Peroxisome division in the yeast Yarrowia lipolytica is regulated by a signal from inside the peroxisome

In the “growth and division” model of peroxisome biogenesis, peroxisomes grow by the posttranslational import of membrane and matrix proteins synthesized on cytosolic polyribosomes and divide to form daughter peroxisomes (Lazarow and Fujiki, 1985). These daughter peroxisomes undergo further growth and division (Purdue and Lazarow, 2001). Although a large body of evidence supports this model (Sacksteder and Gould, 2000; Purdue and Lazarow, 2001; Titorenko and Rachubinski, 2001b), it remains unclear whether the processes of peroxisome growth and division are coordinated. In particular, the model does not specify whether only mature peroxisomes, which contain the complete set of membrane and matrix proteins, are competent for division and, therefore, whether peroxisome growth and maturation occur before the completely assembled mature peroxisomes undergo division. Alternatively, the division of immature peroxisomes carrying only minor amounts of matrix and/or membrane proteins may precede their maturation, which is accomplished by membrane and matrix protein import. This timing of events in peroxisomal development was observed in the yeast Candida boidinii (Veenhuis and Goodman, 1990). In this yeast, the division of small peroxisomes containing only a few matrix proteins occurs before the numerous immature peroxisomes undergo enlargement and maturation by uptake of the bulk of matrix proteins. Another possible scenario is that both mature and immature peroxisomes undergo efficient division, and therefore, the ability of peroxisomes to divide does not depend on whether or not they carry the complete set of matrix and membrane proteins.

Data on purification, protein profiling, and electron microscopic analysis of mammalian and yeast peroxisomes have provided important information regarding the process of peroxisomal development. It was established that the population of peroxisomes in a cell consists of several peroxisomal subforms that differ in their size, morphology, buoyant density, and protein composition (Lüers et al., 1993; van Roermund et al., 1995; Wilcke et al., 1995; Titorenko et al., 1996, 2000; Faber et al., 1998). Furthermore, data on the in vivo dynamics of peroxisomal protein localization to several peroxisomal subforms demonstrated that these subforms also differ in their import competency for various proteins. In fact, newly synthesized peroxisomal proteins in mammalian (Heinemann and Just, 1992) and yeast (Titorenko et al., 2000) cells are imported primarily into small peroxisomal vesicles of intermediate buoyant density that subsequently convert to mature peroxisomes of high density. Recent findings in human (South and Gould, 1999; Gould and Valle, 2000) and yeast (Snyder et al., 1999; Subramani et al., 2000; Titorenko et al., 2000) cells have suggested that several peroxisomal subforms are organized into a multistep peroxisome assembly pathway. The pathway operates by the conversion of subforms in a temporally ordered manner, involves the stepwise import of distinct subsets of matrix and membrane proteins into different intermediates along the pathway, and leads to the assembly of mature peroxisomes (Gould and Valle, 2000; Subramani et al., 2000; Titorenko and Rachubinski, 2001a,b).

The peroxisome assembly pathway operating in the yeast Yarrowia lipolytica leads to the formation of mature peroxisomes, P6 (Titorenko et al., 2000). In this yeast, five immature peroxisomal subforms, termed P1–P5, differ in their import competency for various proteins and are related through a time-ordered conversion of one subform to another. The current study utilizes several approaches to investigate whether growth and division of immature peroxisomal vesicles and mature peroxisomes are coordinated in Y. lipolytica cells. Furthermore, we have previously demonstrated that peroxisome division in this yeast is regulated by the intraperoxisomal peripheral membrane protein Pex16p (Eitzen et al., 1997), a member of the peroxin family of proteins required for peroxisome assembly, division, and inheritance (Sacksteder and Gould, 2000; Subramani et al., 2000; Purdue and Lazarow, 2001; Titorenko and Rachubinski, 2001b). Here, we further investigate the role of Pex16p in peroxisome division. We describe an unusual mechanism that controls peroxisome division from inside the peroxisome. A temporally and spatially regulated interaction between Pex16p and a heteropentameric complex of acyl-CoA oxidase (Aox), one of the proteins imported into the early peroxisomal precursor P2, plays a pivotal role in this control mechanism.

Aox is present as a 443-kD heteropentameric complex composed of one polypeptide chain of each of its five subunits, Aox1p–Aox5p, in the matrix of mature peroxisomes P6 (Titorenko et al., 2002). After centrifugation of osmotically lysed mature peroxisomes, this matrix form of Aox was recovered in the supernatant fraction (Fig. 1 A). In addition, about half of the peroxisome-bound pool of each Aox subunit could be pelleted during centrifugation (Fig. 1 A). The observed recovery of Aox subunits in the pelletable fraction was not due to the greater resistance of mature peroxisomes to osmotic lysis. In fact, none of the three other most abundant peroxisomal matrix proteins tested, namely isocitrate lyase (ICL), thiolase (THI), and malate synthase (MLS), could be pelleted during centrifugation of osmotically lysed mature peroxisomes (Fig. 1 B). The pelletable form of each Aox subunit could float out of the most dense sucrose during centrifugation to equilibrium in a sucrose density gradient (Fig. 2 A). In contrast, temperature-induced aggregates of peroxisomal matrix proteins remained at the bottom of the gradient (Fig. 2 B). Accordingly, all pelletable Aox subunits in mature peroxisomes were present as membrane-associated forms rather than as aggregates. Furthermore, protease protection experiments revealed that all five Aox subunits were degraded by trypsin only when the membrane of mature peroxisomes was disrupted by Triton X-100, whereas the peripheral membrane proteins Pex1p and Pex6p on the cytosolic surface of peroxisomes (Titorenko et al., 2000) were sensitive to trypsin digestion even in the absence of the detergent (Fig. 2 C). Therefore, all Aox subunits were present as membrane-enclosed forms. Extraction of the membrane-associated Aox subunits with various solubilizing agents showed that they fractionated as peripheral membrane proteins that were solubilized completely by either 1 M NaCl or 0.1 M Na₂CO₃, pH 11.0 (Fig. 2 D). Thus, in mature peroxisomes of wild-type cells, Aox was equally distributed between the matrix and the matrix face of the membrane. In contrast, none of the Aox subunits in immature peroxisomes P2–P4, and only a minor portion of each subunit in immature peroxisomes P5, were membrane bound (Fig. 1 A). Immature peroxisomes P1 lack Aox (Titorenko et al., 2000). Immunoaffinity chromatography of membrane proteins from mature peroxisomes of the wild-type strain showed that all five Aox subunits coimmunoprecipitated under native conditions with anti-Aox1p (Fig. 1 F), -Aox3p, or -Aox5p (unpublished data) antibodies. Therefore, like the Aox complex in the peroxisomal matrix, the complex at the matrix face of the peroxisomal membrane in mature peroxisomes of wild-type cells includes all five Aox subunits. Taken together, these results strongly suggest that at the last step of assembly of mature peroxisomes from immature intermediates, a significant portion of the Aox complex inside the peroxisome relocates from the matrix to the membrane.

The relocation of the Aox complex from the matrix to the matrix face of the membrane of mature peroxisomes requires two Aox subunits, Aox4p and Aox5p, but not the three other subunits of the complex. Indeed, none of the Aox subunits was associated with the peroxisomal membrane in the mutant strains aox4KO and aox5KO deleted individually for the AOX4 and AOX5 genes, respectively (Fig. 1, D and E). In contrast, lack of Aox1p (Fig. 1 C), Aox2p, or Aox3p (unpublished data) did not impair the redistribution of other Aox subunits from the matrix to the membrane of mature peroxisomes. In the mature peroxisomes of aox1KO, aox2KO, and aox3KO cells, all remaining Aox subunits form a complex both in the matrix (Titorenko et al., 2002) and at the matrix face of the membrane (Fig. S1).

The membrane-associated Aox complex of mature peroxisomes of wild-type cells coimmunoprecipitated under native conditions with the peroxin Pex16p (Fig. 1 F), which is attached to the matrix face of the peroxisomal membrane (Eitzen et al., 1997). Neither Aox nor Pex16p was recovered in the flowthrough when native immunoprecipitation was done with anti-Aox1p or anti-Pex16p antibodies (Fig. 1 F). Thus, the membrane-bound pools of both Aox and Pex16p in mature peroxisomes of wild-type cells were present only as components of a complex, and none of these proteins could be found in its free form. No other peroxisomal membrane peroxin tested, including Pex2p (Eitzen et al., 1996), Pex5p (Szilard et al., 1995), and Pex8p (Smith et al., 1997), interacted with the membrane-bound Aox or Pex16p (Fig. 1 F).

The different Aox subunits and Pex16p are present in equimolar amounts in their membrane-associated complex, as judged by quantitation of their stoichiometry in l-[³⁵S]methionine–labeled complex immunoprecipitated from mature peroxisomes of wild-type cells (Fig. S2, A and C). No other radiolabeled membrane protein coimmunoprecipitated with the components of the Aox–Pex16p complex under native conditions (Fig. S2 B). Whereas the molecular mass of the Aox complex recovered from the matrix of mature peroxisomes was ∼443 kD (Fig. S3 A) (Titorenko et al., 2002), the molecular mass of the Aox–Pex16p complex attached to the matrix face of the peroxisomal membrane in wild-type cells was ∼900 kD (Fig. S3 B). From these observations and a consideration of the molecular masses of each Aox subunit (∼80 kD) and Pex16p (∼45 kD), we conclude that relocation of a significant portion of the Aox complex from the matrix to the membrane at the last step of the assembly of mature peroxisomes leads to the formation of a supramolecular complex containing two molecules of Aox complex and two molecules of Pex16p. Relocation of Aox complex to the matrix face of the peroxisomal membrane requires two Aox subunits, Aox4p and Aox5p, and Pex16p. In fact, no membrane-bound form of the Aox complex was detected in mature peroxisomes recovered from mutant cells lacking any of these three proteins (Fig. 1, D, E, and G).

Overexpression of the PEX16 gene by the highly active THI promoter in the strain pex16-TH results in a reduced number of greatly enlarged mature peroxisomes (Fig. 3) (Eitzen et al., 1997). Similar to Aox in mature peroxisomes of wild-type cells, a significant portion of all five Aox subunits inside the mature peroxisomes of pex16-TH cells is relocated from the matrix to the membrane (Fig. 1 H), where they form a complex with each other and with Pex16p. In fact, all five Aox subunits and Pex16p recovered from the membranes of mature peroxisomes of pex16-TH cells coimmunoprecipitated under native conditions with anti-Aox3p or anti-Pex16p antibodies (Fig. 1 I). However, unlike the membrane-attached Pex16p in mature peroxisomes of wild-type cells, which is present only as a component of the Aox–Pex16p complex (Fig. 1 F), most of the Pex16p in mature peroxisomes of pex16-TH cells cannot be immunoprecipitated under native conditions with anti-Aox3p antibodies (Fig. 1 I) and, therefore, does not interact with the membrane-bound Aox complex.

These findings suggest that scission of the membrane of mature peroxisomes, which results in their division, can occur only if all the Pex16p inside mature peroxisomes is titrated by its interaction with Aox complex that has relocated from the matrix to the membrane. Thus, Pex16p negatively regulates the membrane scission event required for the division of mature peroxisomes. In wild-type cells, the Aox complex that has relocated from the matrix to the membrane of mature peroxisomes interacts with Pex16p and terminates its negative effect on peroxisome division. This hypothesis is further supported by the observation that lack of either Aox4p or Aox5p prevented such a relocation of Aox complex (Fig. 1, D and E) and resulted in fewer, but greatly enlarged, mature peroxisomes (Fig. 3). On the other hand, lack of Aox1p, Aox2p, or Aox3p did not impair the redistribution of Aox complex from the matrix to the membrane inside mature peroxisomes (Fig. 1 C) and did not affect the division of peroxisomes (Fig. 3). All Pex16p in the membrane of mature peroxisomes recovered from aox1KO, aox2KO, or aox3KO cells was titrated by its interaction with the Aox complex (Fig. S1).

Importantly, the accumulation of greatly enlarged peroxisomes in aox4KO and aox5KO cells was not due to a deficiency in peroxisomal fatty acid β-oxidation. In fact, no mutation knocking out a single Y. lipolytica AOX gene affected the enzymatic activity of Aox, one of the key enzymes of peroxisomal β-oxidation, or impaired the utilization of oleic acid as a carbon source (Wang et al., 1999). Thus, the observed changes in peroxisome size and number in aox4KO and aox5KO cells (Fig. 3) cannot be attributed to a defect in the so-called metabolic control of peroxisome abundance (Chang et al., 1999), which operates in yeast, mammalian, and human cells (Fan et al., 1998; Poll-Thé et al., 1988; Chang et al., 1999; Smith et al., 2000; van Roermund et al., 2000).

Morphometric analysis of random electron microscopy sections was used to evaluate the dynamics of change in the size and number of peroxisomes in wild-type and aox mutant cells transferred from glucose- to oleic acid–containing medium. In wild-type cells of Y. lipolytica, such a transfer greatly increases peroxisome size and number (Fig. S4) (Smith et al., 2000). Data from morphometric analysis further confirmed that the inability of the Aox complex to relocate from the matrix to the membrane at the last step of the assembly of mature peroxisomes impairs their ability to divide. During the first 3 h of incubation in oleic acid–containing medium, the size of peroxisomes in wild-type, aox4KO, and aox5KO cells significantly increased (Fig. 4 A; Figs. S4–S6), while their number did not change (Fig. 4 B and Figs. S4–S6). Similar dynamics of change in peroxisome size and number by 3 h after the shift from glucose- to oleic acid–containing medium was observed in aox1KO, aox2KO, and aox3KO cells (unpublished data). By 6 and 9 h after the shift to oleic acid–containing medium, the number of peroxisomes in wild type (Fig. 4 B and Fig. S4), and in aox1KO, aox2KO, and aox3KO cells (unpublished data), dramatically increased, attaining 14.6 ± 2.0 peroxisomes per μm³ of cell section volume. Concomitantly, the proportion of small peroxisomes in these cells gradually increased, leading to significant variability in peroxisome size by 9 h after the shift (Fig. 4 A and Fig. S4). In contrast, the size of peroxisomes in aox4KO and aox5KO mutant cells continued to increase by 6 and 9 h after the shift to oleic acid–containing medium (Fig. 4 A and Figs. S5 and S6), with only greatly enlarged peroxisomes visible by 9 h after the shift. During the entire period of incubation after the shift from glucose- to oleic acid–containing medium, the number of peroxisomes in aox4KO and aox5KO cells did not change significantly, attaining only 1.7 ± 0.3 and 2.3 ± 0.4 peroxisomes per μm³ of cell section volume, respectively, by the end of the incubation (Fig. 4 B and Figs. S5 and S6). Thus, the inability of the Aox complex lacking either the Aox4p or the Aox5p subunit to relocate from the matrix to the membrane and, therefore, to titrate all membrane-bound Pex16p results in the inability of Aox to prevent the negative effect of Pex16p on the division of mature peroxisomes.

The pex16KO mutant strain deleted for the PEX16 gene accumulates a considerable number of very small peroxisomes (Eitzen et al., 1997) that are reminiscent of immature peroxisomal vesicles (Titorenko et al., 2000). These immature peroxisomal vesicles can be pelleted only by centrifugation at 200,000 g, whereas mature peroxisomes are completely pelletable even at 20,000 g (Titorenko et al., 1998, 2000). In wild-type cells, immature peroxisomes represent only a minor portion of the peroxisome population, as judged from the relative distribution of MLS, a protein marker of both immature and mature peroxisomes, between the low-speed (20,000 g) and high-speed (200,000 g) organellar fractions (Fig. 5 A). Our published data on the relative distribution of other peroxisomal protein markers and from immunofluorescence microscopy support this conclusion (Titorenko et al., 1998, 2000). In pex16KO cells, the steady-state level of immature peroxisomes is dramatically increased compared with their level in wild-type cells, with about half of the peroxisome population present as immature peroxisomal vesicles (Fig. 5 A). These data strongly suggest that Pex16p negatively regulates the membrane scission event required for the division of early peroxisomal precursors, the immature peroxisomal vesicles P1–P5. Inside the immature peroxisomes of wild-type cells, the Aox complex cannot abolish the negative effect of Pex16p on scission of the membrane, because in this peroxisomal population, Aox resides only in the matrix (Fig. 1 A). Therefore, the lack of Pex16p in pex16KO cells results in the excessive proliferation of immature peroxisomal vesicles.

Pulse-chase analysis of the trafficking of MLS in vivo demonstrated that the excessive proliferation of immature peroxisomes in pex16KO cells significantly decreased the rate and efficiency of their conversion to mature peroxisomes (Fig. 5, B and C). In contrast, the pex16KO mutation did not abolish the import of MLS (Fig. 5, B and C) or of other peroxisomal proteins, including Aox1p, -2p, -3p, -4p, -5p, ICL, THI, and Pex2p (unpublished data), from the cytosol to the matrix of immature peroxisomes. The excessive proliferation of immature peroxisomal vesicles could be suggested to substantially decrease the concentration of vesicle-associated complexes required for the conversion of these vesicles to mature peroxisomes.

Comparison of the spectra and relative distributions of peroxisomal matrix and membrane proteins demonstrated that even the earliest intermediates in the multistep peroxisome assembly pathway, the immature peroxisomal vesicles P1 and P2, contain most of the peroxisomal membrane proteins (PMPs) associated with mature peroxisomes, P6 (Fig. 6 A). P1 and P2 undergo fusion to generate larger and more dense immature peroxisomal vesicles, P3 (Titorenko et al., 2000), containing PMPs derived from both fusion partners. The quantities of PMPs in P4, P5, and P6 peroxisomes were significantly lower than in P1, P2, and P3 peroxisomes, and gradually decreased from P4 to P6 (Fig. 6 B). In contrast, only a few matrix proteins found in mature peroxisomes were seen in the immature peroxisomal vesicles P1, P2, and P3 (Fig. 6 A). Most matrix proteins were associated with P4, P5, and P6, and the complexity of their spectra increased from P4 to P6 (Fig. 6 A). The quantities of matrix proteins in P4, P5, and P6 peroxisomes were significantly higher than in P1, P2, and P3 peroxisomes, and gradually increased from P4 to P6 (Fig. 6 B). Taken together, these results strongly suggest that the stepwise import of distinct subsets of matrix proteins into different immature intermediates along the peroxisome assembly pathway provides them with an increasing fraction of the matrix proteins present in mature peroxisomes, P6 (see Fig. 8). P6 peroxisomes contain the highest levels of matrix proteins (Fig. 6, A and B).

Based on these findings, we hypothesized that the observed relocation of Aox complex from the matrix to the membrane of mature peroxisomes (Fig. 1 A) is due to an increase in the total mass of matrix proteins above a critical level, and that overloading mature peroxisomes with matrix proteins is a major factor in the relocation of Aox. To test this hypothesis, we attempted to reconstruct in vitro the relocation of Aox from the matrix to the membrane and its interaction with membrane-bound Pex16p by reconstituting peroxisomal liposomes from matrix proteins, detergent-solubilized PMPs, and membrane lipids of mature peroxisomes. We reconstituted four types of peroxisomal liposomes, termed PLA to PLD (Fig. 7 A). PLA were reconstituted from matrix proteins immunodepleted of Aox, Aox complex purified from the matrix of mature peroxisomes by immunoaffinity chromatography, detergent-solubilized PMPs immunodepleted of Aox, and membrane lipids. Each component used for the reconstitution of PLA was recovered from 1 mg (1 equivalent) of peroxisomes. PLB were reconstituted from 0.2 equivalent of matrix proteins immunodepleted of Aox, 1 equivalent of purified Aox complex, 1 equivalent of detergent-solubilized PMPs immunodepleted of Aox, and 1 equivalent of membrane lipids. PLC were reconstituted from 0.2 equivalent of matrix proteins immunodepleted of Aox, 5 equivalents of purified Aox complex, 1 equivalent of detergent-solubilized PMPs immunodepleted of Aox, and 1 equivalent of membrane lipids. PLD were reconstituted from 1 equivalent of matrix proteins immunodepleted of Aox, 1 equivalent of purified Aox complex, 1 equivalent of detergent-solubilized PMPs immunodepleted of both Aox and Pex16p, and 1 equivalent of membrane lipids.

Electron microscopy revealed that all four types of peroxisomal liposomes were bound by a single membrane (Fig. S7). Comparison of the relative distributions of peroxisomal matrix and membrane proteins demonstrated no significant difference between mature peroxisomes and PLA (Fig. 7 B). In contrast, the quantities of matrix proteins in PLB were significantly lower than in mature peroxisomes or PLA (Fig. 7 B). Even though the total amounts of all five Aox subunits in mature peroxisomes, PLA, and PLB were similar (Fig. 7 C), the Aox complex was attached to the membrane only in mature peroxisomes and PLA (Fig. 7 C), in which membrane-bound Aox formed a complex with Pex16p (Fig. 7 D). In contrast, no Aox subunits were attached to the membrane inside PLB (Fig. 7 C), in which Pex16p was present only in its free form (Fig. 7 D). Like the Aox complex in mature peroxisomes, Aox in PLA and PLB was resistant to digestion by external protease added to intact peroxisomes or liposomes, i.e., was present in membrane-enclosed form (Fig. 7 E). Because PLA and PLB differ only in their total amount of matrix proteins but contain the same amount of Aox, the increase in total mass of matrix proteins above a critical level can cause the observed relocation of Aox complex from the matrix to the membrane inside mature peroxisomes (Fig. 1 A).

A comparative analysis of PLB and PLC, which differ in their amounts of Aox (Fig. 7 C) but contain similar low amounts of other matrix proteins (Fig. 7 B), showed that a significant increase in the amount of matrix-associated Aox did not result in its relocation to the membrane (Fig. 7 C). Taken together, these findings suggest that overloading mature peroxisomes with matrix proteins other than Aox is a major factor in the relocation of Aox complex from the matrix to the membrane.

Finally, although PLA and PLD were loaded with very similar high amounts of matrix proteins (Fig. 7 B), including Aox (Fig. 7 C), the Aox complex was attached to the membrane only inside PLA (Fig. 7 C). PLA contain Pex16p, whereas PLD lack this membrane-bound peroxin (Fig. 7 C). Therefore, Pex16p is the only attachment factor for the Aox complex in the PLA liposomes containing high amounts of matrix proteins and, perhaps, also in mature peroxisomes P6 (Fig. 1 G) containing the greatest percentage of matrix proteins as compared with immature peroxisomal vesicles P1–P5 (Fig. 6, A and B).

It should be noted that the above data cannot rule out the possibility that a distinct, yet unknown, matrix protein or a limited set of such proteins rather than protein mass in the peroxisomal matrix initiates the relocation of Aox complex from the matrix to the membrane, thereby terminating the negative action of Pex16p on peroxisome division. Although Aox in the matrix of mature peroxisomes does not form a stable complex with any protein (Titorenko et al., 2002), even its transient interaction with a specific, yet unknown, soluble factor may promote the redistribution of Aox from the matrix to the membrane. Alternatively, overloading mature peroxisomes with matrix proteins may ultimately lead to the relocation of an unknown specific factor from the matrix to the membrane. Once bound to the peroxisomal membrane, this specific factor may cause perturbations in its physical properties, thereby promoting the assembly of the Aox–Pex16p complex at the matrix face of the membrane. The development of a reliable in vitro assay for reconstructing the relocation of Aox from the matrix to the membrane and its interaction with membrane-bound Pex16p creates the opportunity to test individual peroxisomal matrix proteins for their ability to initiate these processes.

The results of this study and our published data (Titorenko et al., 1998, 2000) can be summarized in the following model for peroxisome growth and division in Y. lipolytica (Fig. 8). Six subforms of peroxisomes, termed P1–P6, are organized into a multistep peroxisome assembly pathway (Titorenko et al., 2000). The pathway operates by conversion of the subforms in a temporally ordered manner from P1 to P6 and leads to the formation of mature peroxisomes, P6, carrying the complete set of matrix and membrane proteins (Titorenko and Rachubinski, 2001a,b). The earliest intermediates in the assembly pathway, the immature peroxisomal vesicles P1 and P2, contain most of the PMPs associated with mature peroxisomes (Fig. 6 A). P1 and P2 are competent for the import of a limited subset of matrix proteins (Titorenko et al., 2000) and, therefore, contain only a few matrix proteins found in mature peroxisomes (Fig. 6 A). P1 and P2 undergo fusion to generate larger and more dense immature peroxisomal vesicles, P3 (Titorenko and Rachubinski, 2000; Titorenko et al., 2000). Conversion of P3 to mature peroxisomes, P6, proceeds through several consecutive steps. At each of these steps, the import of a limited subset of matrix proteins results in the formation of a distinct peroxisomal subform that is larger and more dense than its precursor (Fig. 6 A; Titorenko et al., 2000).

The amounts of immature peroxisomal vesicles P1–P5 are not more than 1–2% that of mature peroxisomes, based on protein mass (Titorenko et al., 1998, 2000). The peroxin Pex16p, which is attached to the matrix surface of the membrane (Eitzen et al., 1997), negatively regulates the division of immature peroxisomal vesicles, thereby preventing their excessive proliferation. The Pex16p-dependent negative control of the membrane scission event required for the division of immature peroxisomal vesicles is essential for cell growth in oleic acid–containing medium. In fact, lack of Pex16p results in the excessive proliferation of immature peroxisomal precursors (Fig. 5 A; Eitzen et al., 1997) and impairs the utilization of oleic acid as a carbon source (Eitzen et al., 1997).

The stepwise import of distinct subsets of matrix proteins into P1–P5 intermediates provides them with an increasing fraction of the matrix proteins present in mature peroxisomes (Fig. 6, A and B). This increase in the total mass of matrix proteins above a critical level causes the redistribution of the heteropentameric complex of Aox, which is imported into the early intermediate P2 (Titorenko et al., 2000), from the matrix to the matrix surface of the membrane (Fig. 1, A and F). A significant redistribution of Aox complex from the matrix to the membrane occurs only in mature peroxisomes (Fig. 1 A), which contain the greatest percentage of matrix proteins (Fig. 6, A and B). Overloading mature peroxisomes with matrix proteins other than Aox can be a major factor in the relocation of Aox complex to the membrane. The available data cannot rule out the possibility that a distinct, yet unknown, matrix protein, rather than protein mass, in the peroxisomal matrix initiates the relocation of Aox complex from the matrix to the membrane.

Inside mature peroxisomes, the membrane-bound pool of Aox complex interacts, via its Aox4p and Aox5p subunits, with Pex16p (Fig. 1 F). This interaction leads to the formation of a supramolecular complex containing two molecules of Aox complex and two molecules of Pex16p (Figs. S2 and S3) and terminates the negative action of Pex16p on scission of the peroxisomal membrane, thereby allowing mature peroxisomes to divide. The temporally and spatially regulated interaction between membrane-attached Aox and Pex16p ensures the temporal and spatial separation of the processes of peroxisome assembly and division in Y. lipolytica. Such a separation may provide an important advantage for the efficient, stepwise assembly of mature, metabolically active peroxisomes.

A combination of morphometric electron microscopic analysis, pulse-chase analysis of the trafficking of peroxisomal proteins in vivo, and the isolation and protein profiling of structurally distinct peroxisomal subforms has convincingly demonstrated that yeast peroxisomes do not grow and divide at the same time (Veenhuis and Goodman, 1990; Tan et al., 1995; Titorenko et al., 2000; this study). It seems that evolution has generated at least two different temporal patterns of peroxisome growth and division. In the yeast C. boidinii (Veenhuis and Goodman, 1990), the massive proliferation of immature peroxisomal vesicles containing only minor amounts of matrix proteins is a primary event in peroxisomal development. This significant increase in the number of immature peroxisomes by their division precedes the growth of these early peroxisomal precursors by membrane and matrix protein import and their conversion to mature organelles containing the complete set of peroxisomal proteins (Veenhuis and Goodman, 1990). We demonstrated that the timing of events of peroxisome growth and division is different in the yeast Y. lipolytica. In this organism, the growth of immature peroxisomal vesicles, which is accomplished by the import of matrix proteins, and their development into mature peroxisomes occur before completely assembled mature peroxisomes undergo division (this study; Titorenko et al., 2000). Similar temporal patterns of peroxisome growth and division have been observed for the yeast Hansenula polymorpha (Tan et al., 1995).

In human cells, both immature peroxisomal vesicles and mature peroxisomes are proposed to be able to divide (Gould and Valle, 2000). However, the division of immature peroxisomes before their growth and maturation by peroxisomal protein import can only be seen in some peroxin-deficient human fibroblasts after reactivation or reexpression of an originally defective peroxin-encoding gene (Matsuzono et al., 1999; South and Gould, 1999; Sacksteder and Gould, 2000). On the other hand, in normal human cells, growth of immature peroxisomal vesicles by membrane and matrix protein import, resulting in their conversion to mature peroxisomes, may occur before peroxisomes undergo division (Gould and Valle, 2000).

This study and our published data (Eitzen et al., 1997; Smith et al., 2000) provide evidence that the membrane scission event required for peroxisome division in Y. lipolytica is regulated by two mechanisms. Both mechanisms control peroxisome division in response to a specific signal transmitted from inside the peroxisome. One mechanism acts through the Pex16p- and Aox-dependent intraperoxisomal signaling cascade (Fig. 8). In this cascade, the ability of Pex16p to inhibit membrane scission is inversely proportional to the level of membrane-bound Aox, which, in turn, depends on the intraperoxisomal level of matrix proteins other than Aox. Only inside mature peroxisomes, which are assembled by the stepwise import of distinct subsets of matrix proteins into different immature peroxisomal vesicles, does the total mass of matrix proteins other than Aox reach its critical level. This triggers the relocation of a significant portion of Aox from the matrix to the matrix face of the peroxisomal membrane, ultimately terminating the negative action of Pex16p on membrane scission. Importantly, whereas the pex16-TH, aox4KO, and aox5KO mutations impair the Pex16p- and Aox-dependent control of peroxisome division and result in a reduced number of greatly enlarged mature peroxisomes (Fig. 3) (Eitzen et al., 1997), they do not affect metabolic flux through the peroxisomal fatty acid β-oxidation pathway (Eitzen et al., 1997; Wang et al., 1999). On the other hand, loss of the activity of multifunctional enzyme type 2, one of the key enzymes of peroxisomal β-oxidation, but not the absence of this protein, causes pronounced changes in peroxisome size and number in Y. lipolytica (Smith et al., 2000). Therefore, the Pex16p- and Aox-dependent control of peroxisome division in Y. lipolytica coexists with another mechanism regulating scission of the peroxisomal membrane from inside the peroxisome, the so-called metabolic control of peroxisome abundance (Chang et al., 1999). This second mechanism depends on metabolic flux through the peroxisomal fatty acid β-oxidation pathway and regulates peroxisome division in yeast (Smith et al., 2000; van Roermund et al., 2000), mammalian (Fan et al., 1998; Poll-Thé et al., 1988), and human (Chang et al., 1999) cells. The metabolic control of peroxisome abundance may be due to the ability of peroxisomes to generate a signaling molecule, perhaps an intermediate of peroxisomal fatty acid β-oxidation, that initiates a cascade of events ultimately promoting peroxisome division (van Roermund et al., 2000; Li and Gould, 2002).

Controlling the size and number of different organelles is essential to the normal physiology and the viability of cells. Here we have described an unusual mechanism that controls the division of peroxisomes from within the peroxisome itself. We have shown that the temporally and spatially regulated interaction between a peroxin, Pex16p, required for peroxisome biogenesis and the heteropentameric complex of the peroxisomal β-oxidation enzyme Aox plays a pivotal role in the control of peroxisome division in the yeast Y. lipolytica. A challenge for the future will be to understand how perturbations in the physical properties of the peroxisomal membrane promote the membrane scission event required for peroxisome division and how the Pex16p- and Aox-dependent intraperoxisomal signaling cascade triggers this process.

The Y. lipolytica wild-type strain P01d (Wang et al., 1999), the mutant strains pex16KO and pex16-TH (Eitzen et al., 1997), the single AOX gene knock-out strains (Wang et al., 1999), and the media, growth conditions, and genetic techniques for Y. lipolytica (Szilard et al., 1995) have been previously described. Antibodies to the Aox1p, -3p, and -5p subunits of the Aox complex (Titorenko et al., 2002), Pex2p (Eitzen et al., 1996), Pex5p (Szilard et al., 1995), Pex16p (Eitzen et al., 1997), ICL (Titorenko et al., 1998), THI (Szilard et al., 1995), and MLS (Titorenko et al., 1998) have been described.

Subcellular fractionation of Y. lipolytica cells grown in oleic acid–containing medium (Szilard et al., 1995), isolation of highly purified mature peroxisomes P6 (Titorenko et al., 1998), and purification of immature peroxisomes P1–P5 (Titorenko et al., 2000) were performed as described previously.

Covalent coupling of affinity-purified antibodies to protein A–Sepharose was performed as described previously (Xu et al., 1998). For immunoaffinity chromatography under native conditions, peroxisomal matrix proteins recovered in the supernatant fraction after centrifugation of osmotically lysed peroxisomes and peroxisomal liposomes were diluted with an equal volume of 50 mM Tris-HCl, pH 7.5, buffer containing 300 mM NaCl, 1% (vol/vol) Triton X-100, and 2× protease inhibitor cocktail (PIC) (Szilard et al., 1995). The pellets of PMPs recovered after centrifugation of osmotically lysed peroxisomes and peroxisomal liposomes were resuspended in 25 mM Tris-HCl, pH 7.5, buffer containing 150 mM NaCl, 0.5% (vol/vol) Triton X-100, and 1× PIC. Samples were cleared of any nonspecifically binding proteins by incubation for 20 min at 4°C with protein A–Sepharose washed five times with 10 mM Tris-HCl, pH 7.5. The cleared samples were then subjected to immunoaffinity chromatography. Bound proteins were washed five times with 25 mM Tris-HCl, pH 7.5, 150 mM NaCl, and 0.5% (vol/vol) Triton X-100 and eluted with 100 mM glycine, pH 2.8. Proteins were precipitated by addition of trichloroacetic acid to 10%, washed in ice cold 80% (vol/vol) acetone, and then subjected to SDS-PAGE followed by immunoblotting or by fluorography (Titorenko et al., 1998).

For immunoaffinity chromatography under denaturing conditions, PMPs purified by immunoaffinity chromatography under native conditions and proteins recovered in the 200KgS (cytosolic), 200KgP, and 20KgP subcellular fractions were diluted with an equal volume of 4% SDS, and samples were warmed at 65°C for 10 min. Samples were then allowed to cool to room temperature, and four volumes of 62.5 mM Tris-HCl, pH 7.5, buffer containing 190 mM NaCl, 1.25% (vol/vol) Triton X-100, and 6 mM EDTA were added. Samples were cleared of any nonspecifically binding proteins by incubation for 20 min at 4°C with protein A–Sepharose washed five times with 10 mM Tris-HCl, pH 7.5. The cleared samples were then subjected to immunoaffinity chromatography. Bound proteins were washed five times with 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% (vol/vol) Triton X-100 and eluted with 2% SDS at 95°C for 5 min. Eluted proteins were subjected to a second immunoprecipitation (recapture) step (Bonifacino and Dell'Angelica, 1998), resolved by SDS-PAGE, and analyzed by immunoblotting or visualized by fluorography (Titorenko et al., 1998).

The pellet of PMPs recovered after centrifugation of osmotically lysed P6 peroxisomes was resuspended in 100 μl of buffer M (10 mM MES-KOH, pH 5.5, 1 mM KCl, 0.5 mM EDTA, 0.1% [vol/vol] ethanol, 1× PIC), transferred to the bottom of a 5-ml ultraclear centrifuge tube (Beckman Coulter), and supplemented with five volumes of 65% (wt/wt) sucrose in buffer M in order to adjust the sucrose concentration of the sample to 54% (wt/wt). The sample was then overlaid with 1.1 ml of 45% sucrose, 1.1 ml of 30% sucrose, 1.1 ml of 10% sucrose (all wt/wt in buffer M), and lastly with 1.1 ml of buffer M alone. After centrifugation at 200,000 g for 18 h at 4°C in a SW50.1 rotor (Beckman Coulter), 18 fractions of 275 μl each were collected.

Peroxisomal matrix proteins recovered in the supernatant fraction after centrifugation of osmotically lysed P6 peroxisomes were incubated for 2 h at 75°C. Under these conditions, all matrix proteins formed insoluble aggregates, as judged by light scattering at 320 nm and as confirmed by SDS-PAGE followed by Coomassie staining. Aggregates of peroxisomal matrix proteins were pelleted by centrifugation at 20,000 g for 30 min at 4°C and resuspended in 100 μl of buffer M. This material was subjected to flotation on a multistep sucrose gradient as described above.

Electron microscopy (Goodman et al., 1990), morphometric analysis of random electron microscopic sections of cells (Titorenko et al., 1998), and electron microscopic analysis of purified peroxisomal liposomes (Titorenko et al., 2000) were performed as previously described.

Preparation of peroxisomal liposomes (supplemental Materials and methods), SDS-PAGE and immunoblotting (Titorenko et al., 1998), pulse-chase analysis (Titorenko et al., 1998), and fractionation of peroxisomal proteins by centrifugation on a linear 5–35% glycerol gradient (Titorenko et al., 2002) were performed as previously described. Osmotic lysis of peroxisomes, protein extraction, and protease protection analysis of purified peroxisomes were performed according to established procedures (Szilard et al., 1995).

The supplemental material includes additional Materials and methods and figures (Figs. S1–S7). The supplemental Materials and methods describe preparation of peroxisomal liposomes. Fig. S1 demonstrates that all remaining Aox subunits form a membrane-attached complex that interacts with Pex16p inside mature peroxisomes of mutant cells lacking Aox1p, Aox2p, or Aox3p. Fig. S2 shows that all five Aox subunits and Pex16p are present in equimolar amounts in their membrane-associated complex inside mature peroxisomes of wild-type cells. Fig. S3 provides evidence that relocation of a 443-kD Aox complex from the matrix to the membrane at the last step of the assembly of mature peroxisomes results in the formation of a 900-kD complex containing Aox and Pex16p. Figs. S4–S6 provide data on electron microscopic analysis of the dynamics of change in the size and number of peroxisomes in wild-type, aox4KO, and aox5KO cells transferred from glucose- to oleic acid–containing medium. Fig. S7 shows the amounts of individual components of mature peroxisomes, P6, used for the reconstitution of peroxisomal liposomes PLA to PLD and provides transmission electron micrographs of these liposomes purified by flotation on a multistep sucrose gradient.

We thank Jeannie Mui (Facility for Electron Microscopy Research, McGill University) for help with electron microscopy.

This work was supported by grants from the Canadian Institutes of Health Research to V.I. Titorenko (MOP-57662) and R.A. Rachubinski (MOP-9208) and by a grant from the Canada Foundation for Innovation to V.I. Titorenko. J.M. Nicaud and M.T. Le Dall were supported by the Institut National de la Recherche Agronomique and by the Centre National de la Recherche Scientifique. R.A. Rachubinski is a Canada Research Chair in Cell Biology and an International Research Scholar of the Howard Hughes Medical Institute.