Biogenesis of nonspecific lipid transfer protein and sterol carrier protein x: studies using peroxisome assembly-defective pex cell mutants.

Nonspecific lipid transfer protein (nsLTP; also called sterol carrier protein 2) with a molecular mass of 13 kDa is synthesized as a larger 15-kDa precursor (pre-nsLTP) with an N-terminal 20-amino acid extension presequence, as well as with the peroxisome targeting signal type 1 (PTS1), Ala-Lys-Leu, at the C terminus. The precursor pre-nsLTP is processed to mature nsLTP by proteolytic removal of the presequence, most likely after being imported into peroxisomes. Sterol carrier protein x (SCPx), a 59-kDa branched-chain fatty acid thiolase of peroxisomes, contains the entire pre-nsLTP moiety at the C-terminal part and is converted to the 46-kDa form and nsLTP after the transport to peroxisomes. We investigated which of these two potential topogenic sequences functions in biogenesis of nsLTP and SCPx. Morphological and biochemical analyses, making use of Chinese hamster ovary cell pex mutants such as the PTS1 receptor-impaired pex5 and PTS2 import-defective pex7, as well as green fluorescent protein chimeras, revealed that both pre-nsLTP and SCPx are imported into peroxisomes by the Pex5p-mediated PTS1 pathway. Nearly half of the pre-nsLTP remains in the cytosol, as assessed by subcellular fractionation of the wild-type Chinese hamster ovary cells. In an in vitro binding assay, only mature nsLTP, but not pre-nsLTP, from the cell lysates interacted with the Pex5p. It is likely, therefore, that modulation of the C-terminal PTS1 by the presequence gives rise to cytoplasmic localization of pre-nsLTP.

The mammalian nonspecific lipid transfer protein (nsLTP) 1 functions in vitro as a carrier protein in the transfer of a variety of lipids, such as phospholipids and cholesterol (for a review, see Ref. 1). It is identical to sterol carrier protein 2, stimulates the enzymatic conversion of lanosterol to cholesterol during the biosynthesis of cholesterol, and enhances the synthesis of bile acids (1). But, the physiological role of nsLTP is not yet well defined. nsLTP is localized in peroxisomes, as well as the cytoplasm, in several organs such as rat liver (2,3) and adrenal gland (1). nsLTP is synthesized as a larger precursor, termed pre-nsLTP, with a mass of 15 kDa on free polyribosomes (4) and then processed to its mature form of 13 kDa (4,5), apparently after transport into peroxisomes (2,4). Two types of cDNA for nsLTP of rat liver were cloned (6 -8), one encoding 15-kDa pre-nsLTP consisting of 143 amino acids, a precursor form of 123-amino acid mature nsLTP (9,10), and the other coding for a 547-amino acid protein with a mass of 59 kDa, termed sterol carrier protein x (SCPx) (see Fig. 1A).
pre-nsLTP and SCPx are expressed from one gene by alternative transcription initiation by two distinct promotors (11). The N-terminal, 20-amino acid extra sequence of pre-nsLTP is cleaved off to form mature nsLTP in peroxisomes (4,12). Interestingly, nsLTP was recently shown to interact with acyl-CoA oxidase (AOx) (13), the first-step enzyme of peroxisomal ␤-oxidation system, suggesting that nsLTP may function in transfer of the substrates such as fatty acyl-CoA derivatives to AOx. The sequence for SCPx contains the full sequence of pre-nsLTP at the C-terminal part and is suggested to be partly converted to the N-terminal part protein of 46 kDa and nsLTP (1,14). It has been demonstrated that SCPx functions as a peroxisomal branched-chain ␤-ketothiolase (15)(16)(17).
The import of most peroxisomal matrix proteins is mediated by well characterized cis-acting peroxisomal targeting signals (PTSs), C-terminal Ser-Lys-Leu (SKL) motif PTS1 (18,19) and N-terminal cleavable presequence PTS2 (20 -22). Other types of PTS have been postulated to exist but have not been identified yet. It is evident that pre-nsLTP contains a cleavable N-terminal 20-amino acid presequence resembling PTS2, as well as PTS1 tripeptide Ala-Lys-Leu (AKL), whereas SCPx possesses internally this 20-amino acid presequence of nsLTP and the C-terminal AKL. However, despite numerous biochemical findings such as those related to the lipid transfer activity, biogenesis of nsLTP and SCPx has not been well defined at the molecular and cellular levels.
To address the underlying mechanisms by which nsLTP and SCPx are transported into peroxisomes, we investigated biogenesis of these proteins at morphological, as well as biochemical, levels using several types of peroxisome biogenesis-defective cell mutants and the PTS1 receptor Pex5p. We herein report that import of nsLTP and SCPx was affected in pex5 mutant of Chinese hamster ovary (CHO) cells deficient in import of PTS1 and PTS2, whereas in PTS2 import-defective pex7 mutant both proteins were imported into peroxisomes as effi-ciently as in the wild-type CHO-K1. We also observed that nsLTP and SCPx bound to Pex5p. Therefore, these findings strongly suggest that nsLTP and SCPx are imported by the PTS1 translocation pathway. The function of the 20-amino acid presequence in intracellular transport of pre-nsLTP is also discussed.
Antibodies-Affinity-purified rabbit anti-rat nsLTP antibody was as described (2,4). Rat cDNA (TM-33/34) encoding the 59-kDa SCPx protein (6) was amplified by polymerase chain reaction using SCPx primers, a forward 5Ј-CCCGGGAACTCTCCGCGCCTGCGC-3Ј (underline, SmaI site) and a reverse 5Ј-GTCGACATCACTGGGACCCAG-GCC-3Ј (underline, SalI site). The polymerase chain reaction product was cloned into pGEM-T Easy vector (Promega) and cleaved with SmaI and SalI. The SmaI-SalI fragment encompassing an N-terminal part consisting of amino acid residues at 7-300 of 59-kDa SCPx was inserted into the SmaI-SalI site of pUEX3 (Amersham Pharmacia Biotech). The resultant plasmid for lacZ fusion protein was expressed in Escherichia coli. Antiserum specific for SCPx was raised in rabbit by conventional subcutaneous injection of ␤-galactosidase-SCPx fusion protein. Rabbit anti-green fluorescent protein (GFP) antibody was purchased from CLONTECH.
Morphological Analysis-nsLTP and SCPx in CHO cells were visualized by indirect immunofluorescence light microscopy using rabbit antibodies to nsLTP and SCPx, respectively, as described (27). We also used goat anti-rat catalase antibody (28) and rabbit anti-malate dehydrogenase antibody (29). Antigen⅐antibody complexes were detected under a Carl Zeiss Axioskop FL microscope by fluorescein isothiocyanate-labeled sheep anti-rabbit IgG antibody (Cappel), Texas Red-labeled sheep anti-rabbit IgG antibody (Cappel), or donkey anti-goat IgG antibody conjugated to rhodamine (Chemicon).
Subcellular Fractionation-Subcellular fractions of rat liver and CHO cells were prepared as described (19,30). A high speed supernatant, termed cytosolic, fraction was prepared by centrifuging a postnuclear supernatant fraction (PNS) for 40 min at 100,000 ϫ g (30). Isopycnic ultracentrifugation on a sucrose density gradient using PNS from CHO cells was done as described (30).
Expression of GFP Fusion Protein-DNA fragment for the N-terminal presequence of nsLTP was obtained by polymerase chain reaction using as a template cDNA encoding pre-nsLTP (6) in pUcD2 and a set of primers, a forward primer nsLTP-XbaI, CCTCTAGAAGAAT-GGGTTTTCCC, and a reverse nsLTP-BamHI, AATGGATCCCCTGCA-GAGCTGGT. After digesting with XbaI and BamHI, the resulting fragment was inserted into the XbaI-BamHI of pUcD2HygGFP(TT) (28), termed pUcD2Hyg⅐pre-GFP. pUcD2Hyg⅐pre-GFP-AKL was constructed as follows: PmaCI-KpnI fragment of pUcD2Hyg⅐GFP-AKL (28) was inserted into the PmaCI-KpnI site of pUcD2Hyg⅐pre-GFP. The resultant plasmids were assessed by nucleotide sequencing using a Dyeterminator DNA sequence kit (Applied Biosystems). Cells expressing GFP that had been grown on a cover glass were observed after cell fixation (28) under a Carl Zeiss Axioskop FL microscope using a number 17 filter.
Other Methods-Western blot analysis on polyvinylidene difluoride membrane (Bio-Rad) was performed using primary antibodies including rabbit anti-AOx antibody (26) and a second antibody, donkey antirabbit IgG antibody conjugated to horseradish peroxidase (Amersham Pharmacia Biotech). Antigen⅐antibody complexes were detected using ECL Western blotting detection reagent (Amersham Pharmacia Biotech).

RESULTS
Anti-SCPx Antibody-We raised antiserum against SCPx by immunizing a rabbit with SCPx (residues at 7-300; see Fig. 1A) fused to E. coli ␤-galactosidase. The antibody specifically recognized 59-and 46-kDa proteins in PNS, as well as in purified peroxisomes (Fig. 1B, lanes 1 and 2). The 46-kDa protein is most likely to be derived from 59-kDa SCPx by proteolytic conversion (1,14). Using anti-nsLTP antibody, 13-kDa, as well as 59-kDa, proteins were detected in PNS and peroxisomes (Fig. 1B, lanes 3 and 4), consistent with our earlier observation (2,4). In cytosolic fraction ϳ15 kDa of protein reacted with this antibody, in addition to 13-and 59-kDa polypeptides (lane 5), both presumably from broken peroxisomes during homogenization of liver (2). We interpreted these results to mean that nsLTP and 59-kDa SCPx cross-reactive to anti-nsLTP antibody are localized in peroxisomes, and pre-nsLTP is present in the cytosol. It is more likely that SCPx is partly processed to produce nsLTP and an N-terminal 46-kDa thiolase protein of SCPx (1,14). These data demonstrate that both antibodies are specific.  Morphological Analysis of nsLTP and SCPx-Wild-type and pex mutant CHO cells were stained using anti-nsLTP antibody. In CHO-K1 and pex7 mutant ZPG207 defective in PTS2 import (25), nsLTP was detected in a punctate staining pattern ( Fig.  2A, a and d), in a superimposable manner with that obtained using anti-catalase antibody (e and h), thereby suggesting that nsLTP was localized to peroxisomes. On the other hand, in pex2 Z65 and pex5 ZP105 mutants defective in PTS1 and PTS2 import (23,24,26,30), nsLTP appeared to be stained partly in a diffused pattern in the cytoplasm ( Fig. 2A, b and c), as was also the case for catalase (f and g) (24,26). A membraneassociated pattern, presumably representing SCPx (see Fig. 3), is also visible by staining with anti-nsLTP antibody in Z65 and ZP105 ( Fig. 2A, b and c). These results collectively suggest that nsLTP and SCPx cross-reacting with anti-nsLTP are imported by the PTS1, rather than PTS2, pathway. Moreover, in a recently isolated pex5 CHO cell mutant ZPG231 with a phenotype showing a defect solely in PTS2 import (31), punctate staining pattern was likewise discernible upon cell staining using anti-nsLTP antibody, in a superimposable manner with catalase-positive structures, hence confirming the import of nsLTP via the PTS1 pathway (data not shown).
These four types of CHO cells were stained using an antibody specific for SCPx. In CHO-K1, SCPx was detected in punctate structures that were superimposable upon those stained for catalase, hence indicating peroxisomal localization (Fig. 2B, a  and e). Similarly, SCPx was localized in particles in ZPG207, as was the case for catalase, strongly suggesting that SCPx is transported into peroxisomes (Fig. 2B, d and h). Similar punctate structures were detected when pex5 ZPG231 was stained with anti-SCPx antibody (data not shown). In contrast, in pex2 Z65 and pex5 ZP105 where catalase (Fig. 2B, f and g) and PTS1 proteins (data not shown) were in the cytoplasm, SCPx was partly, if not entirely, stained in a punctate pattern (b and c). In the same sets of cells that were permeabilized with 25 g/ml of digitonin, under which plasma membranes were selectively permeabilized and intraperoxisomal proteins were inaccessible to exogenous antibodies (28,32), SCPx was not detectable as a particle-associated form in any of four types of CHO cells including wild-type, pex7, pex2, and pex5 cells (Fig. 2B, i-l). Catalase was not accessible in digitonin-permeabilized CHO-K1 and pex7 ZPG207 (m and p), whereas catalase in pex2 Z65 and pex5 ZP105 cells was apparently visible in the cyto-  (36). B, protease protection assay. PNS from pex5 ZP105 was treated with proteinase K for 30 min on ice. The digestion was terminated with 1 mM phenylmethylsulfonyl fluoride. The reaction mixture was centrifuged to separate cytosolic (S) and organellar (P) fractions, which were analyzed by SDS-PAGE and immunoblot using antibodies to SCPx and Pex14p. PNS fractions analyzed were from 5 ϫ 10 5 cells each. PNS was mock-treated (lanes 1 and 2) or treated with 5, 10, and 20 g/ml of proteinase K in the absence (lanes [3][4][5] or presence (lane 6) of 1% Triton X-100. Only P fractions, each from proteinase K-treated PNS, were loaded. In lane 6, total reaction mixture was analyzed. C, isopycnic subcellular fractionation. PNS fractions from CHO-K1 (upper panel) and pex2 Z65 (lower panel) (2 ϫ 10 8 cells each) were fractionated by isopycnic ultracentrifugation on a sucrose density gradient. The gradient was collected into 19 tubes (0.5 ml each). An equal volume (15 l) of each fraction was analyzed by SDS-PAGE, followed by immunoblot with antisera against SCPx, Pex14p, and Pex5p. Results are presented in the direction of lower to higher density of sucrose, from left to right. plasm (n and o), slightly more vividly than the other cell types. We interpreted these data to mean that SCPx is associated with and possibly imported into endomembranes in pex2 and pex5 cell mutants.
Subcellular Fractionation Analysis-Intracellular localization of nsLTP and SCPx was also investigated by subcellular fractionation of CHO cells. When PNS of CHO-K1 cells was fractionated, nsLTP protein was found in both the cytosolic fraction and organelle pellets by immunoblotting with anti-nsLTP antibody (Fig. 3A, top panel, lanes 1 and 2). Subcellular localization of nsLTP and pre-nsLTP was distinct; pre-nsLTP was exclusively in the cytosolic fraction, whereas nsLTP was in the organelle fraction, apparently in peroxisomes (see Fig. 2). SCPx cross-reacting with anti-nsLTP antibody was exclusively in the organelle pellets (lane 2), presumably peroxisomes (see Fig. 2), as reported previously (2,33). A similar distribution pattern of pre-nsLTP, nsLTP, and SCPx was observed in pex7 ZPG207 (Fig. 3A, top panel, lanes 7 and 8). In contrast, only pre-nsLTP, but not mature nsLTP, was present and exclusively in the cytosolic fraction from pex2 Z65 and pex5 ZP105 (top lower panel, lanes 3-6), whereas SCPx was detected by anti-nsLTP antibodies in the organelle fraction in both mutants (top upper panel, lanes 4 and 6) as in CHO-K1 and ZPG207. Next, using SCPx-specific antibody, SCPx and 46-kDa protein, presumably the processed form of 59-kDa SCPx, were discernible in the organelle fraction of both CHO-K1 and ZPG207 (Fig. 3A,  middle panel, lanes 1, 2, 7, and 8), thereby implying that SCPx is imported into peroxisomes and partly converted to 46-kDa protein plus (presumably) nsLTP. In contrast, there was only the 59-kDa form of SCPx detectable in membrane pellet fractions of Z65 and ZP105 cells (middle panel, lanes 3-6), implying that there was no processing activity in these mutants.
AOx, a PTS1 protein, consists of 75-kDa A, 53-kDa B, and 22-kDa C polypeptide components and exists as a hetero-oligomer comprising A 2 , ABC, and B 2 C 2 (26,34). In normal cells, proteolytic conversion of AOx-A component to B and C polypeptides occurs in peroxisomes (26,35). A, B, and C components of AOx were evident in CHO-K1 and ZPG207, indicative of normal biogenesis of AOx (Fig. 3A, bottom panel, lanes 1, 2, 7, and 8), consistent with our earlier observation (25). In Z65 and ZP105, AOx-A, but not the B and C components, was detectable in cytosolic fraction at a lower level (bottom panel, lanes 3-6), indicating failure on the conversion of AOx, as in our earlier observation (27,28,36).
To determine the subcellular, as well as intraorganellar, localization of SCPx in pex mutants such as pex5 ZP105, we took another approach, a so-called protease protection assay. SCPx in PNS of ZP105 cells was resistant to the treatment with exogenously added proteinase K and was sedimentable (Fig.  3B, upper panel, lanes 1-5), whereas a peroxisomal membrane peroxin Pex14p (37) was sensitive to the digestion (lower panel), thereby strongly suggesting that SCPx was inside the organelles. Treatment with Triton X-100 prior to the protease digestion abolished the resistance, thereby confirming that SCPx was localized in membranous vesicles. Essentially the same results were obtained using pex2 Z65 cells (data not shown).
To confirm the findings described above with respect to the intracellular location of SCPx, PNS from CHO-K1 and a pex2 Z65 was fractionated by isopycnic ultracentrifugation on a sucrose density gradient. From CHO-K1, both 59-and 46-kDa forms of SCPx cosedimented with Pex14p, indicating that SCPx was localized and processed in peroxisomes (Fig. 3C,  upper panel). Pex5p mostly remained nearly at the top of gradient, but a small portion of Pex5p cosedimented with peroxisomes, indicating a largely cytoplasmic PTS1 receptor (30,38).
In Pex2p-defective Z65 (23), 59-kDa SCPx was sedimented and distributed from the middle to the bottom of the gradient (Fig.  3C, lower panel). Pex14p representing peroxisomal remnants (37) sedimented to a lighter density part of the gradient, with which a larger part of Pex5p cosedimented, consistent with our earlier observation (30). We interpreted these findings to mean that SCPx was localized to endomembranes, partly associated with peroxisomal remnants, in Z65 cells. SCPx appeared to be associated with mitochondria, as well (see Fig. 2B).
Binding to the PTS1 Receptor Pex5p-To investigate whether nsLTP and SCPx interact with the PTS1 receptor Pex5p, cell lysates from CHO-K1 were incubated with GST-Pex5pL fusion protein of wild-type Pex5pL, as well as ZP105type mutant GST-Pex5pL-G335E severely defective in PTS1binding (30). Bound protein fractions were analyzed by SDS-PAGE and immunoblot using specific antibodies. Only mature nsLTP, not pre-nsLTP, bound to wild-type Pex5pL, but not to GST and GST-Pex5pL-G335E, as probed with anti-nsLTP antibody, although pre-nsLTP and nsLTP were detectable in a nearly equal amount in PNS (Fig. 4, middle panel). This result suggests that C-terminal AKL is recognized by Pex5p in the case of nsLTP but not of pre-nsLTP. The presequence may affect the interaction of PTS1 of pre-nsLTP with Pex5p. SCPx was likewise bound only to normal Pex5pL (Fig. 4, top panel), where the 44-kDa processed form of SCPx lacking the C-terminal AKL was not detectable, even probed with SCPx-specific antibody (data not shown). Essentially the same results were obtained (data not shown) using GST-Pex5pS (30). As a PTS1 control, three components of AOx hetero-oligomer (35) (see Fig.  3A) were pulled down by normal GST-Pex5pL but barely by GST-Pex5pL-G335E, not by GST (Fig. 4, bottom panel, lanes  3-5), consistent with our previous observation (30).
Intracellular Localization of GFP Fusion Protein-To elucidate whether the presequence of pre-nsLTP and/or C-terminal AKL function as a sufficient PTS, we constructed two types of GFP-fusion proteins, one fused N-terminally to the 25-amino acid residues including the 20-amino acid presequence plus 5 downstream-residues of pre-nsLTP and C-terminally linked to AKL (termed pre-GFP-AKL), and the other fused only to the presequence at the N terminus (pre-GFP) (Fig. 5A). These chimeras were transiently expressed in CHO-K1, pex2 Z65, pex5 ZP105, and pex7 ZPG207 cells. In CHO-K1, pre-GFP-AKL was detected in particulates and colocalized with catalase, hence indicating that pre-GFP-AKL was targeted to peroxisomes (Fig. 5B, a and e). pre-GFP-AKL was likewise in particles, in catalase-positive but PTS2-negative "peroxisomes" in ZPG207 (25) (Fig. 5B, d and h). In contrast, pre-GFP-AKL appeared to remain partly in the cytoplasm, like catalase, in pex2 and pex5 mutants (Fig. 5B, b, c, f, and g). In these mutants, pre-GFP-AKL was also discernible in a membrane-associated form as seen for nsLTP and SCPx (see Fig. 2, A and B), which was superimposable on that assessed by cell staining using antibody to malate dehydrogenase, a mitochondrial marker enzyme (Fig. 5C). Accordingly, the data strongly suggested that pre-GFP-AKL was mostly localized to mitochondria in pex2 and pex5 mutants. Taken together, pre-GFP-AKL was transported to peroxisomes in a PTS1-dependent and Pex5pmediated pathway in normal and PTS2-import-defective pex7 cells. Moreover, we likewise investigated whether the presequence of pre-nsLTP functions as a topogenic signal. pre-GFP was detected in a superimposable manner with malate dehydrogenase-positive particles, presumably mitochondria, in both wild-type and pex2 CHO cells (Fig. 5D), suggesting that the 20-amino acid presequence functions per se as a mitochondrial targeting sequence.
Interaction of pre-GFP-AKL with Pex5p-The expression level of pre-GFP-AKL and pre-GFP was also verified by immunoblot of the cell lysates. Both proteins were detected with an expected size and at a similar level in wild-type CHO-K1 and a pex2 Z65 (Fig. 6A, lanes 2, 3, 5, and 6). As a control GFP-AKL (28) was also expressed, which was detectable in both types of cells (lanes 1 and 4), and was targeted to peroxisomes in CHO-K1 but remained in the cytoplasm in Z65 (data not shown), in good agreement with our earlier observation (28). The distinct mobility in SDS-PAGE of these three GFP fusion proteins suggested that the presequence of pre-GFP-AKL and pre-GFP was not cleaved off, even after imported to peroxisomes and mitochondria, respectively. The processing site of these chimera may not be recognized by a potential processing protease. Next, we investigated whether pre-GFP-AKL interacts with Pex5p. Control GFP-AKL bound to GST-Pex5pL, but not to GST (Fig. 6B, lanes 1-3), consistent with peroxisomal localization (data not shown) (28). pre-GFP-AKL, but not pre-GFP, from CHO-K1 cell-lysates was likewise pulled down by GST-Pex5pL, not by GST (Fig. 6B, lanes 4 -9), thereby indicating that, in contrast to pre-nsLTP, the PTS1 of pre-GFP-AKL was readily recognized by Pex5p, in agreement with morphological findings (see Fig. 5B). DISCUSSION pre-nsLTP can be classified as a unique protein among a number of peroxisomal proteins with respect to its topogenic sequence. The cleavable presequence locates at the N terminus, similar to PTS2 of several peroxisomal enzymes such as 3-ketoacyl-CoA thiolase (20,21), and the C terminus carries a typical PTS1 tripeptide motif, AKL. Moreover, the amino acid sequence, -AAPT2SS-(-, continuing to adjacent residues), near the processing site of pre-nsLTP to form mature nsLTP resembles the sequence of PTS2-thiolase, -AAPC2SS-(-, continuing to adjacent residues), implying that pre-nsLTP and PTS2 proteins may share a potential, yet unidentified processing protease present in peroxisomes. The nsLTP presequence is were transfected with each cDNA construct indicated at the top. Cell lysates were analyzed by SDS-PAGE and Western blot using anti-GFP antibody. B, binding assay of GFP-AKL, pre-GFP, and pre-GFP-AKL to Pex5p was done using cell-lysates of GFP-AKL-, pre-nsLTP-, and pre-GFP-AKL-expressing CHO-K1 (1 ϫ 10 6 cells each), as in Fig. 4. Components added in the GST pull-down assay were indicated at the top. GFP fusion protein was detected by Western blot using anti-GFP antibody. One-tenth aliquot of each input was loaded in lanes 1, 4, and 5. relatively positive-charged like PTS2, but it does not contain the consensus PTS2 sequence -(R/K)(L/V/I)X 5 (H/Q)(L/A)-. This implies that the presequence may be a novel-type PTS. We investigated in the present work which of two potential PTS sequences destines pre-nsLTP to its final intracellular location in vivo. With the use of several CHO pex mutants, import of pre-nsLTP was verified. pre-nsLTP was affected in import in pex2 and pex5 mutants with a phenotype showing import defect of both PTS1 and PTS2 proteins, where pre-nsLTP remained in the cytosol. Other groups of investigators reported that nsLTP is likely to be degraded or barely detectable in ZR82, a pex2 CHO mutant (33), as well as in fibroblasts from a patient with Zellweger syndrome (12), although pre-nsLTP was not fully described. In pex7 cell mutant impaired in PTS2 import, pre-nsLTP was translocated as efficiently as in the wild-type CHO-K1. Transport of pre-GFP-AKL was likewise re-established in wild-type and pex7 mutant CHO cells, consistent with the morphological phenotype as assessed for pre-nsLTP. Taken together, we conclude that pre-nsLTP is imported by the Pex5p-mediated PTS1 pathway.
To our surprise, pre-GFP-AKL was targeted to mitochondria when expressed in pex2 and pex5 mutants. These results imply that pre-GFP-AKL behaves differently from pre-nsLTP; PTS1 of pre-nsLTP could not be recognized by Pex5p in vitro, whereas AKL of pre-GFP-AKL readily interacted with Pex5p. Despite such biochemical properties, pre-nsLTP is transported to peroxisomes by a Pex5p-dependent pathway in vivo. Moreover, in mutant cells absent from Pex5p, pre-nsLTP remains in the cytoplasm, whereas pre-GFP-AKL is readily transported to mitochondria at least in an over-expression system. The mitochondrial topogenic activity appears to be interfered with in the presequence of pre-nsLTP. Accordingly, the PTS1 and presequence of pre-nsLTP are most likely to be mutually regulated by unknown mechanisms. One possibility includes modulation of PTS1 of pre-nsLTP by direct steric effect of the presequence or that mediated by a cytosolic factor. Alternatively, PTS1 of pre-nsLTP may be partly modified, whereby a portion of total pre-nsLTP no longer interacts with Pex5p and remains in the cytoplasm. In the case of pre-GFP-AKL, the tripeptide AKL is more likely to be exposed to the surface of the molecule, thereby being readily recognized by Pex5p. The presequence may be unable to modulate the PTS1, possibly because of the configuration of this GFP chimera. Furthermore, we identified the presequence as a mitochondrial targeting signal, based on the observation that pre-GFP was targeted to mitochondria in normal and mutant CHO cells.
It is noteworthy that the N-terminal sequence encompassing the 20-amino acid presequence plus Ser at position 21 was predicted as a mitochondrial targeting sequence (44% probability) using a recognition program, PSORT II, minimally requiring 21 amino acid residues to be analyzed, where the prediction also included cytoplasmic (26%) and peroxisomal (4%) localization. Secondary structure analysis of the 20-amino acid presequence using a Chou-Fasman program predicted an ␣-helix between residues at 3-10 and ␤-sheet structure between the residues 10 -17. These characteristics may contribute as a mitochondrial import signal. Therefore, it is tempting to infer that the presequence may interact with the C-terminal PTS1 and regulate peroxisomal transport of pre-nsLTP, possibly in a concerted manner with other cytosolic factors such as chaperones. How the PTS1 and mitochondrial targeting signal-like presequence are regulated in biogenesis of nsLTP in vivo remains to be defined.
pre-nsLTP is rather stably present in the cytosol as assessed by subcellular fractionation study and morphological analysis. The modulated presequence may function in vivo as a signal for pre-nsLTP to be a cytosolic protein. If synthesized pre-nsLTP remains in the cytoplasm, nsLTP in peroxisomes may be all derived from SCPx instead of pre-nsLTP by proteolytic conversion. Although we can not exclude this possibility, this may be less likely based on the findings in pulse-chase experiments, where processing of pre-nsLTP to nsLTP was noted (2,12). The physiological consequences of pre-nsLTP in the cytosol, nearly at an equal level as nsLTP in peroxisomes in normal CHO cells, also remain to be investigated. The level of cytosolic pre-nsLTP may vary, depending on the cell types from various tissues and organs that are involved in lipid metabolism and steroid biosynthesis (1,14). It is also notable that a role of nsLTP and SCPx in peroxisomal ␣-oxidation of phytanic acid was implicated from the study using nsLTP/SCPx gene null mice (39).
SCPx is exclusively localized in peroxisomes where it is converted to the 46-kDa fragment and nsLTP. SCPx is also transported with use of C-terminal AKL via a Pex5p-mediated PTS1 pathway. In PEX5-deficient mutant cells, SCPx is associated with endomembranes such as peroxisomal remnants and mitochondria. It is noteworthy that not only the 59-kDa form but also the 46-kDa fragment are enzymatically active as a branched-chain fatty acid thiolase (15,17,40), as revealed using specimens from patients with peroxisome-defective disorders such as Zellweger syndrome. Localization of SCPx on endomembranes may be required for such enzymatic activity of the unprocessed form. However, its underlying mechanisms are presently unclear. Exact cleavage sites initially at position 404 -405 or 424 -425 or both (see Fig. 1A), as well as physiological significance of this intraperoxisomal processing of SCPx, are not well understood at present. Interestingly, P-44 with thiolase activity and ZK892.2 apparently corresponding to the 46-kDa form of SCPx and nsLTP in mammals, respectively, have both been identified in Caenorhabditis elegans (41), whereas an orthologue of 59-kDa SCPx has not been detected. Whether accumulation of SCPx in protease-resistant form in the membrane fraction in pex5 and pex2 mutants, as noted in this work, is physiologically relevant or related to clinical phenotypes of peroxisome biogenesis disorders also remains to be defined.