In Vitro Transport of Membrane Proteins to Peroxisomes by Shuttling Receptor Pex19p*

The peroxin Pex19p comprising 299 amino acids functions in peroxisomal membrane assembly. We here developed a cell-free system for transport of membrane proteins to peroxisomes. Pex19p interacts with multiple membrane peroxins, including other membrane biogenesis peroxins, Pex16p and Pex26p, involved in matrix protein import. Cell-free synthesized, 35S-labeled Pex19p was targeted to subcellular fractions containing peroxisomes from Chinese hamster ovary-K1 cells as well as peroxisomes isolated from rat liver in an ATP-dependent manner. Such translocation was also reproduced with in vitro synthesized 35S-Pex16p with two transmembrane segments and C-tail anchor-type 35S-Pex26p, upon incubation with 35S-Pex19p in the reaction mixtures containing isolated peroxisomes. The transported 35S-Pex16p and 35S-Pex26p were integrated into membranes as assessed by the sodium carbonate extraction method. Peroxisome-associated and partly Na2CO3-resistant 35S-Pex19p was released to the cytosolic fraction upon incubation in the absence of ATP, whereas 35S-Pex16p and 35S-Pex26p remained in the membranes. Furthermore, not only 35S-Pex19p but also 35S-Pex19p complexes each with 35S-Pex16p and 35S-Pex26p were bound to 35S-Pex3p in vitro. Together, these results strongly suggested that Pex19p translocates the membrane peroxins from the cytosol to peroxisomes in an ATP- and Pex3p-dependent manner and then shuttles back to the cytosol.

Molecular mechanisms of peroxisome biogenesis have been investigated by mostly focusing on the attempt to elucidate how matrix proteins are imported. In contrast, those involving peroxisomal import of membrane proteins and membrane vesicle assembly remain elusive (1)(2)(3). In peroxisome-deficient mutant cell lines, including pex3, pex16, and pex19 Chinese hamster ovary (CHO) 2 cell mutants and fibroblasts from patients with peroxisome biogenesis disorders, peroxisome membrane assembly is severely impaired; hence, membrane structures, socalled peroxisomal ghosts or membrane remnants, are morphologically and biochemically undetectable (4 -10). Peroxisome membrane assem-bly is initiated upon expression of three peroxins, Pex3p, Pex16p, and Pex19p, in their respective cell mutants. Several distinct steps following the initial stage include import of membrane and matrix proteins, growth, and division or fusion of matured peroxisomes.
As a further step toward understanding molecular mechanisms underlying the Pex19p functions, we here developed an in vitro assay system for Pex19p targeting to peroxisomes. Moreover, we have provided several lines of evidence that Pex19p interacts with newly synthesized PMPs, directs them to peroxisome membranes, and returns to the cytosol.

EXPERIMENTAL PROCEDURES
Expression and Purification of Recombinant Pex19p-Expression plasmid encoding human Pex19p fused to glutathione S-transferase (GST-Pex19p) was constructed as follows. PCR was performed using pUcD2Hyg⅐PEX19 (4) as a template with forward primer PEX19F/ BamHI and reverse primer PEX19R/BamHI. A BamHI fragment of PCR products was subcloned into the BamHI site in the pGEX6P-1 vector. GST-Pex19p was expressed in Escherichia coli DH5a and purified from cell lysates as described (22). Fusion protein-bound glutathione-Sepharose beads were mixed with 40 units of PreScission protease in cleavage buffer consisting of 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM dithiothreitol, 3 mg/ml pepstatin A, and 1 mM phenylmethylsulfonyl fluoride. The mixture was rotated for 3 h at 4°C. Pex19p was eluted by centrifugation at 500 ϫ g for 5 min.
Cell-free Synthesis of Pex Proteins-Tandem double hemagglutinin A (HA 2 )-tagged human PEX19, termed HA 2 -PEX19, was amplified by PCR using pUcD2Hyg⅐PEX19 as a template and a respective set of primers, a forward N2F/SalI (ATCAGTCGACCGCCGCCGCTGAGGAA-* This work was supported in part by a SORST grant from the Science and Technology Agency of Japan (to Y. F.), grants-in-aid for scientific research (to Y .F.) from National Project on Protein Structural and Functional Analyses (to Y. F.), and grants from The 21st Century COE Program from The Ministry of Education, Culture, Sports, Science, and Technology of Japan, the Uehara Memorial Foundation (to Y.F.), Japan Foundation for Applied Enzymology, and Takeda Science Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 35 S-cysteine (Amersham Biosciences) as labels. Translation mixture was centrifuged at 20,000 ϫ g for 2 min at 4°C, and the supernatant fraction was used for assays.
In Vitro Membrane Targeting Assays-Pex19p membrane-targeting assays were performed at 26°C for 60 min in 250 l of the homogenizing buffer (see above) or in the import buffer: 10 mM Hepes-KOH, pH 7.5/0.25 M sucrose/5 mM methionine/50 mM KCl/3 mM MgCl 2 /1 mM ATP/50 g/ml creatine kinase/10 mM creatine phosphate, using PNS from CHO-K1 cells (3 ϫ 10 5 ) or 6 g of rat liver peroxisomes and 0.2 l of reticulocyte-lysate translation mixture containing 35 S-HA 2 -Pex19p. Targeting reaction was also carried out under an ATP-depleting condition, in the presence of 10 mM AMP-PNP in place of 1 mM ATP. The assay mixtures were separated into high speed supernatant and organelle fractions by centrifugation at 100,000 ϫ g for 30 min. An aliquot of the organelle fraction in the import buffer was treated with 0.1 M Na 2 CO 3 on ice for 30 min, and soluble and integral membrane proteins were separated by ultracentrifugation at 100,000 ϫ g for 30 min (28). All the fractions were analyzed by SDS-PAGE using 12% gels. 35 S proteins were detected by a Fluoro Imaging Analyzer FLA-5000 (Fuji Film, Tokyo, Japan).
For Pex19 export assays, aliquots of the organelle pellet fractions obtained in the Pex19p membrane-targeting assays described above were resuspended in the homogenizing buffer, import buffer with or without 10 g of recombinant Pex19p (rPex19p), or ATP-depleted import buffer in the presence of cytosolic fraction from CHO-K1 (3 ϫ 10 5 cells). These assay mixtures were incubated at 26°C for 30 min and separated into supernatant and membrane fractions by centrifugation at 100,000 ϫ g for 30 min.
Pex19p PMP-recruiting assays were done at 26°C for 60 min in 250 l of import buffer containing 6 g of rat liver peroxisomes and 0.2 l of 35 S-Pex16p or 35 S-Pex26p with 35 S-HA 2 -Pex19p. The assay mixtures were separated into high speed supernatant and organelle fractions by centrifugation at 100,000 ϫ g for 30 min. The organelle pellet was resuspended in the homogenization buffer or the import buffer with or without 10 g of rPex19p.
Binding of Cargo-loaded Pex19p to Pex3p-35 S-HA 2 -Pex19p and 35 S-FLAG-Pex16p or 35 S-FLAG-Pex26p were separately translated using a coupled transcription/translation system as described above. Respective sets were incubated at 4°C for 2 h with 35 S-FLAG-Pex3p-EGFP that had been likewise synthesized. 35 S-FLAG-Pex3p-EGFP was immuno-precipitated using anti-GFP antibody (Santa Cruz Biotechnology). Immunoprecipitates were analyzed by SDS-PAGE and Fluoro Imaging Analyzer as described above. 35 S-HA 2 -Pex19p-bound 35 S-FLAG-Pex26p and free 35 S-FLAG-Pex26p, if any, were immunoprecipitated using anti-FLAG immunoglobulin G (IgG)-conjugated agarose. Bound proteins were eluted by addition of a large excess of FLAG peptides. The eluate was incubated at 26°C for 1 h with 35 S-FLAG-Pex3p-EGFP. Immunocomplexes of 35 S-FLAG-Pex3p-EGFP isolated by immunoprecipitation with anti-GFP antibody were analyzed as above.
Other Methods-Western blot analysis was performed using electrophoretically transferred samples on polyvinylidene difluoride membrane (Bio-Rad) with primary antibodies and a second antibody, donkey anti-rabbit IgG antibody conjugated to horseradish peroxidase (Amersham Biosciences). Antigen-antibody complexes were visualized with ECL-plus Western blotting detection reagent (Amersham Biosciences) and a Can-Get-Signal Enhancer (Toyobo, Osaka, Japan).

RESULTS
In Vitro Translocation of Pex19p to Peroxisomes-To investigate whether Pex19p targets peroxisome membranes from the cytosol, we attempted to establish a cell-free translocation system of Pex19p. In vitro synthesized 35 S-Pex19p was incubated at 26°C for 1 h in the import assay buffer with PNS fraction from wild-type CHO-K1, pex2 mutant Z65 (31), and pex3 ZPG208 (6) cells and with rat liver peroxisomes. The reaction mixture was separated into organelle and cytosol fractions and analyzed by SDS-PAGE. Pex14p, a peroxisomal membrane protein, was detected in organelle fraction in respective sets of the incubation mixtures, indicating the adequate separation of soluble and organelle fractions (Fig. 1A, left panels). Two types of 35 S-Pex19p, authentic and farnesylated forms (4), were detected in both cytosol and organelle fractions (Fig. 1A, lanes 4 and 5), presumably in peroxisomes. To verify this result, we expressed HA 2 -Pex19p in CHO-K1 cells. HA 2 -Pex19p was discernible in particles superimposable on PMP70-stained peroxisomes and in the cytosol (Fig. 1B), thereby suggesting the transport of 35 S-HA 2 -Pex19p to peroxisomes. 35 S-Pex19p was likewise detected, but in lesser amounts, in membrane fraction when incubated with PNS fraction of peroxisomal membrane remnant-positive pex2 Z65, whereas it was barely associated with the membranes of pex3 ZPG208 defective in peroxisomal membrane biogenesis (lower panels, lanes 4 and 5), consistent with morphological findings. 3 In immunoblot, Pex14p was detected in membrane fraction from both types of CHO cell mutants, consistent with our earlier report (6,32), suggesting the proper separation in the subcellular fractionation (Fig. 1A, lower panels). 35 S-Pex19p was also detectable at a level similar to the result with Z65 in the organelle fraction of other CHO cell mutants defective in matrix protein import, including pex1 ZP107 (33), pex6 ZP164 (5), pex13 ZP128 (34), and pex26 ZP167 (27) (data not shown). 35 S-Pex19p associated with the organelles of CHO-K1 was in a significantly higher amount in the assays done in the import assay buffer containing 1 mM ATP and the ATP regeneration system as compared with that incubated in the cell-homogenizing buffer with no addition of ATP (right panel, compare lanes 3 and 5). However, the level of 35 S-Pex19p in organelle fraction was ϳ50% reduced in the presence of 10 mM AMP-PNP in place of ATP (right panel, compare lane 7 with 5). 35 S-Pex19p was indeed transported to peroxisomes purified from rat liver, apparently with a much higher amount of 35 S-Pex19p in peroxisome fraction (Fig. 1A, lanes 8 and 9), presumably because ϳ50-fold peroxisomes were used as compared with the assay using CHO-K1-derived PNS, based on calculation with catalase-specific activity (23,35). However, the farnesylated 35 S-Pex19p was less detected than that in the assays using PNS fraction in which the farnesylating activity was conceivably higher. Next, 35 S-Pex19p in the organelle fraction (Fig. 1C, upper panel, lane 3, equivalent to lane 5 in Fig. 1A, upper panel) was treated with 0.1 M Na 2 CO 3 (28). 35 (Fig. 1C, lower panel), hence confirming the finding with 35 S-Pex19p in a cell-free system.
To investigate whether Pex19p could be exported from peroxisomes, we resuspended Pex19p-targeted peroxisomes (Fig. 2, upper panel, lane  1) in the homogenizing buffer or the import buffer supplemented with cytosol fraction from CHO-K1 in the absence or presence of AMP-PNP or rPex19p. After a 30-min incubation at 26°C, the reaction mixtures were separated to organelle and cytosol fractions and analyzed as in Fig.  1A. Pex14p was detected in all of the peroxisome fractions as in the input (upper panel), indicating that peroxisomes in the respective incubation mixtures were sedimented into organelle pellet fractions. 35 S-Pex19p in each of the fractions was quantitated (Fig. 2, lower panel). In the homogenizing buffer, 35 S-Pex19p was dissociated, up to two thirds, from the organelles into cytosolic fraction (Fig. 2, upper and lower panels, lanes 2 and 3), while about half of the 35 S-Pex19p was released in the import buffer containing ATP (lanes 4 and 5). Addition of AMP-PNP instead of ATP apparently reduced the level of 35 S-Pex19p in the organelle fraction (lane 7), consistent with the finding in the in vitro targeting assay (see Fig. 1, lane 7), hence implying that a retargeting step  5), and ATP-depleted import buffer (lanes 6 and 7) or with rat liver peroxisomes (Ps, 6 g) in import buffer (lanes 8 and 9). PNS fractions from CHO pex2 Z65 and pex3 ZPG208 mutant cells were likewise used for the assay (lower panels in lanes 4 and 5). Reaction mixtures were separated to soluble (S) and organelle (P) fractions and were analyzed by SDS-PAGE. Separation of S and P fractions was assessed by immunoblotting Pex14p, a peroxisomal integral membrane protein.  lanes 4 and 5). 35 S-Pex19p was detected as in panel A. Soluble and membrane protein markers, AOx and Pex14p, were detected by immunoblotting. Only AOx-B chain is shown. Lower panel, PNS fraction from CHO-K1 cells expressing HA 2 -Pex19p was likewise separated (lanes 6 and 7). Soluble and membrane fractions (lanes 8 and 9) were isolated by the Na 2 CO 3 treatment of organelle fraction (lane 7). HA 2 -Pex19p was detected by immunoblotting with anti-HA antibody.  Fig. 1A as well as one containing recombinant Pex19p (rPex19p, 40 g/ml) in the presence of CHO-K1-derived cytosol, incubated at 26°C for 30 min, and then separated to cytosolic and organelle fractions. The fractions were analyzed as in Fig. 1A. 35 S-Pex19p in each fraction was quantitated as in Fig. 1A and was represented by taking the input as 1. rPex19p was detected by immunoblot at an exposure under which only exogenous rPex19p added at a much higher level was detectable. of 35 S-Pex19p was affected. Together, these results suggested that cytosolic Pex19p was targeted to and associated with peroxisomes, translocated back to the cytosol fraction, and retargeted to peroxisomes. Moreover, 35 S-Pex19p-associated peroxisomes were likewise incubated in the presence of rPex19p (Fig. 2, upper and lower panels, lanes 8 and 9). Significantly, a reduced amount of 35 S-Pex19p was detected in the organelle fraction (compare lanes 5 and 9), with a concomitant increase of 35 S-Pex19p in the cytosol fraction (lanes 4 and 8). We interpreted this result to mean that 35 S-Pex19p was released from peroxisomes but was not efficiently retargeted to peroxisome membranes because of a larger excess of unlabeled Pex19p, in good agreement with the above observation (lanes 2-7). Moreover, the exogenously added rPex19p was indeed translocated and bound to peroxisome membranes (Fig. 2, upper panel,  inset, lanes 8 and 9).

Pex19p-dependent PMP Import to Peroxisomes and Pex19p
Recycling-We investigated topogenesis of PMPs by taking as a model PMP two integral membrane peroxins, Pex16p with two transmembrane segments (7) and C-tail anchor-type Pex26p (27). In vitro synthesized 35 S-Pex16p and 35 S-Pex26p were separately incubated with rat liver peroxisomes in the presence or absence of 35 S-Pex19p. After the reaction, peroxisome fractions were analyzed by SDS-PAGE (Fig. 3A). Peroxisomes used were recovered in the organelle pellet fractions as assessed with Pex14p (left panels). Both 35 S-Pex16p and 35 S-Pex26p were recovered in the organelle fractions, at 3-and 10-fold higher levels upon the reaction in the presence of 35 S- Pex19 (right panels, compare lanes 3 and 4). Moreover, 35 S-Pex16p and 35 S-Pex26p in the organelle fraction were resistant to the sodium-carbonate extraction where Pex14p and AOx-B, markers  Fig. 1 and, respectively, represented in an arbitrary unit. Note that 35 S-Pex19p was released to the cytosol fraction.
for membrane and soluble proteins of peroxisomes, were detected in the membrane and soluble fractions, respectively (Fig. 3A, upper and  lower left panels, lanes 5 and 6), hence indicating that 35 S-Pex16p and 35 S-Pex26p were integrated to peroxisome membranes. Interestingly, about half or less of 35 S-Pex19p was also resistant to the alkaline extraction (lanes 5 and 6), implying its apparently membraneintegrated form. Next, 35 S-Pex16p-and 35 S-Pex19p-targeted peroxisomes (Fig. 3B, upper panel, lane 1) were resuspended in the import assay buffer containing CHO-K1-derived cytosol and were incubated at 26°C for 30 min as in Fig. 2. The input peroxisomes were likewise recovered in organelle pellet fractions as verified by Pex14p immunoblot (Fig. 3B). About one half of 35 S-Pex19p was released from peroxisomes, whereas 35 S-Pex16p mostly remained in peroxisome membranes (Fig. 3B, upper left panel, lanes 2 and 3). Addition of a larger excess of rPex19p to the reaction mixture significantly reduced the amount of 35 (Fig. 3B, lower right panel, lanes  1-5; lower right panel) were distinct. Taken together, these results strongly suggested that PMPs such as the membrane peroxins of two different types are recruited by Pex19p in the cytosol, transported to peroxisomes, and integrated into peroxisome membranes. Upon or after integration of PMPs, Pex19p is likely released to the cytosol and recycled for another round of recruiting and targeting of PMPs.

DISCUSSION
In earlier studies by several groups including ours (4,12,13), Pex19p was shown to be essential at early steps of peroxisomal membrane biogenesis such as that involving the recognition of multiple PMPs (14,16 -18,20). 4 Defect of Pex19p function indeed impairs peroxisomal mem-4 Y. Matsuzono, and Y. Fujiki, to be reported elsewhere. brane assembly occurring prior to the import of peroxisomal matrix enzymes in yeast (12,13) and mammalian cells such as CHO cells and human fibroblasts (4,5,14).
In the present work, we established a cell-free Pex19p translocation system. Moreover, in this system, PMPs including two membranespanning Pex16p and a C-tail-anchored-type Pex26p were transported in a Pex19p-dependent manner and integrated to peroxisome membranes. Furthermore, Pex19p complexes with PMPs such as Pex16p and Pex26p bound to Pex3p in vitro, indicating that the ternary complexes were formed. The profiles of targeting to peroxisomes of Pex19p and its truncated variants were superimposable to those of their binding to Pex3p, as assessed by co-immunoprecipitation and/or yeast two-hybrid assays. 4 By taking this observation into consideration, we interpreted the findings on the ternary complex formation to mean that Pex19p-PMP complexes were transported to their docking site Pex3p on peroxisomes, in good agreement with the recent suggestion that Pex3p is required for Pex19p docking to peroxisomes (36). 3 With regard to the physiological role, Pex19p has been thought to function as a PMP chaperone (17,37) and an import receptor that mediates the transport of PMPs to peroxisomes (15,38) by interacting with their peroxisome membrane-targeting signal (19,20,39), except for Pex3p (36). Fang et al. (36) recently reported that Pex3p was required for recruiting and docking of Pex19p-PMP complexes to peroxisomes. Our findings in the present work favor such models.
The translocation to peroxisomes of Pex19p requires ATP and is affected by AMP-PNP. Whether any factor is involved in the hydrolysis or binding of ATP remains to be defined. Meanwhile, the AAA ATPase family peroxins, Pex1p and Pex6p, essential for the import of matrix proteins, may be candidates for such factors (27,40). It is noteworthy, however, that Pex19p was targeted to peroxisomal membrane remnants in pex1 and pex6 cell mutants (data not shown). Therefore, the AAA ATPase peroxins are less likely candidates. In contrast, ATP was dispensable for the release of Pex19p from peroxisomes. Again, the mechanistic insight to this important issue needs to be addressed. Likewise to be elucidated are physiological role(s) of the apparent integration of Pex19p to peroxisome membranes, consistent with the findings that Pex19p is indeed partly associated with peroxisomes (4,12,14,20). Such issues may be addressed by making use of the in vitro Pex19p transport system reported here in combination with other assay systems such as the in vivo analyses.
We were not able to readily distinguish the farnesylated and nonfarnesylated forms of Pex19p in the activities in translocating, transporting PMPs including at least Pex16p and Pex26p examined in this report, and returning to the cytosol. Incidentally, both forms apparently behaved in an indistinguishable manner. The farnesylation may not play a major role for peroxisome-restoring activity of Pex19p. Rather, it may promote a proper conformational change of the C-terminal domain of Pex19p by which the function of Pex19p is regulated and may not be involved in the direct interaction with PMPs and peroxisome restoration activity. Together, physiological consequences for the farnesylation of Pex19p may include 1) to regulate Pex19p in binding to and releasing from cargo PMPs and 2) to control the steps of binding of Pex19p-PMPs complexes to peroxisome membranes and releasing the unloaded Pex19p from peroxisome membranes. It has been suggested that the farnesylation is important for the Pex19p binding to PMPs (39,41). Such issues may also be addressed by investigating biogenesis of other PMPs, including membrane peroxins.
It is conceivable that molecular dynamism of Pex19p resembles Pex5p, a shuttling receptor for proteins with the peroxisome targeting signal type 1 (22,29,42). In this report, we showed by the in vitro membrane-targeting assay that Pex19p functions as a cytosolic recycling receptor in recruiting PMPs, at least Pex16p and Pex26p. Collectively, our present work is the first report providing in vitro evidence that Pex19p is the shuttling receptor for the membrane peroxins. Further investigation, by making use of in vivo and in vitro systems involving dissection of Pex19p and Pex3p functions, should shed light on the mechanisms underlying the peroxisome assembly.