known as the "P-loop" or "A-motif," represents the consensus sequence for the most conserved of these motifs and is believed to form a flexible phosphate binding loop between a p-sheet and an a-helix (Walker et al., 1982;Saraste et al., 1990). This motif was found twice in Pay4p at amino acid positions 477-484 and 760-767. This arrangement of duplicated ATP-binding domains separated by 200-300 amino acids is characteristic of a recently described family of presumptivr MgY+-dependent Amases that include the peroxisomal assrm-  Pay4p and Pas5p sequences also revealed a high degree of sequence conservation over the ATP-binding domains. These two sequences also display a significant sequence conservation along the entire length of the proteins (Fig. 9). The nine most highly conserved domains are aligned in Fig. 10. The high degree of sequence conservation among the domains is evidenced by the fact that 16% of the amino acid residues are identical in all nine sequences, and an additional 29% of the amino acids are similar.
The similarities in Pay4p, Paslp, and Padp, together with the fact that these proteins were identified through their role in peroxisome biogenesis, might suggest that these proteins are functional homologues in the yeasts k: lipolytica, S. cerevisiae, and Rpastoris, respectively. However, comparison of the Pay4p and Paslp reveals that the similarity between the two sequences is limited primarily to the second ATP-binding domains of the two proteins (49.4% identity/l6.7% similarity), with overall sequence identity of 21.5% and sequence similarity of 14.9%. The primary sequence of Pas5p required for peroxisome assembly in the yeast R pastoris2 has been found to exhibit a high degree of conservation with Pay4p, with 59.0% identity and 15.3% similarity between their carboxyl-terminal ATP-binding domains and 46.0% identity and 15.3% similarity over the entire length of the polypeptides. Pas5p shows a lesser degree of conservation with Paslp, with 24.1% identity and 15.1% similarity, with the alignment being concentrated in the ATP-binding domain. Therefore, while Pay4p and Pas5p are closely related at the primary sequence level and may be functional homologues, Paslp appears to be a related, but nonhomologous, member of a superfamily of putative ATPases involved in peroxisome biogenesis.
Immunological Detection and Characterization of Pay4p "Rabbit polyclonal antibodies raised against a maltose binding protein-Pay4p fusion were used to identify the product of the PAY4 gene. The antibodies detected a polypeptide of -112 kDa in Y lipolytica grown in oleic acid that was not recognized by the preimmune serum (data not shown). The anti-Pay4p antibodies were used in a Western blot analysis to probe cell lysates of oleic acid-grown parental strain E122, the pay4 mutant strain, the PAY4 gene disruption strain P4-KO, and a second peroxisome assembly mutant strain pay2 (Fig. 11). The immunoreactive 112-kDa polypeptide was detected in cell lysates of the E122 parental strain and the pay2 mutant strain but was absent in cell lysates of the pay4 mutant strain and the knockout strain P4-KO. These data suggest that the mutation in the pay4 strain does not result in the production of a nonfunctional form of Pay4p but is most probably a nonsense mutation resulting in premature termination of translation.
Immunological detection of Pay4p was used to investigate the regulation of the PAY4 gene. Parental E122 cells were transferred from glucose medium to medium containing oleic acid as the sole carbon source. The presence of oleic acid as the sole carbon source allows for the proliferation of peroxisomes and the induction of peroxisomal enzymes in various yeast species. Cell lysates were prepared at various time points after transfer of E122 cells to oleic acid medium. The lysates were analyzed by western blotting with anti-Pay4p and anti-SKL antibodies (Fig. 12). In E122 cells grown in glucose (T = 01, a small amount of Pay4p could be detected in the cell lysate, but no anti-SKI, immunoreactive polypeptides were seen. Pay4p was detectably induced (5-fold) after 2 h in oleic acid medium and was induced a maximum 20-fold after 3.5 h in oleic acid medium. This induction of Pay4p was mirrored by the induction of the major anti-SKI, immunoreactive polypeptides after the shift of E122 cells into oleic acid medium.

DISCUSSION
A genetic approach using yeast has led to the identification of a number of genes called sec genes that encode proteins involved in the secretory process (for a review, see Pryer et al.   Morphologically identifiable peroxisomes are absent in pay4 cells, and both PTS-1 and PTS-2 containing proteins are mislocalized to the cytoplasm. In these respects, pay4 cells appear to share many characteristics ascribed to cells of patients with inborn errors affecting peroxisome assembly, like Zellweger syndrome, and therefore represent an excellent model system to investigate the genetic basis of such disorders.

GVLFYGPWCGKTLLAXAIANECQANFISI-KGPELLTMWFGESEANVRE 561 GVLWGPPGTGKTLHARAVANETGAFFFLr-NGPEVMSKMAGSSESNLRK 297 G V L P Y O P P O T C K T L L A K A V A T E V S~F I S V -K G P E L L S E 570 GVLKYGPPGTGKTLLARACAAQTKATFL~-AGPCtLVQ~rGDGAKL~ 235
Three genes encoding proteins required for peroxisome assembly in s. cerevisiae have been isolated and characterized (Erdmann et al., 1991;Hohfeld et al., 1991;Wiebel and Kunau, 1992). In this paper we report the cloning and characterization of the PAY4 gene, the first description of a gene required for peroxisome assembly in I: Zipolytica. Functional complementation of the pay4 mutant strain with the PAY4 gene reestablished peroxisome assembly in the mutant as evidenced by morphological (Figs. 4 and 5) and biochemical (Fig. 6) criteria.
Genetic proof for the authenticity of the cloned gene was obtained from gene disruption and genetic analysis of the resulting strain. Gene disruption experiments also demonstrated that the PAY4 gene was not required for cell viability and that peroxisomes are therefore not required for cell viability.
Nucleotide sequence analysis of the PAY4 gene identified a large open reading frame that could encode a polypeptide of 112,258 Da (Fig. 7). There is substantial evidence indicating that the product of the PAY4 gene is indeed a polypeptide of -112 kDa. Firstly, antibodies raised against a fusion between maltose binding protein and the putative PAY4 gene product recognize a protein with an M, of 112,000 in Western blots of cell lysates of the parental strain E122 and of a second pay mutant strain pay2, which is in a different complementation group from that of pay4 (Fig. 11). This protein of -112 kDa was absent in the pay4 mutant strain and the gene disruption strain P4-KO. Secondly, the synthesis of the -112-kDa polypeptide was induced by oleic acid and repressed by glucose in a manner similar to that of peroxisomal proteins recognized by anti-SKL antibodies (Fig. 12). Many yeast peroxisomal proteins (Kamiryo and Okazaki, 1984;Fujiki et al., 1986;Nuttley et al., 19901, and also proteins required for peroxisome assembly in S. cerevisiae (Erdmann et al., 1991;Hohfeld et al., 1991), have been shown to be induced by growth on oleic acid.
The hydropathy profile of Pay4p did not provide strong evidence for any membrane-spanning region and indicated rather a hydrophilic polypeptide (Kyte and Doolittle, 1982). Western blot analysis of subcellular fractions showed that Pay4p was not localized to the fraction enriched for peroxisomes (data not shown). This would suggest a cytoplasmic localization for Paylp, although at this time we cannot exclude the possibility of a peripheral association of this polypeptide with the peroxisomal membrane.
Pay4p shows remarkable partial sequence similarity with a number of proteins involved in important and diverse biological functions such as vesicle-mediated protein transport (NSF and SeclSp), control of cell cycle (Cdc48p, VCP, and P97), modulation of gene expression (TBP-11, and peroxisome biogenesis (Paslp and Pas5p) (Figs. 8 and 10). All these proteins have in common a stretch of approximately 175 amino acids that contains a consensus sequence for ATP binding. Therefore, it would appear that the PAY4 gene represents a member of a multigene family that has evolved to encode a large number and variety of proteins that control a diversified set of biological functions. The only common function that can be attributed to all these proteins appears to be the hydrolysis of ATP. ATPase activity has been demonstrated for the VCP-like protein of Xenopus oocytes (Peters et al., 1990) and has been suggested to be important for the functioning of NSF (Clary et al., 1990).
Paslp has also been suggested to be an ATPase based on the conservation of its ATP binding sites (Erdmann et al., 1991). We therefore propose that Pay4p is also a putative ATPase required for peroxisome assembly and that its tentative localization to the cytosol suggests that it acts at an early step in peroxisome assembly.
A comparison of sequence conservation would suggest that Pay4p of k: lipolytica, Paslp of S. cerevisiue, and Pas5p of I! pastoris represent members of a multigene family encoding putative ATPases required for peroxisome assembly. Whether or not these proteins are true functional homologues can best be addressed by heterologous complementation experiments. It will be interesting to determine whether other such ATPases are found in these yeasts and whether such ATPases are involved in peroxisome assembly in other yeast species and in mammals. An approach based on the polymerase chain reaction might be used to address this question most easily, with primers derived from regions conserved in the genes encoding these three proteins. An alternative approach is to attempt to complement the yeast lesions using mammalian cDNA libraries expressed in the mutants. Such an analysis opens up the possibility of rapidly identifying ATPases required for peroxisome assembly in humans and potential mutations of these proteins that could lead to genetic diseases of peroxisome assembly such as Zellweger syndrome.