Báo cáo khoa học: The common phospholipid-binding activity of the N-terminal domains of PEX1 and VCP/p97 doc

The common phospholipid-binding activity of the N-terminal domains of PEX1 and VCP/p97 Kumiko Shiozawa 1 , Natsuko Goda 1 , Toshiyuki Shimizu 1 , Kenji Mizuguchi 2,3 , Naomi Kondo 4 , Nobuyuki Shimozawa 4,5 , Masahiro Shirakawa 1,6 and Hidekazu Hiroaki 1 1 International Graduate School of Arts and Sciences, Yokohama City University, Tsurumi-ku, Yokohama, Kanagawa, Japan 2 Department of Biochemistry, University of Cambridge, UK 3 Department of Applied Mathematics and Theoretical Physics, Centre for Mathematical Sciences, University of Cambridge, UK 4 Department of Pediatrics, Gifu University School of Medicine, Japan 5 Division of Genomic Research, Life Science Research Center, Gifu University, Japan 6 Department of Molecular Engineering, Graduate School of Engineering, Kyoto University, Japan The peroxisome is a single-membrane organelle involved in various metabolic pathways [1]. Biogenesis and maintenance of the peroxisome require at least 32 proteins, known as PEX gene products or peroxins [2]. Autosomal recessive mutations in any of 12 of the PEX genes cause peroxisome biogenesis disorders, such as Zellweger syndrome, neonatal adrenoleukodystro- phy and infantile Refsum disease [3]. Peroxisomal membrane fusion and protein transloca- tion are both ATP-dependent processes, but among over 32 peroxins, only PEX1 and PEX6 are AAA- ATPases. Dysfunction of the ATPase activity of either of the two proteins results in peroxisome biogenesis disorders [4–7]. The ATPase activities of PEX1 and PEX6 are believed to be indispensable for normal peroxisomal biogenesis, including the critical step of Keywords AAA-ATPase; N-terminal domain; PEX1; phospholipid; valosine-containing protein Correspondence H. Hiroaki, Division of Molecular Biophysics, Graduate School of Integrated Sciences, Yokohama City University, 1-7-29, Suehirocho, Tsurumi, Yokohama, Kanagawa, Japan 230-0045 Fax: +81 45 508 7361 Tel: +81 45 508 7214 E-mail: hiroakih@tsurumi.yokohama-cu.ac.jp (Received 21 June 2006, revised 6 September 2006, accepted 11 september 2006) doi:10.1111/j.1742-4658.2006.05494.x PEX1 is a type II AAA-ATPase that is indispensable for biogenesis and maintenance of the peroxisome, an organelle responsible for the primary metabolism of lipids, such as b-oxidation and lipid biosynthesis. Recently, we demonstrated a striking structural similarity between its N-terminal domain and those of other membrane-related AAA-ATPases, such as valo- sin-containing protein (p97). The N-terminal domain of valosine-containing protein serves as an interface to its adaptor proteins p47 and Ufd1, whereas the physiologic interaction partner of the N-terminal domain of PEX1 remains unknown. Here we found that N-terminal domains isolated from valosine-containing protein, as well as from PEX1, bind phosphoinos- itides. The N-terminal domain of PEX1 appears to preferentially bind phosphatidylinositol 3-monophosphate and phosphatidylinositol 4-mono- phosphate, whereas the N-terminal domain of valosine-containing protein displays broad and nonspecific lipid binding. Although N-ethylmaleimide- sensitive fusion protein, CDC48 and Ufd1 have structures similar to that of valosine-containing protein, they displayed lipid specificity similar to that of the N-terminal domain of PEX1 in the assays. By mutational analy- sis, we demonstrate that a conserved arginine surrounded by hydrophobic residues is essential for lipid binding, despite very low sequence similarity between PEX1 and valosine-containing protein. Abbreviations AAA, ATPase associated with a diversity of cellular activities; ER, endoplasmic reticulum; GST, glutathione-S-transferase; IPTG, isopropyl thio-b- D-galactoside; ND, N-terminal domain; NSF, N-ethylmaleimide-sensitive fusion protein; PA, phosphatidic acid; PC, phosphatidylcholine; PE, phosphatidylethanolamine; PhA, phytic acid; PS, phosphatidylserine; PtdIns, phosphatidylinositol; QCM, quartz crystal microbalance; SKD1, suppressor of K + transport growth defect 1; SNAP, soluble NSF attachment protein; VCP, valosine-containing protein. FEBS Journal 273 (2006) 4959–4971 ª 2006 The Authors Journal compilation ª 2006 FEBS 4959 peroxisomal protein import and peroxisomal mem- brane fusion [8,9,49]. The two proteins have been shown to associate with each other both in vitro and in vivo [5,10–12], and it has been found that ATP bind- ing and hydrolysis are important for their interaction in Saccharomyces cerevisiae [12]. Recently, it has been reported that the PEX1–PEX6 complex is involved in the dissociation of ubiquitinated PEX5 from the per- oxisomal membrane [13]. PEX5 binds the peroxisomal proteins (cargo proteins), which contain peroxisome targeting signal 1 at the C-terminus in the cytosol, tar- gets them to the protein complex machinery at the peroxisomal membrane and releases them into the peroxisomal lumen before returning to the cytosol. Several groups have reported that PEX5 can return to the cytosol in a ubiquitin-dependent manner [13–17], and it is believed that the PEX1–PEX6 complex is indispensable to this step [15]. AAA-ATPases are found in three domains of all liv- ing organisms [18,19], and play an important role as molecular chaperones, including in the dissociation of protein complexes and protein translocation. PEX1 belongs to the class of type II AAA-ATPases, which contain two copies of the AAA cassette. Extensive structure–function studies of N-ethylmaleimide-sensi- tive fusion protein (NSF) and valosine-containing pro- tein (VCP) (p97), which also belong to this class, have been reported [20–31,33–35]. These enzymes form a hexameric ring and can act as protein unfoldases. Type II AAA-ATPases are located in organelle membranes, where they are involved in specific functions, such as membrane fusion [20–22] and protein transport across the membrane [23]. NSF, and its yeast ortholog Sec18, are responsible for heterotypic membrane fusion mainly in exocytic pathways [20,24,25]. a-soluble NSF attachment protein (SNAP), b-SNAP and c-SNAP are known to be the most important targets of NSF [21]. VCP and yeast CDC48 are involved in endoplasmic reticulum (ER)-associated protein degradation [23,26,27] as well as in remodeling of the Golgi and nuclear membrane [20,28]. Although the specific target of VCP unfoldase activity remains unclear, there are several adaptor molecules, p47 [28], Ufd1 ⁄ Npl4 [27], VCIP135 [29], Derlin-1 and VIMP [30,31], which may determine VCP functions. Previously, we have determined the crystal structure of the N-terminal domain (ND) of mouse PEX1 (PEX1-ND) [32]. It bears a striking resemblance to those of VCP (p97) [33], NSF [34,35], the archaeal homolog VAT [36] and Sec18 [40], despite the low level of sequence similarity. The domain architecture of all five proteins contains an ND followed by the tandem AAA domains D1 and D2. This architecture is known as a ‘supradomain’, and is found in many cases where two or three domains (in this case, ND, D1 and D2) are persistently conserved in terms of their sequen- tial order and biological context [37]. In the crystal structure of PEX1-ND, the characteristic crevice, sim- ilar to that of NSF, is conserved. NSF-ND is assumed to be a binding site for a-SNAP [34]. In the case of VCP-ND, a flat hydrophobic surface provides an inter- face to the ubiquitin-like domain of p47 [38], and the surface might be used for binding other adaptor pro- teins, such as Npl4–Ufd1 complex and VCIP135. Inter- estingly, the ND of Ufd1, which is similar to VCP-ND as well as PEX1-ND, also shares a common hydropho- bic interface for polyubiquitin binding [39]. Ufd1 is a non-ATPase-type adaptor protein associated with VCP and Npl4. This hydrophobic surface is not conserved in PEX1-ND, and the molecular function of this pro- tein remains to be resolved. The structural similarity between VCP, NSF and PEX1 suggested that, as VCP binds phospholipids [22,58], PEX1 could have similar properties. We there- fore investigated phospholipid binding and the binding sites of PEX1 and VCP. Evolutionary trace analysis revealed a conserved charged residue surrounded by hydrophobic residues in the ND of PEX1 and VCP, which may bind to the phospholipid. It is shown below that both PEX1-ND and VCP-ND can bind phospho- lipids in vitro with broad specificity for phosphatidyl- inositol (PtdIns) monophosphate species. By analogy to PEX1, we examined phospholipid binding and bind- ing sites in the NDs of NSF, CDC48 and Ufd1. As their surface properties differ from those of PEX1 and VCP, no lipid-binding site was found by computa- tional analysis, although experimentally, they were all found to bind phospholipids. Results Evolutionary trace analysis of PEX1-ND Based on the structure-restrained sequence alignment of PEX1 orthologs with other AAA-ATPase NDs, VCP is the closest neighbor of PEX1. To identify potential protein–protein and protein–lipid interaction surfaces, we have analyzed the available structural data for PEX1 [32] and VCP [33] and searched for con- served charged residues as well as hydrophobic resi- dues exposed at the protein surface (Figs 1B and 2; supplementary Fig. S1) [44,45]. R135 in PEX1 and R144 in VCP are the only exposed charged residues, conserved across the two protein families, whereas K174 is relatively conserved among other PEX1 ortho- logs (Fig. 1B). These positively charged residues are Phospholipid-binding activity of adaptor domains K. Shiozawa et al. 4960 FEBS Journal 273 (2006) 4959–4971 ª 2006 The Authors Journal compilation ª 2006 FEBS located at the end of the shallow groove between the N-lobe and C-lobe of PEX1-ND. An exposed constel- lation of hydrophobic residues is found along a longi- tudinal line of the kidney-shaped PEX1-ND, which surrounds the conserved basic residues. In addition, the corresponding R144 in VCP-ND, equivalent to R135 in PEX1, was found to be surrounded by exposed hydrophobic residues (supplementary Fig. S1). These residues are relatively well conserved among VCP orthologs, although the shape of the hydrophobic constellation differs (Fig. 2B). In order to unravel the characteristics of these com- mon charged and hydrophobic interfaces, protein interface predictions were made based on optimal docking areas (ODAs) in PEX1-ND and VCP-ND [46] (supplementary Fig. S2). ODAs are a set of continuous surface patches with optimal protein–protein docking desolvation energy. A previous analysis correctly located known protein–protein interfaces in 80% of cases [46]. Indeed, the known p47 interaction surface in VCP-ND was detected with a significant low-energy value (< ) 10 kcal Æmol )1 ; shown in red in supplement- ary Fig. S2). In contrast, the areas near R135 in PEX1-ND and the corresponding region around R144 in VCP-ND displayed no significant ODAs despite their hydrophobic nature, suggesting that this surface may be used for interactions with nonprotein mole- cules. This observation led to the hypothesis that PEX1-ND and VCP-ND may directly associate with phospholipids. PEX1-ND as a phosphoinositide-binding domain One of the common features of the type II AAA-ATP- ases is their role in organellar membrane fusion. To the best of our knowledge, there are no reports indica- ting that PEX1-ND is responsible for recruitment of PEX1 to the peroxisomal membrane. Several lines of evidence indicate that the N-terminal region of PEX1 may directly associate with the membrane. The binding of PEX1-ND to phospholipids as well as its specificity have been investigated by incubating a purified gluta- thione-S-transferase (GST)-tagged PEX1-ND on nitro- cellulose membranes spotted with different lipid species (PIP Strip). Binding of protein to the membrane was quantified using anti-GST serum (Fig. 3A). Mouse PEX1-ND (residues 3–180) clearly represented the region that binds phospholipid. PEX1-ND bound most strongly to PtdIns monophosphates (PtdIns3P, PdsIns4P, and PtdIns5P) with approximately equal affinity. PEX1-ND weakly bound to PtdIns bisphos- phates [PtdIns(4,5)P 2 , PtdIns(3,4)P 2 and PtdIns(3,5)P 2 ] with lower or negligible affinity compared to PtdIns monophosphate. Very weak binding of PEX1-ND to PtdIns, phosphatidic acid (PA) and phosphatidylserine (PS) was observed, while no binding to phosphatidyl- choline (PC) and phosphatidylethanolamine (PE) could be detected. GST alone did not bind to phospholipids under these conditions (Fig. 3A, right). The same experiment was carried out in the presence of 20 mm phytic acid (inositol hexakisphosphate), which is a competitive inhibitor of various phosphoinositide-bind- ing domains with a broad specificity [47]. Nonspecific electrostatic interactions between the characteristic positively charged residues of PEX1-ND and the phospholipid are likely to contribute significantly to this binding. Indeed phytic acid inhibited the binding of PEX1-ND to the phosphoinositides, thereby demon- strating that lipid binding occurs specifically to the inositol phosphate moiety (Fig. 3A, middle). In addi- tion, we found that PtdIns binding to PEX1-ND is Ca 2+ -independent (data not shown), in contrast to the behavior of some other phosphoinositide-binding domains, such as annexin and the C2 domain, whose lipid binding is Ca 2+ -dependent [69]. The lipid-binding specificity of GST–PEX1-ND was studied in detail using a liposome recruitment assay. PC + PE (1 : 1)-based liposomes containing up to 5% PtdIns3P, PtdIns4 P, PtdIns5P or PtdIns were prepared (Fig. 3B). The proteins that sedimented as well as those that remained in solution were analyzed by Coomassie brilliant blue-stained SDS ⁄ PAGE and quantified. PEX1-ND bound most strongly to PtdIns3P and PdsIns4P, and weekly to PtdIns5P. The amount of PEX1-ND recruited to the liposomes increased with the amount of PtdIns monophosphate included in the liposome preparation. The apparent binding constant K d for binding between PEX1-ND and PtdIns3P was 1 lm, as judged from its saturation curve. The inhibitory effect of phytic acid is also demonstrated in Fig. 3B. We assumed that PEX1-ND interacts with PtdIns4P more strongly and more spe- cifically than with PtdIns3 P, as PtdIns4P binding is maintained in the presence of excess phytic acids, since phytic acid is not a perfect mimic of the head- group of PtdIns4P. Although PEX1-ND is recruited to the liposomes in the presence of a large excess of PtsIns5P or PtdIns, there is no steep saturation curve, suggesting that the binding is nonspecific. The liposome interaction of PEX1-ND was further confirmed by quartz crystal microbalance (QCM) assays. Figure 3C shows some typical results for GST– PEX1-ND immobilized on a sensor chip and titrated by liposomes. The frequency change was only observed when the liposome contained PtdIns4P, which is con- sistent with the results above. K. Shiozawa et al. Phospholipid-binding activity of adaptor domains FEBS Journal 273 (2006) 4959–4971 ª 2006 The Authors Journal compilation ª 2006 FEBS 4961 A B Phospholipid-binding activity of adaptor domains K. Shiozawa et al. 4962 FEBS Journal 273 (2006) 4959–4971 ª 2006 The Authors Journal compilation ª 2006 FEBS Determination of the residues responsible for phospholipid binding in PEX1-ND To determine the residues responsible for phosphoinos- itide binding, two mutants of PEX1-ND, R135A and K174A, were generated to eliminate the conserved pos- itive residues, which may serve as acceptors of the phosphate moiety of phosphoinositides. Both mutants (R135A and K174A) were subjected to the PIP Strips assay (Fig. 4A). Whereas K174A was found to bind to PtdIns monophosphates with approximately the same affinity and specificity as the wild-type protein, R135A had effectively lost all phosphoinositide-binding activ- ity. In order to estimate the binding ability, R135A was subjected to the liposome recruitment assay (Fig. 4B). R135A did not bind the PC + PE-based liposome containing PtdIns4P. Phosphoinositide-binding activity of ND of mouse VCP By analogy to PEX1, VCP may directly attach to the membrane via the ND, despite the low level of sequence similarity. As mentioned above, R144 in VCP-ND, equivalent to R135 in PEX1-ND, is con- served among VCP orthologs and surrounded by exposed hydrophobic residues. A comparison of these hydrophobic residues of VCP with the corresponding residues of PEX1-ND is shown in Fig. 2B. In contrast, some hydrophobic residues, colored in green, are only conserved among VCP orthologs but not in PEX1. To test possible binding of VCP-ND to phospholi- pids through this surface, mouse VCP-ND was purified as a GST fusion product and subjected to a lipid-bind- ing assay using PIP Strips, and, as shown in Fig. 4C, significant phospholipid binding was found. VCP-ND binds PtdIns monophosphates (PtdIns3P, PtdIns4P and PtdIns5P) as well as PtdIns bisphosphates, such as PtdIns(3,4)P 2 (i.e. it has a lower specificity than PEX1- ND). By analogy to PEX1-ND, the R144A mutant of VCP was generated to confirm the residue responsible for phosphoinositide binding. In the PIP Strips assays (Fig. 4), R144A displayed no detectable binding to PtdIns monophosphates, indicating the importance of R144 for phospholipid binding. Other NDs also bind phosphoinositides By analogy to PEX1, other proteins containing NDs with the same folding topology may attach to the membrane via this domain, despite the low level of sequence conservation. To test this hypothesis, we purified NDs derived from mouse NSF, yeast CDC48, mouse Ufd1 and yeast Ufd1 as GST fusion proteins, and subjected them to a lipid-binding assay using PIP Strips. As shown in sup- plementary Fig. S3, all GST fusion proteins of NDs bound phospholipids to some extent. The specificity of binding of these NDs to phosphoinositides was again vague, as shown in supplementary Fig. S3. Most of the NDs bound preferentially to PtdIns monophos- phates, whereas some bound weakly to PtdIns bis- phosphates. In addition, the phospholipid-binding properties of NSF were sensitive to an excess of phytic acid, suggesting that the interaction is specific for the inositol phosphate molecules of phosphoinositides. Discussion Phospholipid-binding interface on PEX1-ND Our results, which show that isolated PEX1-ND binds phosphoinositides with broad specificity in vitro, are consistent with the proposed function of PEX1 at the peroxisomal membrane [10,48,49,62]. We have identi- fied a specific surface region of PEX1-ND implicated in phospholipid interactions. Although the surface charge distribution of PEX1-ND is rather acidic, the protein does not bind phospholipids containing basic head- groups, such as PE and PC. From a multiple sequence alignment analysis of PEX1 orthologs, we found a single well-conserved basic residue (R135) located at one end of a shallow groove, which is assumed to be a substrate- binding site in NSF [34,50]. Furthermore, a constella- tion of well-conserved hydrophobic residues, such as V36, V49, W60, W116, L121, L130, L131, W146, V147, L175, L176 and I177, is positioned vertically across the surface of the groove, with R135 and K174 near one end (Fig. 2A). The surfaces where the hydrophobic patch Fig. 1. Sequence comparison of the N-terminal domain (ND) of PEX1 with that of other related proteins. (A) Schematic representation of the domain architecture of AAA-ATPases PEX6, PEX1, N-ethylmaleimide-sensitive fusion protein (NSF) and valosine-containing protein (VCP). The ND AAA-ATPase domains (D1, D2) are shown. Physical interactions among the domains are indicated by arrows. (B) Multiple sequence alignment of PEX1-ND and related NDs with secondary structure elements of PEX1-ND. For sequence names, see Experimental procedures. The alignment was generated with CLUSTALX [60] and manually edited. The secondary structure elements are shown at the top, with thick line segments for the a-helices (a1–a4) and thin line segments for b-strands (b1–b14). The conserved residues of PEX1-ND found in the putative phospholipid-binding site are shown on the second line. Conserved hydrophobic residues and basic residues of PEX1-ND are indica- ted by closed circles and closed squares, respectively (see text). The conserved residues and class-specific residues of PEX1-ND and VCP- ND are colored yellow and green, respectively. Well-conserved basic residues on the surfaces of PEX1-ND and VCP-ND are colored blue. K. Shiozawa et al. Phospholipid-binding activity of adaptor domains FEBS Journal 273 (2006) 4959–4971 ª 2006 The Authors Journal compilation ª 2006 FEBS 4963 and the conserved basic residues are found are flat. The peroxisomal membrane is composed of approximately 10% PC, the head-group of which contains hydrophobic methyl groups [51]. It is noteworthy that phospholipids with hydrophilic head-groups such as phosphoinositides may be sparsely distributed on the membrane, whereas PC may form a hydrophobic cluster. Thus, the juxtapo- sition of positively charged and hydrophobic residues on a flat surface of PEX1-ND suggests an interaction with such a membrane surface. The results obtained with the R135A and K174A mutants clearly demon- strate that this basic and hydrophobic surface is the phospholipid-binding interface and that arginine 135 is responsible for specific recognition of the phosphate group. The importance of the conserved arginine residue in the molecular recognition of phosphoinositides in other phospholipid-binding domains, such as FYVE [52] and PX [53], has been reported. We thus conclude that PEX1-ND is one of the functional modules for membrane binding. Involvement of phosphoinositides in peroxisomal biogenesis The interaction between PEX1-ND and phospholipids could be specific to an inositol phosphate moiety, because lipid-binding activity is abolished in the pres- ence of excess phytic acid as a competitor (Fig. 3A). In terms of phospholipid specificity, there is only a limited preference for PtdIns monophosphate species over other phosphoinositides (Fig. 3A,B). Some of these nonspe- cific lipid-binding domains are believed to act as scaffolds for a specific membrane rather than in the PtdIns-mediated signal transduction systems. There are a few reports in the literature suggesting the involvement of PtdIns-mediated signal transduction systems in peroxisomal biogenesis, such as those involving phos- phatidylinositol-3-kinase and phosphatidylinositol- 4-phosphate-5-kinase. Pexophagy, an event in the autophagic degradation of excess of peroxisomes, may be an exceptional phenomenon [54]. More recently, peroxisome fusion, the critical early step in peroxisome assembly, for which PEX1 and PEX6 are essential, has been reported to require the phosphoinositides PtdIns4P and PtdIns(4,5)P 2 , as well as a distinct set of peroxisom- al membrane proteins that specifically bind to these two lipids [62]. This supports our finding that PEX1-ND interacts with phosphoinositides. Nevertheless, phos- phoinositide-specific regulation of the peroxisome, akin to receptor signaling systems or endocytosis, therefore appears to be unlikely, because PtdIns4P is the most abundant phosphoinositide in the cell. In contrast, the number of nonspecific membrane-binding domains is R135 V36 V49 W60 W146 L121 W116 I177 L176 L131 L175 K174 L130 L169 V147 N C I70 V108 Y110 Y143 A142 R144 F139 L140 V181 I182 F131 V154 F152 Y138 C N C A B Fig. 2. Constellation of hydrophobic residues with conserved basic residues on the surface of the N-terminal domains (NDs) of PEX1 and valosine-containing protein (VCP). (A) Conserved hydrophobic residues within the PEX1 orthologs shown in Fig. 1B are colored yellow. Conserved basic residues R135 and K174 are colored blue. (B) Hydrophobic residues conserved among PEX1 and VCP (V108, Y138, F139, F152, V154, I182) are colored yellow. Conserved hydrophobic residues within VCP orthologs only (I70, Y110, F131, A142, Y143, L140 and V181) are colored green. The conserved basic residue R144 is colored blue. The inset is a ribbon diagram of the molecule. Phospholipid-binding activity of adaptor domains K. Shiozawa et al. 4964 FEBS Journal 273 (2006) 4959–4971 ª 2006 The Authors Journal compilation ª 2006 FEBS often correlated with that of other domains responsible for specific protein–protein interactions. For example, more than 63% of pleckstrin homology-domain con- taining proteins also contain one or more protein-inter- acting domains such as SH3 [55]. This may enhance specificity for a particular subcellular compartment. PEX1-ND meets the criteria for such a protein, as it forms a scaffold on the peroxisomal membrane through specific interactions with phospholipids, whereas PEX6 interacts with either PEX26 (mammals) or PEX15 (yeast) on the membrane (Fig. 5) [56,57]. Some mem- brane-binding proteins bind multiple lipids and ⁄ or pro- teins simultaneously [63]. The binding of PEX1-ND to PtdIns might also be attributed to coincident PtfIns monophosphate binding. Common phospholipid-binding activity of VCP-ND and PEX1-ND We have also demonstrated that the NDs of another type II AAA-ATPase, VCP, can directly bind phos- pholipids in vitro. Although this is the first report of direct interaction between the isolated NDs and phos- phoinositides, several lines of evidence concerning VCP and NSF support our findings. Affinity purifica- tion of NSF using immobilized phospholipids, such as PtdIns(4,5)P 2 and PA, has been reported [67]. Purified recombinant NSF was shown to promote fusion of synthetic liposomes in the absence of a-SNAP and SNAP receptors, although the required lipid content was critical [41,42]. Involvement of PtdIns4P and PtdIns(4,5)P 2 during Sec18-dependent homotypic vacu- ole fusion in vitro has been reported [43]. All these results suggest that NSF and Sec18 possess some abil- ity to bind phospholipids. VCP, which promotes homo- typic fusion of Golgi or nuclear membranes [20], in concert with its adaptor protein p47 [28] and ⁄ or VCIP135 [29], can also promote the fusion of PE- based liposome vesicles in the absence of p47, with a reduced but nevertheless substantial measurable effi- ciency [58]. This suggests that a specific region of VCP binds to PE-based liposomes. More recently, ATP- mPEX1 ND (3-180) A GST only 20m M PhA B Bound protein (%) PI5PI3 PI4 Lipid content (%) 0 123450 1 2 3 4 5 0 1 2 3 4 5 012345 PI 30 25 20 15 10 5 0 C (a) (b) 0 -5 -10 -15 delta F / Hz Time (s) 030-15 604515 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Fig. 3. The N-terminal domain (ND) of PEX1 is a phosphoinositide-binding domain. (A) Phospholipid binding of PEX1-ND. The puri- fied glutathione-S-transferase (GST)–PEX1- ND fusion proteins were examined by PIP Strip assay and detected by anti-GST serum. The spots correspond to lyso-phosphatidic acid (PA) (1), lyso-phosphatidylcholine (PC) (2), phosphatidylinositol (PtdIns) (3), PtdIns3P (4), PtdIns4P (5), PtdIns5P (6), phosphatidylethanolamine (PE) (7), PC (8), sphingosine 1-phosphate (9), PtdIns(3,4)P 2 (10), PtdIns(3,5)P 2 (11), PtdIns(4,5)P 2 (12), PtdIns(3,4,5)P 3 (13), PA (14), PS (15), blank (16). The negative control was GST only (right panel). The middle panel corresponds to the same experiments conducted in the presence of 20 m M phytic acid (PhA). (B) Liposome-binding assay of GST–mPEX1-ND. Fraction of protein bound to PC ⁄ PE (1 : 1)- based liposomes was plotted against increasing amounts of PtdIns3P (PI3), PtdIns4P (PI4), PtdIns5P (PI5) and PtdIns (PI) in the liposomes, in the absence (filled bar) or presence (open bar) of 20 m M PhA as a competitor. Solid lines indicate standard deviation of three experiments. (C) Quartz crystal microbalance (QCM) assay of lipo- somes, which were added onto GST– mPEX1-ND immobilized on the QCM electrode. A typical time-dependent drop in frequency after injection of PC ⁄ PE (1 : 1)-based liposome is shown. (a) 5% PtdIns(4)P. (b) PC ⁄ PE only. K. Shiozawa et al. Phospholipid-binding activity of adaptor domains FEBS Journal 273 (2006) 4959–4971 ª 2006 The Authors Journal compilation ª 2006 FEBS 4965 hydrolysis-deficient mutants of VCP were shown to accumulate on reticular membranes in vivo [22]. On the other hand, colocalization of VCP and PtdIns-4-kinase IIa on the buoyant subfraction of ER-derived mem- brane was observed [66]. The ND of VCP, rather than the D1 or D2 domains, was assumed to act as an interface for the phospholipid membrane. The data shown here may partly provide the molecular basis of attachment of VCP to the ER-derived membrane. It is noteworthy that other AAA-ATPase proteins, belong- ing to the SF6 subfamily, possess transmembrane heli- ces at their N-terminus [19]. Sequence similarities in D1–D2 domains as well as the entire domain architectures of PEX1 and VCP sug- gest that PEX1 can form a stable hexameric ring struc- ture via the tandem D1–D2 AAA-ATPase domains such as VCP. Although there are few data concerning direct protein–lipid interaction involving AAA-ATP- ases that would prove a common molecular function of the ND, our results are largely consistent with pre- vious observations. The apparent association of PEX1 and VCP with the subcellular membranes could, of course, be attributed not only to protein–lipid interac- tion, but also to protein–protein interactions via other membrane-associated proteins. We do not rule out this possibility, as simultaneous binding of a single ND to both a lipid and a ligand may occur [63]. In addition, binding of PtdIns to PEX1-ND may also coincide with binding of PtdIns phosphates. Protein interface predic- tions based on ODAs [46] (supplementary Fig. S2) enabled detection of the known p47 interaction site of VCP-ND, but the corresponding surface in PEX1-ND produced no significant energy values. Therefore, PEX1-ND is unlikely to bind the Ubx domain or ubiquitin in a manner similar to that of the VCP-ND– Ubx (p47) complex, consistent with our previous ana- lyses [32]. Interestingly, a small patch of low-energy surface was found in a-helix 3 of PEX1-ND, which is highly conserved among its orthologs. This area might be involved in protein–protein interaction. As this Fig. 5. Schematic representation of the binding of the N-terminal domain (ND) of PEX1 to the peroxisomal membrane. PEX1 is pre- dicted to interact with the PEX6 portion of AAA domains. PEX1-ND binds to the peroxisomal membrane by recognizing specific lipids, whereas PEX6 interacts with either PEX26 or PEX15 on the mem- brane. mPEX1 ND (3-180) R135A K174A GST only A Wild-type B Lipid content (%) 012345 PI(4)P Bound protein (%) 30 25 20 15 10 5 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 GST onlyR144A Wild-type mVCP ND (1-200) Wild-type C 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 20mM PhA Fig. 4. Phospholipid binding of wild-type and R144A mutant of the N-terminal domain (ND) of valosine-containing protein (VCP)-ND. (A) Two mutants of the conserved basic residues, R135A and K174A, were analyzed by PIP Strips assay. K174A retained the ability to bind phosphatidylinositol (PtdIns) monophosphates, whereas R135A did not. (B) The R135 mutant was subjected to the lipo- some recruitment assay as in Fig. 3B. The fraction of protein bound to PC ⁄ PE (1 : 1)-based liposomes containing increasing amounts of PtdIns4P is plotted. Solid lines indicate the standard deviation of three experiments. (C) PIP Strips assay for wild-type VCP-ND and its R144A mutant, which does not bind PtdIns monophosphates. Phospholipid-binding activity of adaptor domains K. Shiozawa et al. 4966 FEBS Journal 273 (2006) 4959–4971 ª 2006 The Authors Journal compilation ª 2006 FEBS ODA area and the putative phospholipid-binding sur- face do not overlap, simultaneous binding of a single ND to both a membrane and a protein substrate might occur. We are currently testing this hypothesis. In the present study, we verified that the NDs of PEX1, VCP, NSF and Ufd1 bind PtdIns. Ufd1-ND, which has structural similarity with PEX1-ND, is the only known example of a protein with an ND and no AAA-ATPase domain. Furthermore, there are several reports indicating that other AAA-ATPases bind phos- phoinositides (e.g. in the ER and Golgi) [43,64,65,68]. Examples include phosphoinositide binding of SKD1, which is a type I AAA-ATPase, whose ND (MIT domain) is structurally different from that of PEX1 [68,70]. The binding mode and binding, however, remain unknown and may differ from those of PEX1. Experimental procedures Expression and mutational study of PEX1-ND The construction of a GST fusion protein containing the mouse PEX1 gene (encoding residues 3–180) in pGEX-4T3- PRESAT, according to the PRESAT vector methodology [59], has been previously described [32]. Ala-substituted mutants R135A and K174A were prepared with a Gene- Editor site-directed mutagenesis system (Promega, Madison, WI, USA), according to the manufacturer’s instructions. The coding regions were sequenced after introduction of the mutations. Mouse PEX1-ND and their mutants were expressed as GST fusion proteins in LB medium containing 1% glucose (LBG) containing ampicillin (50 lgÆmL )1 ). The cells were grown to a D 600 of 0.3, and heterologous gene expression was induced by addition of 1 mm isopropyl thio-b-d-galactoside (IPTG). The cells were collected 12–16 h after IPTG induction, washed, and disrupted by sonication. Recombinant proteins were purified by a single chromatography step using glutathione Sepharose (Amersham Bioscience, Uppsala, Sweden). The GST fusion proteins were dialyzed against buffer containing 150 mm KCl and 50 mm Hepes (pH 7.5) prior to the phospholipid- binding assays. Expression and mutational study of VCP-ND The plasmids for GST fusion proteins containing the mouse VCP (residues 1–200) were constructed by a standard pro- tocol using PCR, and subcloned into pGEX-4T3 (Amer- sham Bioscience). Ala-substituted mutants R144A were prepared with a Gene-Editor site-directed mutagenesis sys- tem (Promega), according to the manufacturer’s instruc- tions. The coding regions were sequenced after introduction of the mutations. Mouse VCP and mutant were expressed as GST fusion proteins in LBG containing ampicillin (50 lgÆmL )1 ). The cells were grown to a D 600 of 0.5, and heterologous gene expression was induced by addition of 1mm IPTG. The cells were collected after 3–5 h of IPTG induction, pelleted, washed, and disrupted by sonication. Recombinant proteins were purified by a single chromato- graphy step using glutathione Sepharose. The GST fusion proteins of the NDs were dialyzed against buffer containing 150 mm KCl and 50 mm Hepes (pH 7.5) prior to the phospholipid-binding assays. Expression of other NDs of NSF, CDC48 and Ufd1 A GST fusion expression construct containing the mouse NSF gene (encoding residues 5–200) was constructed in a manner similar to that for PEX1-ND. The plasmids for GST fusion proteins containing yeast CDC48 (residues 1–210), mouse Ufd1 (residues 14–194) and yeast Ufd1 (resi- dues 1–210) were constructed by a standard PCR protocol, and subcloned into pGEX-4T3 (Amersham Bioscience). The cells were grown to a D 600 of 0.5, and heterologous gene expression was induced by addition of 1 mm IPTG. The cells were collected after 3–5 h of IPTG induction, pel- leted, washed, and disrupted by sonication. Recombinant proteins were purified by a single chromatography step using glutathione Sepharose. The GST fusion proteins of the NDs, except mouse ⁄ yeast Ufd1, were dialyzed against buffer containing 150 mm KCl and 50 mm Hepes (pH 7.5) prior to the phospholipid-binding assays. Mouse and yeast Ufd1 were dialyzed against buffer containing 85 mm KCl, 50 mm Hepes (pH 7.5) and 10% glycerol. Phospholipid-binding assays For assessment of phospholipid-binding properties, PIP Strips (Echelon Bioscience Inc., Salt Lake City, UT, USA) were blocked with binding buffer containing 150 mm NaCl, 10 mm Hepes (pH 7.4), supplemented with 3% fatty acid- free BSA for 1 h at room temperature. The strips were then incubated with purified GST fusion proteins at a concentra- tion of 300 lgÆmL )1 in blocking buffer at room temperature for 3 h. After three washes in the binding buffer, PIP Strips were incubated for 3 h at room temperature with anti-GST (Nacalai Tesque, Kyoto, Japan) serum in the same buffer. Secondary antibody incubation and 3,3¢,5,5¢-tetra- methylbenzidine staining were performed to detect GST- tagged proteins bound to the phospholipid spots on the membrane. Liposome recruitment assay Liposome recruitment reactions were performed in 150 mm KCl and 50 mm Hepes (pH 7.5). Liposomes were made from a 1 : 1 mixture of PC and PE (Sigma Aldrich, Tokyo, Japan) in the presence or absence of PtdIns3P, PtdIns4P, K. Shiozawa et al. Phospholipid-binding activity of adaptor domains FEBS Journal 273 (2006) 4959–4971 ª 2006 The Authors Journal compilation ª 2006 FEBS 4967 PtdIns5P, or PtdIns (5% weight ratio to the base liposome; Sigma Aldrich). Purified GST fusion proteins (50 lgÆmL )1 ) were incubated with the liposomes (1 mg lipidÆmL )1 ). After 5 min of incubation at room temperature, membranes were recovered by centrifugation at 35 000 g at 4 °C with a Beckman TL-100 ultracentrifuge, rotor type TLA-100 (Beckman Coulter, Fullerton, CA, USA). The reaction was carried out on a 50 lL scale. Supernatants and pellets were resuspended in SDS sample buffer and analyzed by SDS ⁄ PAGE, followed by Coomassie brilliant blue staining. The gel images were analyzed using the imagej 1.31v soft- ware (http://rsb.info.nih.gov/ij/). QCM assay of in vitro protein–lipid interaction The in vitro liposome binding of a mutant of PEX1-ND was determined by frequency change in a 27-MHz QCM using an AFFINIX-Q (QCM2000) instrument (Initiam Co., Tokyo, Japan). A GST fusion protein of wild-type PEX1- ND was immobilized on a QCM gold electrode (diameter 4.5 mm, QCMST27) according to the manufacturer’s instructions. A 2 lL aliquot of a 50 lgÆmL )1 solution of GST–PEX1-ND was used. The electrode was washed sev- eral times in buffer containing 150 mm KCl and 50 mm Hepes (pH 7.45), soaked in the same buffer (2 mL cuvette), and then monitored continuously for QCM frequency chan- ges at 25 °C. Once the frequency had stabilized, 8 lLof the PC + PE (1 : 1) liposomes (1 mgÆmL )1 ) with or with- out PtdIns4P (5% weight ratio to the base liposome) were injected. Evolutionary trace analysis The four sequences of PEX1-ND orthologs and six sequences of VCP-ND orthologs were obtained from NCBI; they are PEX1_HUMAN (human PEX1; AAB99758), PEX1_MOUSE (Mus musculus PEX1; XML131895), PEX1_ARATH (Arabidopsis thaliana PEX1; AAG44817), PEX1_YEAST (yeast PEX1; CAA82041), VCP_MOUSE (Mus musculus VCP; NP_033529), VCP_HUMAN (human CAH70993), VCP_XENOPUS (Xenopus VCP; AAH74716), VCP_ZEBRAFISH (zebrafish VCP; NP_958889), CDC48_ YEAST (yeast cdc48; CAA98694) and VAT_THEAC (Thermoplasma acidophilum VAT; AAC45089). The multiple sequence alignment was produced using the pro- gram clustalx [60] and then refined manually. The evolu- tionary trace analysis [45] was then carried out on the Evolutionary Trace Server (tracesuite II) [46]. Here, the evolutionary trace method defined groups of proteins using an evolutionary time cut-off in a phylogenetic tree, and divided all residues of aligned sequences into three classes: conserved (residues invariant throughout the groups), class- specific (residues invariant within each group but that change between groups), and neutral (all others) (supple- mentary Fig. S1). Gaps were counted as an extra residue type. The class-specific residues are the most interesting in terms of the development of functional innovation. Trace residues were mapped onto the structure by pymol [61]. The surfaces of PEX1-ND and VCP-ND were further subjected to ODA analysis by using program pydock (J. Fernandez- Recio, unpublished results). Acknowledgements This work was supported by grants to MS and HH from the Japanese Ministry of Education, Science, Sports and Culture (Protein3000). We thank Dr Y. Fujiki for a vector containing PEX1 cDNA. We thank Dr M. H. J. Koch and Dr W. Schliebs for many valu- able discussions and for critical reading of the manu- script. References 1 van den Bosch H, Schutgens RB, Wanders RJ & Tager JM (1992) Biochemistry of peroxisomes. Annu Rev Biochem 61, 157–197. 2 Eckert JH & Erdmann R (2003) Peroxisome biogenesis. Rev Physiol Biochem Pharmacol 147, 75–121. 3 Wanders RJ & Waterham HR (2005) Peroxisomal dis- orders I: biochemistry and genetics of peroxisome bio- genesis disorders. Clin Genet 67, 107–133. 4 Imamura A, Tamura S, Shimozawa N, Suzuki Y, Zhang Z, Tsukamoto T, Orii T, Kondo N, Osumi T & Fujiki Y (1998) Temperature-sensitive mutation in PEX1 moderates the phenotypes of peroxisome defi- ciency disorders. Hum Mol Genet 7 , 2089–2094. 5 Geisbrecht BV, Collins CS, Reuber BE & Gould SJ (1998) Disruption of a PEX1–PEX6 interaction is the most common cause of the neurologic disorders Zellwe- ger syndrome, neonatal adrenoleukodystrophy, and infantile Refsum disease. Proc Natl Acad Sci USA 95, 8630–8635. 6 Tamura S, Okumoto K, Toyama R, Shimozawa N, Tsukamoto T, Suzuki Y, Osumi T, Kondo N & Fujiki Y (1998) Human PEX1 cloned by functional com- plementation on a CHO cell mutant is responsible for peroxisome-deficient Zellweger syndrome of complementation group I. Proc Natl Acad Sci USA 95, 4350–4355. 7 Reuber BE, Germain-Lee E, Collins CS, Morrell JC, Ameritunga R, Moser HW, Valle D & Gould SJ (1997) Mutations in PEX1 are the most common cause of per- oxisome biogenesis disorders. Nat Genet 17, 445–448. 8 Dodt G & Gould SJ (1996) Multiple PEX genes are required for proper subcellular distribution and stability of Pex5p, the PTS1 receptor: evidence that PTS1 protein import is mediated by a cycling receptor. J Cell Biol 135, 1763–1774. Phospholipid-binding activity of adaptor domains K. Shiozawa et al. 4968 FEBS Journal 273 (2006) 4959–4971 ª 2006 The Authors Journal compilation ª 2006 FEBS [...]... valosine-containing protein (VCP), colored according to the energy values of their respective optimal docking areas (ODAs) [46] Fig S3 PIP Strip assay of the phospholipid binding of the glutathione-S-transferase (GST) fusion protein with the other N-terminal domains of type II AAAATPases and the related adaptor proteins This material is available as part of the online article from http://www.blackwell-synergy.com... 460–465 Supplementary material The following supplementary material is available online: Fig S1 (A) Phylogenetic tree constructed from the four sequences of PEX1 N-terminal domain (ND) orthologs and six sequences of valosine-containing protein (VCP)-ND orthologs (B) Location of trace residues in PEX1- ND and VCP-ND Fig S2 Surface representations of N-terminal domains (NDs) of PEX1 and valosine-containing protein... peroxin, Pex1p, interacts with Pex6p Biochem Biophys Res Commun 245, 883–886 12 Birschmann I, Rosenkranz K, Erdmann R & Kunau WH (2005) Structural and functional analysis of the interaction of the AAA-peroxins Pex1p and Pex6p FEBS J 272, 47–58 13 Platta HW, Grunau S, Rosenkranz K, Girzalsky W & Erdmann R (2005) Functional role of the AAA peroxins in dislocation of the cycling PTS1 receptor back to the cytosol... al (2005) Structural and Phospholipid-binding activity of adaptor domains mechanistic studies of VPS4 proteins EMBO J 24, 3658–3669 69 Rizo J & Sudhof TC (1998) C2 -domains, structure and function of a universal Ca2+-binding domain J Biol Chem 273, 15879–15882 70 Takasu H, Jee JG, Ohno A, Goda N, Fujiwara K, Tochio H, Shirakawa M & Hiroaki H (2005) Structural characterization of the MIT domain from human... (2004) Structure of the N-terminal domain of PEX1 AAA-ATPase: characterization of a putative adaptorbinding domain J Biol Chem 279, 50060–50069 33 Zhang X, Shaw A, Bates PA, Newman RH, Gowen B, Orlova E, Gorman MA, Kondo H, Dokurno P, Lally J et al (2000) Structure of the AAA ATPase p97 Mol Cell 6, 1473–1484 34 Yu RC, Jahn R & Brunger AT (1999) NSF N-terminal domain crystal structure: models of NSF function... Meyer HH & Rapoport TA (2003) Function of the p97–Ufd1–Npl4 complex in retrotranslocation from the ER to the cytosol: dual recognition of nonubiquitinated polypeptide segments and polyubiquitin chains J Cell Biol 162, 71–84 27 Meyer HH, Shorter JG, Seemann J, Pappin D & Warren G (2000) A complex of mammalian ufd1 and npl4 links the AAA-ATPase, p97, to ubiquitin and nuclear transport pathways EMBO J... B, Butland G, Greenblatt J, Emili A & Houry WA (2006) Formation of a distinctive complex between the inducible bacterial lysine decarboxylase and a novel AAA+ ATPase J Biol Chem 281, 1532–1546 Ikonomov OC, Sbrissa D, Yoshimori T, Cover TL & Shisheva A (2002) PIKfyve kinase and SKD1 AAA ATPase define distinct endocytic compartments Only PIKfyve expression inhibits the cell-vacuolating activity of Helicobacter... & Weis WI (1999) Crystal structure of the amino-terminal domain of N-ethylmaleimide-sensitive fusion protein Nat Cell Biol 1, 175–182 36 Coles M, Diercks T, Liermann J, Groger A, Rockel B, Baumeister W, Koretke KK, Lupas A, Peters J & FEBS Journal 273 (2006) 4959–4971 ª 2006 The Authors Journal compilation ª 2006 FEBS 4969 Phospholipid-binding activity of adaptor domains 37 38 39 40 41 42 43 44 45... The solution structure of VAT-N reveals a ‘missing link’ in the evolution of complex enzymes from a simple betaalphabetabeta element Curr Biol 9, 1158–1168 Vogel C, Berzuini C, Bashton M, Gough J & Teichmann SA (2004) Supra -domains: evolutionary units larger than single protein domains J Mol Biol 336, 809–823 Dreveny I, Kondo H, Uchiyama K, Shaw A, Zhang X & Freemont PS (2004) Structural basis of the. .. D & Rapoport TA (2004) A membrane protein complex mediates retrotranslocation from the ER lumen into the cytosol Nature 429, 841–847 31 Ye Y, Shibata Y, Kikkert M, van Voorden S, Wiertz E & Rapoport TA (2004) Inaugural article: Recruitment of the p97 ATPase and ubiquitin ligases to the site of retrotranslocation at the endoplasmic reticulum membrane Proc Natl Acad Sci USA 102, 14132– 14138 32 Shiozawa . The common phospholipid-binding activity of the N-terminal domains of PEX1 and VCP/p97 Kumiko Shiozawa 1 , Natsuko. representation of the binding of the N-terminal domain (ND) of PEX1 to the peroxisomal membrane. PEX1 is pre- dicted to interact with the PEX6 portion of AAA domains.