Functional domain mapping of peroxin Pex19p: interaction with Pex3p is essential for function and translocation.

The peroxin Pex19p functions in peroxisomal membrane assembly. Here we mapped functional domains of human Pex19p comprising 299 amino acids. Pex19p mutants deleted in the C-terminal CAAx farnesylation motif, the C-terminal 38 amino acid residues and the N-terminal 11 residues, maintained peroxisome-restoring activity in pex19 cells. The sequence 12-261 was essential for re-establishing peroxisome activity. Pex19p was partly localized to peroxisomes but mostly localized in the cytosol. Pex19p interacted with multiple membrane proteins, including the other two membrane biogenesis peroxins, Pex3p and Pex16p, those involved in matrix protein import such as Pex14p, Pex13p, Pex10p, and Pex26p, peroxisome morphogenesis factor Pex11pβ, and a PMP70 peroxisome-targeting signal region at residues 1-123. In yeast two-hybrid assays, Pex10p and Pex11pβ interacted only with full-length Pex19p. Of various truncated Pex19p variants active in translocating to peroxisomes, the mutants with the shortest sequence (residues 12-73 and 40-131) were localized to peroxisomes and competent in binding to Pex3p. Furthermore, membrane peroxins were initially discernible in a cytosolic staining pattern in pex19 cells only when co-expressed with Pex19p and were then localized to peroxisomes in a temporally differentiated manner. Pex19p probably functions as a chaperone for membrane proteins and transports them to peroxisomes by anchoring to Pex3p using residues 12-73 and 40-131.


Introduction
Peroxisomal proteins are encoded by nuclear genes, translated on free polyribosomes in the cytosol, and imported to peroxisomes (Lazarow and Fujiki, 1985). The molecular mechanisms of peroxisomal import of matrix proteins are well understood, whereas those involving membrane protein transport and membrane vesicle assembly remain elusive (Fujiki, 2000;Lazarow, 2003;. In peroxisome-deficient mutant cell lines, including pex3, pex16 and pex19 Chinese hamster ovary (CHO) cell mutants and fibroblasts from patients with peroxisome biogenesis disorders (PBD) of complementation groups (CGs) 12, 9, and 14, respectively, peroxisome membrane assembly is severely impaired, hence membrane structures such as the so-called 'peroxisomal ghosts' or membrane remnants are morphologically and biochemically undetectable (Ghaedi et al., 2000b;Honsho et al., 2002;Honsho et al., 1998;Jones et al., 2001;Kinoshita et al., 1998;Matsuzono et al., 1999;Muntau et al., 2000;South and Gould, 1999). It is of interest to note that membrane particles are observed in pex19 mutant cells of Pichia pastoris (Snyder et al., 1999), Yarrowia lipolytica (Lambkin and Rachubinski, 2001) and Y. lipolytica pex16 cells (Eitzen et al., 1997). Peroxisome membrane assembly is initiated by at least three peroxins, Pex3p, Pex16p and Pex19p, followed by several distinct steps, including the import of membrane and matrix proteins as well as growth and division of peroxisomes. We earlier cloned human PEX19 cDNA encoding the 299 amino acid, hydrophilic peroxin Pex19p with the farnesylation motif CAAx at the C-terminus, by functional complementation strategy using a mutant CHO cell line, ZP119, defective in the import of both matrix and membrane proteins (Kinoshita et al., 1998;Matsuzono et al., 1999). Pex19p is localized, mostly in the cytosol and partly associated with peroxisomes (Goette et al., 1998;James et al., 1994;Matsuzono et al., 1999;Snyder et al., 1999). Pex19p binds multiple peroxisomal integral membrane proteins (PMPs), including several peroxins such as Pex3p and Pex13p (Fransen et al., 2001;Ghaedi et al., 2000b;Snyder et al., 2000). Recent studies also reported that Pex19p specifically bound to peroxisome membrane targeting signal (mPTS) regions of multiple PMPs (Jones et al., 2001;Jones et al., 2004;Rottensteiner et al., 2004) or at regions distinct from the sorting sequences (Fransen et al., 2001). Furthermore, the farnesylation of Pex19p is an important determinant for the higher affinity of Pex19p to several peroxins such as Pex10p and Pex13p (Fransen et al., 2001). Thus, Pex19p has been proposed to function in recruiting newly synthesized PMPs from their site of synthesis on free polyribosomes to the peroxisomes as a soluble receptor and/or a chaperone in targeting of PMPs (Fransen et al., 2001;Jones et al., 2001;Matsuzono and Fujiki, 2006;Snyder et al., 2000).
The peroxin Pex19p functions in peroxisomal membrane assembly. Here we mapped functional domains of human Pex19p comprising 299 amino acids. Pex19p mutants deleted in the C-terminal CAAx farnesylation motif, the Cterminal 38 amino acid residues and the N-terminal 11 residues, maintained peroxisome-restoring activity in pex19 cells. The sequence 12-261 was essential for reestablishing peroxisome activity. Pex19p was partly localized to peroxisomes but mostly localized in the cytosol. Pex19p interacted with multiple membrane proteins, including the other two membrane biogenesis peroxins, Pex3p and Pex16p, those involved in matrix protein import such as Pex14p, Pex13p, Pex10p, and Pex26p, peroxisome morphogenesis factor Pex11p␤ ␤, and a PMP70 peroxisometargeting signal region at residues 1-123. In yeast twohybrid assays, Pex10p and Pex11p␤ ␤ interacted only with full-length Pex19p. Of various truncated Pex19p variants active in translocating to peroxisomes, the mutants with the shortest sequence (residues 12-73 and 40-131) were localized to peroxisomes and competent in binding to Pex3p. Furthermore, membrane peroxins were initially discernible in a cytosolic staining pattern in pex19 cells only when co-expressed with Pex19p and were then localized to peroxisomes in a temporally differentiated manner. Pex19p probably functions as a chaperone for membrane proteins and transports them to peroxisomes by anchoring to Pex3p using residues 12-73 and 40-131.
As a further step toward understanding the molecular mechanisms of Pex19p function, in the present work we searched for the regions required for restoring peroxisome biogenesis, translocation to peroxisomes, and interaction with a wide-range of PMPs. We report that the sequence encompassing amino acid residues 12-261 is responsible for peroxisome-restoring activity in pex19 cells. Pex19p variants comprising only amino-acid residues 40-131 and 12-73 were translocated to peroxisomes and were responsible for the interaction between Pex19p and Pex3p. Moreover, upon coexpression with Pex19p, integral membrane peroxins were initially maintained in the cytosol in pex19 ZP119 cells and were then transported to newly formed peroxisomes. Taken together, Pex19p functions in the membrane-protein transport to peroxisomes, probably as a chaperone-like factor and a transporter.

Functional domain mapping of Pex19p
As a step toward understanding the molecular mechanisms involved in peroxisome membrane biogenesis, we first searched for functional regions of Pex19p. We verified various truncated mutants of tandem double hemagglutinin A (HA)tagged Pex19p, termed HA 2 -Pex19p ( Fig. 1), for the activity in restoring the impaired peroxisomal membrane biogenesis in CHO pex19 ZP119 cells (Kinoshita et al., 1998;Matsuzono et al., 1999), deficient in endogenous Pex19p (supplementary material Fig. S1C, lane 1).
The expression level of HA 2 -Pex19p variants in ZP119 was comparable between the mutants and similar to the fulllength Flag-Pex19p (supplementary material Fig. S1A, lanes 1-7). Numbers designate the positions in the amino acid sequence. Two tandem epitopes of hemagglutinin (HA 2 ) were tagged to the N-terminus of respective Pex19p variants. Several mutants with shorter sequences were fused to hexa-Myc tag, Myc 6 , and HA 2 . Farnesylation CAAx motif is shaded. The activities of the constructs in restoring peroxisomes in pex19 ZP119 cells and translocating to peroxisomal membranes in CHO-K1 are summarized on the right: +++, strong; ++, medial; +, positive; +/-; weakly positive; -, negative.
Deletions in the N-terminal region N-terminal truncation mutants of HA 2 -Pex19p ( Fig. 1B) were likewise verified in pex19 cells as in Fig. 2A. In the cells expressing a Pex19p mutant with deletion of 11 amino acid residues from the N-terminus, termed 12-299 (the same as ⌬N11), numerous PTS1-and Pex14p-positive punctate structures were observed (Fig. 2B,a,b), demonstrating that 12-299 was active in peroxisome restoration. By contrast, expression of 24-299 (⌬N23), 40-299 (⌬N39), and 70-299 (⌬N69) resulted in no punctate staining for PTS1 and Pex14p (Fig. 2Bc,d; Table 1), indicative of abrogation of peroxisomerestoring activity. ⌬12-23 lacking the potential sequence involved in ␣-helix folding (supplementary material Fig. S2) also failed to re-establish peroxisome assembly (Fig. 2Be,f), suggesting the functional importance of this region. In ZP119 cells transfected with M12-261 encoding the Pex19p mutant with deletion of amino-acid residues at 1-11 and 262-299 (  Table 1). These constructs were expressed at levels similar to the C-terminal deletion mutants (supplementary material Fig. S1A, lanes 8-12). Collectively, these results strongly suggested that the minimal sequence required for restoring the peroxisome assembly resides in 12-261, including functionally important regions at 12-23 and 255-261. It is also noteworthy that Pex19p and its biologically active truncation mutants assessed above were partly localized to peroxisomes upon restoration of peroxisomes (data not shown).
[ 35 S]Pex3p-EGFP was likewise co-immunoprecipitated with 1-131, M12-131, 1-73 and M12-73, but not with M24-131, M40-131, M24-73 and M40-73 (Fig. 5B, lanes 13-20), suggesting that the N-terminal short peptide sequence at 12-23 is a direct Pex3p-binding site. The interaction of M40-131 with Pex3p detected by co-immunoprecipitation using cell lysates (Fig. 5A) is more likely to be mediated by other factor(s). Accordingly, M12-73 and M40-131 are the shortest sequences that are competent in translocation to the peroxisome (see Fig. 3) and binding to Pex3p. These results suggest that Pex19p interacts with Pex3p at its two regions and that Pex3p functions in recruiting Pex19p to peroxisomal membranes as a potential anchoring site in peroxisome membrane biogenesis. Such two-site interaction of Pex19p with Pex3p was recently suggested by a bacterial two-hybrid assay (Fransen et al., 2005).

The N-terminal region of Pex19p interacts with Pex14p and Pex16p
As model PMPs, we investigated Pex14p, the Pex5p-docking receptor (Albertini et al., 1997;Otera et al., 2000;Otera et al., 2002), and Pex16p, one of the membrane peroxins essential for peroxisomal membrane assembly (Honsho et al., 1998;South and Gould, 1999). We did not detect interaction of Pex19p with Pex14p and Pex16p in our yeast two-hybrid assays (data not shown), whereas such bindings were previously shown by mammalian and yeast two-hybrid assays (Fransen et al., 2001;. Therefore, we assessed any such interaction by co-immunoprecipitation assay from the lysates of COS7 cells expressing HA 2 -or Myc 6 -HA 2 -Pex19p truncation mutants together with Flag-Pex14p or Flag-Pex16p (supplementary material Fig. S3; Table 1 (Table 1). Taken together, the minimum length sequences of Pex19p required for the interaction with Pex16p and Pex14p apparently reside at positions 40-255 and M40-73, respectively. With respect to Pex19p translocation to peroxisomes, Pex16p does not appear to be responsible, because several Pex19p truncation variants such as 1-131 and M12-131 are translocated to peroxisomes despite elimination of their activity in binding to Pex16p (Table 1). Pex14p is also less likely to be essential for peroxisomal localization of Pex19p, as noted that Pex19p is detectable in peroxisomal remnants in Pex14p-deficient CHO pex14 mutant cells (see below, Fig. 6).

Pex19p functions as chaperone and transporter of PMPs in peroxisome biogenesis
In an attempt to address the Pex19p interaction with multiple PMPs at an early stage of peroxisome biogenesis, we separately expressed Flag-tagged PMPs, two transmembranetype Pex14p (Shimizu et al., 1999) and Pex16p (Honsho et al., 1998), and C-terminal-tailed type-II Pex26p (Matsumoto et al., 2003), in pex19 ZP119 cells and determined their intracellular localization at 12, 24 and 36 hours post-transfection (Ghaedi et al., 2000b;Matsuzono et al., 1999). These Flag-peroxins were mostly localized to mitochondria and/or some membraneparticle-like structures discernible in proximity to mitochondria at 12 hours post-transfection, as verified by staining with MitoTracker (Fig. 7A, left panels). By contrast, when these Flag-PMPs were co-expressed with HA 2 -Pex19p, they were not localized to any intracellular membrane structures at 12 hours, apparently remaining in the cytosol as seen for HA 2 -Pex19p (Fig. 7A, middle left panels). At 24 hours, Flag-Pex14p and Flag-Pex16p were partly visible in a punctate staining pattern, coincident with HA 2 -Pex19p in punctate structures, thereby suggesting that peroxisome membranes were assembled, whereas Flag-Pex26p was mostly in the cytosol with HA 2 -Pex19p (Fig. 7A, middle, right panels). At 36 hours, these PMPs became distinct in a manner superimposable on particle-bound HA 2 -Pex19p, indicating that peroxisome membranes were assembled (Fig. 7A, right panels). Essentially the same results were obtained with Pex10p and Pex11p␤ as for Pex14p and Pex16p (data not shown). Particles stained with Flag-PMP and HA 2 -Pex19p were more readily discernible when the cytosol was washed out before cell fixation by digitonin treatment where only plasma membranes are permeabilized (Okumoto et al., 1998b) (data not shown). Flag-Pex26p-and HA 2 -Pex19p-positive particles at 36 hours were confirmed as the re-established peroxisomes by dual staining of Pex14p and HA 2 -Pex19p (Fig.  7As,t). HA 2 -Pex19p and Pex14p were likewise discernible in a superimposable manner in ZP119 cells expressing the respective Flag-PMP peroxins (data not shown). Moreover, endogenous Pex14p in ZP119 showed the same profile as the ectopically expressed Flag-Pex14p upon transfection of HA 2 -PEX19 (Fig. 7B). We interpreted these results to mean that Pex19p binds to newly synthesized PMPs in the cytosol, preventing them from mistargeting to other organelle membranes such as mitochondria, and then transports PMPs to peroxisome membranes, possibly in a temporally differentiated manner at least in the case of Pex26p, where peroxisomes are assembled. These findings are in good agreement with our most recent results (Matsuzono and Fujiki, 2006) reporting that Pex19p transports PMPs such as Pex16p and Pex26p to peroxisomes in vitro in an ATP-dependent manner. Furthermore, EGFP-fused Pex3p, a strong binding partner of Pex19p (Ghaedi et al., 2000b) (see Figs 4, 5; Table 1), was also mistargeted to mitochondria in the absence of Pex19p when expressed in ZP119 cells (Fig. 7C, left panels), consistent with the observation in human pex19 fibroblasts . Even upon co-expression with HA 2 -Pex19p, Pex3p-EGFP remained mostly mislocalized to mitochondria together with Pex19p (Fig. 7C, right panels). These results implied that Pex19p was less likely to bind to newly synthesized Pex3p as a chaperone as seen in other PMPs at the early steps of peroxisome assembly (Fig. 7A). Strong interaction of Pex19p with Pex3p may occur at other distinct steps such as that involving Pex19p docking on peroxisome membranes.

Discussion
In several earlier studies (Goette et al., 1998;Matsuzono et al., 1999;Snyder et al., 1999), Pex19p is shown to play a central role in the early steps of peroxisomal membrane biogenesis. Defects of Pex19p function results in the impairment of peroxisome biogenesis in mammalian cells such as CHO cells and human fibroblasts, including the failure of peroxisomal membrane assembly occurring before the import of peroxisomal matrix enzymes (Kinoshita et al., 1998;Matsuzono et al., 1999;. Expression of Pex19p restores the impaired peroxisome biogenesis in CHO pex19 mutant, ZP119 (Matsuzono et al., 1999) and CG-J (CG14) PBD patient-derived fibroblasts (Matsuzono et al., 1999;. In the present work, we showed that the deletion of the CAAx box motif from the C-terminus of Pex19p slightly reduced the peroxisome-restoring activity, as noted in PTS1 proteins, which partly remained in the cytosol. Our findings in yeast two-hybrid assays indicated that the CAAx box was important for the interaction of Pex19p with several peroxins, such as Pex10p, Pex11p␤, and Pex13p, but not with Pex3p (Table 1), consistent with the previous observation (Fransen et al., 2001). Thus, the slight reduction in peroxisome-restoring activity of 1-295(⌬C4) of Pex19p apparently giving rise to fewer re-established peroxisomes is possibly due to the abrogation of its interaction with such peroxins. Several explanations can be considered for this observation. First, membrane targeting or assembly of these peroxins, including Pex10p, Pex11p␤ and Pex13p, might not be a prerequisite for the early stages of peroxisomal membrane biogenesis. Pex19p may have an auxiliary role in targeting to peroxisome membranes or assembly of these peroxins during peroxisomal biogenesis. Second, Pex19p might not interact with these peroxins during the membrane biogenesis step but the interaction occurs at other biological steps, such as the maintenance of peroxisomes. Third, 1-295 might interact weakly with these peroxins, but such low activity is undetectable in the yeast two-hybrid system. It is also plausible that the impaired peroxisome membrane assembly in pex19 ZP119 is restored simply by overexpression of 1-295. Fransen et al. (Fransen et al., 2002) reported that farnesylation is an important step for the affinity of Pex19p with Pex10p and Pex13p, and that the tetrapeptide CAAx is required for interaction with Pex11p␤. In the present work, Flag-Pex11p␤ and Flag-Pex13p expressed in COS7 cells were indeed not coimmunoprecipitated with 1-295 (data not shown). However, both ⌬256-260 and ⌬256-295, lacking internal regions but containing the CAAx motif, lost the peroxisome-restoring activity and the interaction with Pex10p, Pex11p␤ and Pex13p (Table 1). Therefore, the farnesylation may not play a major role in the 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, hence may not be involved in the direct interaction with PMPs and peroxisome-restoring activity of Pex19p. It is noteworthy that Pex19p mutants truncated in the C-terminal region and defective in the pex19 cell-complementing activity can still interact with Pex3p (Figs 2, 4; Table 1). Together, Journal of Cell Science 119 (17)  physiological consequences of the farnesylation of Pex19p may include: enhancement of the membrane targeting efficiency of Pex19p-PMP complexes; regulation of Pex19p in binding and release from cargo PMPs; control of the binding steps of Pex19p-PMP complexes to peroxisome membranes; and release of the PMP cargo-unloaded Pex19p from peroxisome membranes.
The Pex19p mutant 1-261 re-established peroxisomes with a reduced efficiency, whereas 1-255 was inactive. This is reminiscent of the finding that a homozygous inactivating, onebase insertion frameshift in a codon for Met255, inducing a 24amino-acid sequence entirely distinct from normal Pex19p, is the genetic cause of Pex19p dysfunction in a patient with Zellweger syndrome of CG-J (CG14) (Matsuzono et al., 1999). Therefore, it is more likely that the region of highly conserved amino acid residues at 256-260, apparently responsible for folding the ␣-helix structure (Fig. S2) (Fransen et al., 2005), has an important role in peroxisome assembly. In the yeast twohybrid assay, 1-261, but not 1-255, bound to Pex26p, the recruiter of Pex1p-Pex6p complexes (Matsumoto et al., 2003). This might also explain the difference in peroxisome-restoring activity between these two Pex19p variants. Moreover, the mutant 12-299, but not 24-299, re-established peroxisomes in pex19 cells, indicating that the region of residues 12-23 involved in ␣-helix folding was essential for the complementing activity of Pex19p. The 12-299, not 24-299 and ⌬12-23, was indeed as active as the full-length protein in binding to Pex3p in the yeast two-hybrid assays. Thus, the interaction of Pex19p with Pex3p is probably mediated by this ␣-helix region and is important for early stages of peroxisome membrane biogenesis. Taken together, we conclude that M12-261 comprises the minimal sequence for biologically active Pex19p. It is noteworthy that ⌬12-23, 24-299 and 40-299, all defective in peroxisome-restoring activity, were coimmunoprecipitated with Pex3p. This may be explained by the findings that these three variants contain the sequence encompassing the residues at 40-131 apparently involved in the indirect interaction with Pex3p (see below).
Hydrophilic Pex19p is localized mostly in the cytosol and partly associated with peroxisomes (Goette et al., 1998;Jones et al., 2004;Matsuzono and Fujiki, 2006;Matsuzono et al., 1999; (this study) and peroxisomal remnants in pex mutants defective in matrix protein import (Matsuzono and Fujiki, 2006) (this study). It is possible that Pex19p translocates to peroxisome membranes via association with other factors such as a Pex19p docking factor in peroxisome membranes. Such candidates could be Pex3p and Pex16p, the peroxins essential for membrane assembly, and Pex14p (Table 1). Pex3p is a strong interacting partner of Pex19p (Ghaedi et al., 2000b;Muntau et al., 2003;Snyder et al., 1999;Soukupova et al., 1999) (Fig, 4; Table 1). Pex19p also interacts with Pex16p (Fransen et al., 2001;Jones et al., 2004) (Table 1) and Pex14p (Fransen et al., 2002; (Table 1). Two Pex19p variants, M12-73 and M40-131, are sufficient for translocation to peroxisomes in normal cells such as CHO-K1 (Fig. 3) and human fibroblasts (data not shown). M12-73 and M40-131 were co-immunoprecipitated with Pex3p and Pex14p, but not with Pex16p ( Fig. 5 and supplementary material Fig. S3), suggesting that Pex3p and Pex14p are required for Pex19p docking to peroxisomes. Of note, direct binding of M12-73 to Pex3p is apparent, whereas M40-131 binding to Pex3p is probably indirect (Fig. 5). However, in a CHO pex14 mutant, ZP110, which is deficient in Pex14p, Pex19p translocated to peroxisomal membrane remnants (see Fig. 6A), strongly suggesting that Pex14p is dispensable for Pex19p localization to peroxisomes and the peroxisomal ghost. The N-terminal residues 1-51 and 1-56 were shown to bind to Pex3p in the yeast (Fransen et al., 2001) and mammalian (Fang et al., 2004) two-hybrid assays, respectively. Pex3p is probably responsible for peroxisomal localization of Pex19p. This was supported by the findings using matrix-protein import-defective, CHO pex mutants and fibroblasts from patients with PBD, where Pex19p was localized to peroxisomal membrane remnants harboring PMPs such as Pex3p (Ghaedi et al., 2000b). On the other hand, Pex19p was exclusively localized in the cytosol in pex3 CHO mutant cells and PEX3-deficient PBD patient-derived fibroblasts. Moreover, we found that co-expression with Pex19p of several membrane peroxins such as Pex14p, Pex16p, and Pex26p in CHO pex19 cells prevented their mistargeting to mitochondria, rather maintaining them in the cytosol (Fig.  7A,B). Pex19p aided their translocation to peroxisome membranes, in a temporally differentiated manner, where peroxisomes were then assembled (Fig. 7A,B). PMPs such as Pex26p may be required at a later stage of biogenesis of functional peroxisomes. By contrast, co-expression of Pex3p with Pex19p in the pex19 cells did not confer the cytosolic localization of Pex3p at the early step of restoration of peroxisome assembly. Pex3p remained mostly mistargeted to mitochondria, where Pex19p was colocalized (Fig. 7C), thereby implying a distinct mechanism of Pex3p transport.
With regard to its physiological role, Pex19p is thought to function as a PMP chaperone (Hettema et al., 2000;Shibata et al., 2004) and an import receptor that mediates the transport of PMPs to peroxisomes (Fransen et al., 2004;Snyder et al., 2000) by interacting with their PTS . However, Fang et al. (Fang et al., 2004) 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. Furthermore, in our recently established in vitro Pex19p-dependent PMP transport system (Matsuzono and Fujiki, 2006), we also showed the release of Pex19p to the cytosolic fraction from peroxisomes, suggesting that Pex19p was a shuttling PMP receptor (Matsuzono and Fujiki, 2006).