Dynamic changes of membrane structure are intrinsic to organelle morphogenesis and homeostasis. Ectopic expression of proteins of the PEX11 family from yeast, plant or human lead to the formation of juxtaposed elongated peroxisomes (JEPs),which is evocative of an evolutionary conserved function of these proteins in membrane tubulation. Microscopic examinations reveal that JEPs are composed of independent elongated peroxisomes with heterogeneous distribution of matrix proteins. We established the homo- and heterodimerization properties of the human PEX11 proteins and their interaction with the fission factor hFis1, which is known to recruit the GTPase DRP1 to the peroxisomal membrane. We show that excess of hFis1 but not of DRP1 is sufficient to fragment JEPs into normal round-shaped organelles, and illustrate the requirement of microtubules for JEP formation. Our results demonstrate that PEX11-induced JEPs represent intermediates in the process of peroxisome membrane proliferation and that hFis1 is the limiting factor for progression. Hence, we propose a model for a conserved role of PEX11 proteins in peroxisome maintenance through peroxisome polarization, membrane elongation and segregation.
Peroxisomes are highly versatile organelles whose size, shape, number and protein content adapt to the cell type or metabolic requirements (Subramani, 1993; Wanders and Waterham, 2006). This dynamic behavior is essential for cell survival. Peroxisomes are able to proliferate by growth of pre-existing organelles followed by division, but they might also form de novo from the endoplasmic reticulum (ER) (Geuze et al., 2003; Motley and Hettema, 2007; Toro et al., 2009). In human cells, both pathways seem to contribute to the cellular peroxisome pool (Huybrechts et al., 2009; Kim et al., 2006). Failures in peroxisome formation lead to biogenesis disorders (PBDs) such as the Zellweger syndrome, which belongs to a group of lethal metabolic diseases (Steinberg et al., 2006). Similarly, peroxisomes in plants fulfill vital functions in photorespiration and in the metabolism of essential growth hormones such as auxin and jasmonic acid. The absence of functional peroxisomes has been associated with mutations in PEX genes whose products, the peroxins, are required for protein import and organelle maintenance (Distel et al., 1996; Wanders and Waterham, 2005). Peroxins are involved in peroxisomal biogenesis and proliferation, membrane and matrix protein import, and recycling of import receptors to the cytosol. Orthologs of these proteins are found in all eukaryotic organisms.
Proteins of the PEX11 family were shown to directly participate in peroxisome proliferation in yeasts, plants and mammals (Abe and Fujiki, 1998; Erdmann and Blobel, 1995; Lingard and Trelease, 2006; Marshall et al., 1995; Orth et al., 2007; Schrader et al., 1998). The yeast Saccharomyces cerevisiae harbors three members of the PEX11 protein family, the founding member PEX11, plus PEX25 and PEX27 (Erdmann and Blobel, 1995; Rottensteiner et al., 2003; Smith et al., 2002; Tam et al., 2003). Five orthologs of PEX11 were found in Arabidopsis thaliana (Lingard and Trelease, 2006; Orth et al., 2007) and three in mammals (Abe et al., 1998; Tanaka et al., 2003). The five plant PEX11 proteins (PEX11a–PEX11e) fall into two clades. Members of the first clade (PEX11c–PEX11e) are nearly identical in their amino acid composition. Members of the second clade (PEX11a, PEX11b) are more divergent, with approximately 50% similarity. Interestingly, proteins of the first clade carry a C-terminal amino acid sequence–KXKXX, known as the ER retrieval motif, which is thought to facilitate binding of coatomer (Andersson et al., 1999; Cosson et al., 1998).
In mammalian cells, three PEX11-related genes have been identified, PEX11α, PEX11β and PEX11γ. The expression of the PEX11 genes has been extensively studied and animal models have been generated lacking either PEX11α or PEX11β (Li et al., 2002; Li and Gould, 2002). Mice lacking PEX11β display many of the pathologic characteristics of a Zellweger syndrome mouse, which includes neonatal lethality. Mice lacking PEX11α developed normally, showed no obvious defect in constitutive peroxisome division, and displayed a normal response to peroxisome proliferating agents (Li et al., 2002). However, overproduction of PEX11α is sufficient to induce peroxisome proliferation in mouse and human cultured cells (Li and Gould, 2002).
Conservation in the targeting of PEX11 proteins between kingdoms was suggested because heterologously expressed PEX11 from Trypanosoma brucei localizes to peroxisomes in mammalian cells (Lorenz et al., 1998). So far, all identified PEX11 proteins share common features: they are small, very basic, and harbor putative transmembrane regions, as calculated using the algorithm HMMTOP 2.0 (Tusnady and Simon, 2001). PEX11 is the most abundant peroxin at the peroxisomal membrane (Erdmann and Blobel, 1996). PEX11 proteins from S. cerevisae, T. brucei and human cells (ScPEX11, TbPEX11 and HsPEX11γ, respectively) were reported to remain insensitive to protease digestion, which suggests membrane embedding (Lorenz et al., 1998; Marshall et al., 1996; Tanaka et al., 2003). Although a consensus membrane peroxisomal targeting signal (mPTS) exists in PEX11 proteins and facilitates interaction with the mPTS receptor, PEX19 (Fransen et al., 2005; Lorenz et al., 1998; Rottensteiner et al., 2004; Sacksteder et al., 2000), most aspects of peroxisome membrane protein insertion and membrane proliferation remain unknown (Brocard and Hartig, 2006; Girzalsky et al., 2009).
Peroxisome proliferation seems to be a multistep process including elongation, constriction and fission (Koch et al., 2004). PEX11 proteins are thought to participate in the first two steps, whereas DRP1 (dynamin-related protein) and Fis1 (mitochondrial fission protein 1) seem to facilitate fission (Koch et al., 2005; Motley et al., 2008; Zhang and Hu, 2009). Molecular interactions have been established between HsPEX11β and hFis1, a tail-anchored protein recruiting the DRP1 GTPase to the peroxisome membrane (Kobayashi et al., 2007). Fis1 and DRP1, originally found as components of the mitochondrial fission machinery, also localized to the peroxisomal membrane (Koch et al., 2005; Kuravi et al., 2006; Lingard et al., 2008; Motley et al., 2008; Wells et al., 2007). Only a subset of PEX11 proteins was shown to interact with Fis1 in plant and mammals (Kobayashi et al., 2007; Lingard et al., 2008), raising the question of the specific role of each PEX11 protein.
To address this question, we performed a functional study on proteins of the PEX11 family. We present cross-species studies with all PEX11 proteins from yeast, plant and human that illustrate the evolutionary conservation of the molecular mechanism that governs peroxisome proliferation. Revealing new aspects on peroxisome membrane proliferation and inheritance, we demonstrate that PEX11 proteins from yeast, plant and human stimulate the formation of juxtaposed elongated peroxisomes (JEPs) in human cells. We established specific homo- and heterodimerization properties of human proteins and interaction with fission factors. Live-imaging and biochemical analyses exposed specific roles of distinct PEX11 proteins in membrane tubulation and suggest the involvement of microtubules in peroxisome maintenance.
Heterologous PEX11 proteins localize to peroxisomes in human cells and plants
In general, eukaryotic organisms contain a number of distinct PEX11 proteins. To assess PEX11 expression levels in human embryonic kidney cells (HEK293T), we performed quantitative RT-PCRs with mRNAs. The data revealed that all three PEX11 genes are expressed, with the ratio 2:5:1 for PEX11α, PEX11β and PEX11γ, respectively. To investigate the effects on the shape, size and number of peroxisomes, we expressed EGFP appended with PEX11 sequences originating from the yeast S. cerevisae (Sc), the plant A. thaliana (At) and human (Hs) in HEK293T cells. Each recombinant EGFP–PEX11 fusion protein produced showed the correct size on a western blot (supplementary material Fig. S1).
We first examined cells expressing either the marker protein EGFP or EGFP–PEX11 fusion proteins (Fig. 1A) using confocal laser scanning microscopy (CLSM). The expression of the PEX11 fusion proteins resulted in the appearance of diverse fluorescent structures, either punctate or elongated, that could also be decorated with antibodies specific to genuine peroxisomal membrane (HsPEX14; Fig. 1A) proteins or matrix (catalase; supplementary material Fig. S2) proteins. This demonstrates that PEX11 proteins of yeast, plant and human origin could all traffic to peroxisomes in human cells (Fig. 1A). Next, to visualize peroxisomes in living cells we concomitantly expressed the peroxisomal matrix marker mCherry–SKL and EGFP–PEX11 fusion proteins and obtained similar results (supplementary material Fig. S3). To scrutinize whether the PEX11 family members present a similar localization pattern in plant tissue, we transiently coexpressed YFP-tagged PEX11 proteins together with the red fluorescent mCherry–SKL in leaf epidermal cells. Except for HsPEX11β, all PEX11 fluorescent fusion proteins were expressed and detected in association with peroxisomes (Fig. 2). Together, these results suggest that targeting of PEX11 proteins to peroxisomes is evolutionarily conserved.
Ectopic expression of PEX11 proteins leads to peroxisome proliferation in human and plant cells
Because our experiments show that all PEX11 fusion proteins localized to peroxisomes, we evaluated their individual effects on the peroxisome number and size and performed statistical analyses. Here, we coexpressed EGFP–PEX11 fusion proteins together with mCherry–SKL in human cells and quantified peroxisomes 24 hours after transfection. This time point was chosen to observe early effects on peroxisomes (Fig. 1B).
The expression of all EGFP–PEX11 fusion proteins had an effect on peroxisome (Px) number and size, although with various intensities. Although expression of ScPEX11 (281±34 Px), HsPEX11β (405±41 Px) or AtPEX11d (264±32 Px) led to an increase in the number of peroxisomes counted per cell as compared to control (187±17 Px), the cells expressing other proteins such as ScPEX25 (60±6 Px), AtPEX11e (50±9 Px) and HsPEX11γ (63±3 Px) presented fewer but larger peroxisomes per cell (Fig. 1B). The expression of AtPEX11a (186±22 Px) and HsPEX11α (188±17 Px) was associated with the incidence of smaller peroxisomes, whereas expression of AtPEX11b (140±14 Px) and AtPEX11c (137±17 Px) led to slightly larger ones. By contrast, peroxisomes in cells expressing ScPEX27 (180±26 Px) were indistinguishable from control cells. On a similar line, ectopic expression of individual PEX11 fusion proteins in plant tissue altered the appearance of peroxisomes to different degrees (Fig. 2; supplementary material Table S1). Our data confirm that overexpression of all PEX11 proteins affects peroxisome appearance in plant and human cells.
Ectopic expression of PEX11 proteins leads to the formation of JEPs in human cells
Expression of PEX11 proteins induced the formation of unusually large peroxisomal structures. Within these, sub-structures were observed that could represent juxtaposed elongated peroxisomes (JEPs). The appearance of such structures could be due to intracellular membrane proliferation and aggregation unrelated to peroxisomes, as a consequence of the overproduction of membrane proteins (Wright et al., 1988). Therefore, we analyzed whether such membrane alterations also occurred in cells devoid of peroxisomes. We expressed each one of the EGFP–PEX11 proteins in skin fibroblasts obtained from Zellweger patients lacking a functional PEX19 (Matsuzono et al., 1999). In these cells devoid of detectable peroxisomes, no fluorescent membranous structures were detected. Some of the ectopically expressed proteins appeared essentially in the cytosol, whereas others were not visible by confocal microscopy (our unpublished data), demonstrating that the clusters observed in wild-type cells upon ectopic expression of PEX11 proteins were indeed associated with the presence of peroxisomes and represent JEPs. These structures could also be the consequence of the N-terminal tagging of PEX11 with EGFP. To rule out this possibility, we expressed untagged PEX11 proteins in HEK293T and observed the same effects on peroxisome morphology (supplementary material Fig. S4). Our results clearly show that JEP formation is solely due to overexpression of PEX11 proteins.
At the time point chosen for the statistical analysis (24 hours), JEPs were visualized only in cells expressing certain EGFP–PEX11 fusion proteins. Therefore, we performed time-course experiments and analyzed JEP formation in cells expressing the different PEX11 fusion proteins. After 4 days, expression of all PEX11 fusion proteins led to the occurrence of JEPs, but the kinetics and progression greatly varied depending on which PEX11 was overexpressed (supplementary material Table S1). The expression of ScPEX25, HsPEX11β, HsPEX11γ, AtPEX11a and AtPEX11e led to the formation of JEPs 18–24 hours after transfection. HsPEX11α led to the occurrence of small JEPs 24 hours after transfection, whereas ScPEX27 or AtPEX11d induced large JEPs only after 2–3 days (supplementary material Table S1; Fig. 3A). Note that in JEPs fluorescence of peroxisomal matrix proteins did not entirely overlap with the appearance of PEX11 fusion proteins. Instead, the PEX11 fusion proteins localized to structures surrounding the matrix, as indicated with EGFP–ScPEX27 (Fig. 3B,C).
It has been reported that the excess of peroxisomes is selectively degraded by pexophagy (Klionsky, 1997) involving the microtubule-associated protein I light chain 3 (LC3), an essential factor for autophagy in mammalian cells (Hara-Kuge and Fujiki, 2008). To test whether JEPs represent intermediates in the process of pexophagy, we coexpressed the mRFP–LC3 and PEX11 fusion proteins. As expected, mRFP–LC3 associated with few peroxisomes; however, JEPs did not colocalize with mRFP–LC3 (supplementary material Fig. S5), indicating that they do not constitute pexophagy intermediates.
In time-course experiments, we observed that overexpression of PEX16, a peroxisomal membrane peroxin required for peroxisome membrane biogenesis, did not lead to any change in peroxisome morphology. Thus, the detected alterations are correlated with PEX11 protein function (Fig. 1A; Fig. 3D). We analyzed the changes in peroxisomal membrane appearance using human PEX11 proteins and observed that, prior to JEP formation (30 hours), the peroxisomes formed protrusions as visualized by GFP fluorescence (Fig. 3D). Note that mCherry–SKL did not fully colocalize with EGFP–PEX11-labeled protrusions. Similar observations were made using immunofluorescence after staining the cells with anti-catalase antibodies (supplementary material Fig. S2). Hence, peroxisomal matrix proteins seem to be excluded from the PEX11-induced membrane protrusions.
Peroxisome clustering is a membrane dynamic event leading to JEP formation
Live-imaging of cells expressing the PEX11 proteins revealed a dynamic trafficking and clustering of peroxisomes towards the center of the cells (supplementary material Movie 1). Peroxisome remodeling and elongation occurred up to the formation of JEPs, which were inherited during cell division. To determine whether the JEPs are autonomous entities or represent continuous membrane structures we performed fluorescence recovery after photobleaching (FRAP) assays (Fig. 4A,B). Whereas in control cells the expressed ER membrane protein EGFP–Sec61β showed rapid and full fluorescence recovery (t1/2~7 seconds), the EGFP–HsPEX11β signal did not show significant fluorescence recovery. This clearly shows that the PEX11-containing structures do not share a common membrane. Thus, it might well be that the clusters that emerge upon overexpression of PEX11 proteins represent peroxisomes that remain juxtaposed due to interorganellar protein interactions. Analysis of PEX11γ-induced structures via electron microscopy illustrates the presence of tubular smooth membranes that are absent in wild-type HEK293T cells (Fig. 4C). The abundant presence of PEX11 proteins enhances peroxisome elongation and concomitantly delays peroxisome fission, generating the opportunity to follow several steps of peroxisome proliferation. JEPs seem to represent intermediates formed during peroxisome proliferation. Interaction of PEX11 with components of the cytoskeleton might lead to such clustering and influence peroxisome inheritance during cell division, as proposed for yeast (Krikken et al., 2009).
JEPs are the result of exhausted fission machinery
To test whether JEPs are the consequence of incomplete peroxisome fission, we analyzed the in vivo effect of fission factors on JEP formation. Coexpression of myc-hFis1 and EGFP–PEX11 fusion proteins gave rise to well-separated peroxisomes (Fig. 5A). Thus, the fission factor hFis1 counterbalanced PEX11-induced JEP formation. Supporting the notion that membrane elongation is part of the proliferation process coexpression of ECFP–DRP1 and EGFP–PEX11 proteins led to the formation of even more elongated JEPs than expression of EGFP–PEX11 fusions alone. Moreover, as exemplified in Fig. 5B, three-dimensional reconstructions illustrate that JEPs are composed of individual elongated peroxisomal structures. Thus, hFis1 seems to be the limiting factor for continuing the proliferation process induced by ectopic expression of human PEX11 proteins. The effects of ectopic expression of yeast and plant PEX11 fusion proteins were counterbalanced by hFis1, as well (Fig. 5A and supplementary material Fig. S2), which suggests that the molecular mechanism underlying peroxisome proliferation has been conserved through evolution. Although our experiments indicate that overexpression of hFis1 restored peroxisome fission in PEX11-overexpressing cells we could not rule out the possibility that hFis1 simply overrode PEX11-initiated peroxisome proliferation through another mechanism.
To distinguish between the two possibilities, we expressed the peroxisomal marker protein EGFP–Scp2 or PEX11 fusion proteins. When JEPs appeared in the latter (48 hours), cells were transfected a second time with plasmids coding for hFis1, DRP1, or both, and grown for another 48 hours. Whereas DRP1 expression enhanced peroxisome elongation and JEP-formation (Fig. 5C), hFis1 or hFis1 and DRP1 expression led to well-separated and normal-sized peroxisomes regardless of the abundant presence of a yeast, plant or human PEX11 protein (Fig. 5). Obviously, proteins of the PEX11 family induce peroxisome proliferation, and hFis1 function is essential to progress into fission and to finalize the proliferation process. The levels of hFis1 do not vary upon PEX11 overexpression (supplementary material Fig. S1). To assess whether the hFis1 levels are influential in the context of our experiments, we evaluated the effects of hFis1 knockdown in human cells using siRNA. Cells expressing unspecific siRNA targeted to firefly luciferase (control) presented well-separated round-shaped peroxisomes. But, in cells expressing the hFis1–siRNAs, peroxisomes were elongated and formed small JEPs. This effect was tremendously enhanced by the overexpression of HsPEX11β (Fig. 5D). In that case, JEPs appeared much faster, which correlates with hFis1 levels becoming limited. As a consequence, DRP1 cannot be properly recruited on peroxisomes and the fission process does not occur. Considering the difference between a knockdown and an overexpression, this result is consistent with the hypothesis that the PEX11 to hFis1 ratio determines the rate of peroxisome proliferation.
Interplay between components of the proliferation machinery in human cells
The yeast PEX11 is thought to act as a monomer (Marshall et al., 1996), whereas PEX11β seems to require homodimerization for its proliferating activity (Kobayashi et al., 2007). It has been reported that HsPEX11α and HsPEX11β could form homophilic complexes but that these two proteins could not interact with each other (Li and Gould, 2003), which suggests that they might be part of different pathways. Yet, only HsPex11β has been shown to be part of a ternary complex containing hFis1 and DRP1 (Kobayashi et al., 2007). To reveal the interplay between PEX11 proteins and factors of the fission machinery, we performed a comprehensive molecular interaction study. Co-immunoprecipitations were performed on HEK293T cell lysates 48 hours after coexpression of: (1) HsPEX11α–FLAG and EGFP–HsPEX11α; (2) HsPEX11α–FLAG and EGFP–HsPEX11β; (3) HsPEX11γ–FLAG and EGFP–HsPEX11α; (4) HsPEX11β–FLAG and EGFP–HsPEX11β; (5) HsPEX11γ–FLAG and EGFP–HsPEX11β; and (6) HsPEX11γ–FLAG and EGFP–HsPEX11γ. In the presence of 0.2% digitonin, all pairwise interactions were detected, except between HsPEX11α and HsPEX11β (Fig. 6A) and, in addition, all three human PEX11 proteins co-precipitated with hFis1 (Fig. 6B). Note that none of these interactions could be detected in the presence of 1% Triton X-100. It seems that the PEX11 proteins act as heteromeric pairs consisting of HsPEX11α–HsPEX11γ and HsPEX11β–HsPEX11γ, representing two separate proliferation pathways, which both require interaction with fission factors to fulfill their function. That immunoprecipitations can occur in the presence of digitonin but not of Triton X-100 suggests that the hydrophobic regions of each PEX11 protein and their membrane integration are required for interaction.
Microtubules and hydrophobic regions of PEX11 proteins are required for JEP formation
The oriented movement of JEPs during cell division suggests that microtubules and PEX11 proteins are involved in peroxisome inheritance (see supplementary material Movie 1). To test this, we treated cells expressing EGFP–HsPEX11γ with Nocodazole, a chemical that blocks the self-assembly of tubulin leading to microtubule depolymerization. As shown in Fig. 7, a functional microtubule cytoskeleton is required for JEP formation.
To assess whether hydrophobic regions of PEX11 proteins are essential for the formation of JEPs, we expressed EGFP fusions of HsPEX11α, HsPEX11β or ScPEX11 lacking their putative transmembrane regions in HEK293T cells (Fig. 8A). In cells overexpressing any one of the three proteins, the morphology and number of peroxisomes were indistinguishable from those in wild-type cells and no JEPs were formed. Interestingly, although EGFP expressing HsPEX11α lacking its C-terminal hydrophobic region appeared predominantly soluble in the cytosol, a significant portion of the truncated EGFP–ScPEX11 and EGFP–HsPEX11β did localize to peroxisomes. We performed co-immunoprecipitations using these truncated PEX11 proteins. Here, HsPEX11αΔ20 did not interact with the full-length HsPEX11α and HsPEX11γ. Although, HsPEX11βΔ26 was able to interact with the full-length HsPEX11β, interaction with HsPEX11γ or with hFis1 could not be detected (Fig. 8B). Together, these data indicate that PEX11 proteins rendered less hydrophobic lose their function in peroxisome proliferation because they can neither form specific heterodimers nor bind hFis1.
Proteins of the PEX11 family are known to be essential regulators of peroxisome proliferation in all organisms studied. By shifting the delicate balance between proliferation factors we were able to observe intermediate stages of peroxisome proliferation. We show that ectopic expression of PEX11 proteins of yeast, plant and human induces a dynamic change in peroxisomal appearance in plant and human cells. Transiently expressed PEX11 proteins localize to peroxisomes and lead to the formation of membrane protrusions and to membrane elongation, which finally develop into JEP structures (Figs 1, 2). This model is supported by three-dimensional reconstructions of confocal images showing JEPs (Fig. 5B), which correspond in size and shape to the tubular membrane structures observed in electron micrographs (Fig. 4C). The size of these structures depends on which PEX11 protein is ectopically expressed (supplementary material Table S1). Similar changes in peroxisome appearance are induced upon a decrease in hFis1 levels, implying that peroxisome fission is hampered in cells overexpressing either of the PEX11 protein fusions. This scenario is strengthened by the observation that an increase in hFis1 fission factor levels leads to the dissolution of JEPs (Fig. 5). In addition, PEX11-driven peroxisome membrane elongation coincides with the segregation of PEX11 from matrix proteins, which suggests that PEX11 proteins are key factors that induce remodeling of the entire peroxisomal compartment.
Our data demonstrate that both the targeting and the function of PEX11 proteins have been conserved throughout evolution. Mammalian cells possess three PEX11 proteins, PEX11α, PEX11β and PEX11γ, and little is known regarding their precise function and interactions. We show that PEX11γ interacts with both PEX11α and PEX11β, but that these two proteins do not associate with each other. These results place PEX11γ at the crossroad of PEX11-induced peroxisome proliferation pathways. Using co-immunoprecipitation assays we confirmed the previously reported homodimerization of PEX11α and PEX11β (Li and Gould, 2003) and show that PEX11γ can similarly participate in homotypic interactions. A possible explanation for this finding is that the function of PEX11 proteins is regulated through the formation of homodimers. Indeed, homodimerization could constitute a molecular switch that allows PEX11 proteins to change from an active (monomer) to an inactive state (dimer), as already proposed for the yeast PEX11 protein (Marshall et al., 1996). Conversely, stimulus-driven PEX11 protein heterodimerization could allow the formation of PEX11-rich patches on the membrane, thereby promoting membrane protrusion and elongation at a distinct location on the peroxisomal surface. PEX11β might be needed for constitutive peroxisome proliferation, whereas PEX11α might be required for peroxisome proliferation in response to external stimuli. Our findings that overexpression of PEX11γ induces the early formation of JEPs (supplementary material Table S1) suggest that this protein either acts upstream of PEX11α and PEX11β or is the limiting factor, in agreement with the results of our quantitative RT-PCR. PEX11γ seems to recruit the other PEX11 proteins and position PEX11-rich patches on the peroxisomal membrane to facilitate further molecular associations. Our data suggest that PEX11γ is always required for elongation of the peroxisome membrane, and that PEX11α or PEX11β might support peroxisome proliferation and division only under inducing or non-inducing growth conditions, respectively.
PEX11γ was found to expose its N- and C-termini to the cytosol (Tanaka et al., 2003), and the same type of membrane topology was suggested for PEX11α and PEX11β (Abe and Fujiki, 1998; Passreiter et al., 1998; Schrader et al., 1998), implying that PEX11 proteins might possess a functional domain in their cytosol-oriented N- or C-terminal part. Two hybrid assays with PEX11β suggested that, whereas the presence of its N-terminus is important for dimerization, its C-terminus might interact with hFis1 and counteract homodimerization (Kobayashi et al., 2007). Here, we expanded this theme on other proteins of the PEX11 family and found that overexpression of PEX11 proteins, either untagged or tagged with EGFP at their N-termini, greatly affected peroxisomal structure (Figs 1, 2; supplementary material Fig. S4) but this effect was abolished when their hydrophobic domain was deleted (Fig. 8A). Interestingly, although C-terminal tagging with small tags such as haemagglutinin (HA) or FLAG did not hinder PEX11 function and JEPs were formed in the cells, large C-terminal tags, e.g., EGFP abolished the PEX11-driven effects. Thus, it is likely that the C-termini of PEX11 proteins play an essential role in promoting peroxisome membrane proliferation. In addition, the hydrophobic region close to the C-termini might contribute to trafficking events associated with their correct insertion into the peroxisomal membrane.
In lower eukaryotes, physiological levels of PEX11 are sufficient to cause fragmentation of peroxisomes, and peroxisomes are enlarged in cells lacking PEX11 (Erdmann and Blobel, 1995; Voncken et al., 2003). In human cells, high levels of PEX11 proteins lead to tubulation and elongation of the peroxisomal membrane (Figs 3, 4). Thus, PEX11 proteins might regulate the overall membrane curvature or associate with specific lipids to determine the correct composition of the peroxisomal membrane, which are two functions that might also be needed for de novo formation of peroxisomes.
Peroxisome movement has been reported to depend on microtubules in mammalian cells and on actin filaments in yeast and plants (Fagarasanu et al., 2006; Mathur et al., 2002; Rapp et al., 1996; Wiemer et al., 1997). The formation of PEX11-induced JEPs could, directly or indirectly, depend on the microtubular transport machinery. A study on the role of microtubules in peroxisome proliferation in fibroblasts from patients with Zellweger syndrome (PEX1-null) showed that the overexpression of PEX11β restored the alignment of peroxisomal structures along microtubules (Nguyen et al., 2006). Accordingly, we show that during cell division, JEPs have an ordered movement, are inherited (see supplementary material Movie 1) and their formation depends on the presence of intact microtubules (Fig. 7). Thus, an essential aspect of peroxisome proliferation could be the interaction of the organelle with the cytoskeleton, a process in which PEX11 proteins might fulfill a primordial function through association with microtubule binding proteins.
The known mitochondrial fission factors, dynamin-related-proteins and Fis1, also play a role in peroxisome fission in yeast, plant and mammalian cells (Kobayashi et al., 2007; Koch et al., 2005; Zhang and Hu, 2009). HsPEX11β has previously been shown to bind hFis1 (Kobayashi et al., 2007), raising the question of whether PEX11β is the sole factor involved in the recruitment of the fission factors to the peroxisomal membrane. Addressing this, we found that hFis1 associates with all three human PEX11 proteins (Fig. 6B). Heteromeric PEX11 protein complexes, such as PEX11α–PEX11γ and PEX11β–PEX11γ dimers, might recruit hFis1 and thereby initiate DRP1 self-assembly. This interaction cascade might induce constriction and scission of the peroxisomal membrane, as proposed for mitochondria (Fukushima et al., 2001). In yeast, it has been reported that the number of peroxisomes doubles shortly before cell division (Hoepfner et al., 2001). Similar to mitochondria (Taguchi et al., 2007), peroxisomes might become fragmented in early mitotic phase. Our results support the notion that proteins of the PEX11 family are essential for initiation of peroxisomal fission by anchoring DRP1 through hFis1.
On the basis of our observation that PEX11 is unequally distributed in the peroxisomal membrane (Figs 1, 3; supplementary material Figs S2, S3), we suggest that this polarization represents a key step in the initiation of elongation of the membrane. We propose a model (Fig. 8C) for a conserved role of PEX11 proteins in peroxisome polarization, membrane protrusion, and elongation. In this model, PEX11γ initiates the proliferation by determining the site of protrusion through formation of PEX11-rich patches at the membrane. Consistent with this view, PEX11γ displays low constitutive expression level, interacts with both PEX11α and PEX11β and, upon ectopic expression, triggers the earliest appearance of JEPs. In the next stage, PEX11γ (with the help of PEX11α or PEX11β) initiates the formation of protrusions leading to elongation of the peroxisomal membrane. By these means, PEX11 proteins might coordinate peroxisome proliferation to the metabolic requirements of the cell. The final stages of membrane constriction and fission require additional factors and have to be coordinated with the delivery of matrix proteins and membranous components to the nascent organelle. Hence, by altering the composition of the PEX11 complex at the peroxisomal membrane, an effect on the import and distribution of peroxisomal matrix proteins can also be observed (Fig. 3). Our studies on PEX11 proteins provide new insights into the mechanism of peroxisome proliferation. We have revealed an intermediate morphological stage in the formation of JEPs, which became detectable as a consequence of a change in the equilibrium between PEX11 proteins and hFis1.
Materials and Methods
All PEX11 proteins were N-terminally tagged with EGFP (human cells) or YFP (plant cells). The PCR fragments representing the cDNA of ScPEX11, ScPEX25 and ScPEX27 (YOL147C, YPL112C, YOR193W) and AtPEX11a–AtPEX11e (TAIR Acc. AT1G47750, AT3G47430, AT1G01820, AT2G45740 and AT3G61070) or HsPEX11α, HsPEX11β and HsPEX11γ (Acc. NM_003847, NM_003846 and NM_080662) were cloned into pENTR4 (Invitrogen) using NcoI and XhoI restriction sites. The resulting pENTR4–PEX11 plasmids were sequenced and served as entry constructs for Gateway (Invitrogen) recombination-mediated cDNA transfer into the pDEST53 vector (Invitrogen), allowing for expression of EGFP–PEX11 fusions under the control of the CMV promoter in human cell culture. Plasmids coding for EGFP–HsPEX11αΔ20 (AA220 to AA239) and EGFP–HsPEX11βΔ26 (AA230 to AA255) as well as EGFP–ScPEX11Δ18 (AA215 to AA232) were engineered by PCR and cloned into pENTR4, followed by recombination as described above. HsPEX11γ–3×FLAG in a pReceiverM14 was purchased from GeneCopoeia (Rockville, MD). HsPEX11α, HsPEX11β and HsPEX11βΔ26 were exchanged against HsPEX11γ in the pReceiverM14 (KpnI/NheI) to create HsPEX11α–3×FLAG, HsPEX11β–3×FLAG and HsPEX11βΔ26–3×FLAG, respectively. For mammalian cells, pmCherry–SKL was cloned via PCR by appending the tripeptide SKL to mCherry and replacing it against EYFP–ER in the plasmid pEYFP–ER (NheI/BglII, Clontech).
For plants, the coding sequences from the pENTR4–PEX11 plasmids were recombined into the binary plant expression vector pEarlyGate104 (Earley et al., 2006). The resulting vector allowed 35S promoter-driven in planta expression of N-terminally tagged YFP fusions. For estradiol-inducible expression of untagged AtPEX11d, the according entry vector was recombined with pMDC7 (Curtis and Grossniklaus, 2003). The red fluorescent peroxisomal marker construct mCherry–SKL cDNA was produced by PCR and transferred via a BP Gateway reaction into the pDONOR plasmid (Invitrogen), which served as template for an LR gateway recombination with the pMDC7 vector. For transient plant expression experiments, the according binary vectors were transferred by electroporation into the A. tumefaciens strain AGL1 (Lazo et al., 1991). To create the EGFP–myc-hFis1 expression plasmid, EGFP (NdeI-BglII) from pEGFP-C1 (Clontech) was inserted into the myc-hFis1 (Yoon et al., 2003) expression vector. Plasmids expressing HsPEX16–CFP (Brocard et al., 2005) and GFP–Scp2 (Stanley et al., 2006) have been described before. For primers used in this study see supplementary material Table S2.
Cell culture, transfection, RNA INTERFERENCE and Immunofluorescence
Human embryonic kidney cells (HEK293T) were cultured in DMEM (+10% FCS, +1% penicillin/streptomycin; PAA Laboratories, Pasching, Austria) at 37°C (5% CO2). Cells were transfected using FuGene6 (Roche) or nucleofected (Lonza). For microscopic analysis, cells were fixed with 3.7% formaldehyde in PBS (15 minutes) and embedded in Mowiol supplemented with 25 mg/ml DABCO (Carl Roth, Karlsruhe, Germany). Prior to immunofluorescence staining, cells were fixed, permeabilized (0.1% Triton X-100 in PBS, 10 minutes) and blocked (2% BSA in PBS, 30 minutes). Subsequently, cells were incubated with the primary antibody, e.g. rabbit-anti-HsPEX14 (1:400) or sheep-anti-catalase (1:250) for 1 hour, washed three times with PBS (10 minutes) and incubated with the appropriate secondary antibody (1:200) for 30 minutes followed by three washing steps (10 minutes) and embedded. Hoechst 33342 (1 μg/ml) was used for counterstaining the nuclei. For knockdown of hFis1, siRNAs were transfected as described previously (Koch et al., 2005). Control experiments were performed using endoribonuclease-prepared siRNAs directed to firefly luciferase (FLuc) mRNAs (Sigma).
Transient expression in N. benthamiana leaves
Six-week old Nicotiana benthamiana plants, grown in the greenhouse at 22–25°C and 16 hours light, were used for leaf infiltration experiments with agrobacterial solutions harboring the relevant binary plasmids prepared as described (Winter et al., 2007). In short, for single expression or double expression studies, agrobacterial suspensions with an OD600 of 0.15 or 0.3 were infiltrated into leaves, respectively. For estradiol-induced expression, a final concentration of 10 μM estradiol (Sigma) was applied by addition from a 50 mM stock solution to the infiltration solution.
Cells expressing the appropriate proteins were processed 48 hours after transfection. Cells were washed in PBS, incubated in 1 ml lysis buffer (50 mM Tris-HCl pH 7.4, 150 mM NaCl, 1 mM EDTA) containing either 1% Triton X-100 or 0.2% digitonin. The lysates were transferred onto columns containing anti-FLAG M2 affinity gel (Sigma Aldrich), incubated (2 hours, 4°C) and washed extensively (0.5 M Tris-HCl pH 7.4, 1.5 M NaCl). Immune-precipitates were eluted using 3×FLAG peptides (150 ng/μl; Sigma Aldrich) for 30 minutes at 4°C. Western blot analyses were performed on aliquots with anti-FLAG, anti-GFP or anti-myc antibodies. Signals were visualized using HRP-conjugated secondary antibodies and Super Signal West Pico chemiluminescence kit (Thermo Scientific).
Rabbit-anti-HsPEX14 antibodies were a kind gift from Ralf Erdmann (Ruhr-Universität, Bochum, Germany). Sheep-anti-human catalase antibodies were purchased from The Binding Site (Heidelberg, Germany). Alexa Fluor 594 donkey-anti-sheep and Alexa Fluor 594 donkey-anti-rabbit antibodies were purchased from Molecular Probes (Invitrogen). Rabbit-anti-GFP antibodies were a kind gift from Michael Rout (The Rockefeller University, New York, NY). Rabbit-anti-calnexin antibodies were kindly provided by Erwin Ivessa (MFPL, Vienna, Austria). Rabbit-anti-hFis1 antibodies were a kind gift from Jean-Claude Martinou (University of Geneva, Geneva, Switzerland). Anti-FLAG M2 monoclonal antibodies (HRP-conjugated) and the HRP-conjugated donkey-anti-sheep antibodies were purchased from Sigma-Aldrich. Mouse-anti-α-tubulin and Texas-Red-conjugated goat-anti-mouse antibodies were a kind gift from Gerhard Wiche (MFPL). HRP-conjugated sheep-anti-mouse and donkey-anti-rabbit antibodies were purchased from GE Healthcare.
Microscopy and statistical analysis
For human cells, confocal images were acquired on a LSM510META, Zeiss (Neofluar 100×1.45, pixel size 45×45 nm, z-stacks 200 nm, 1.6 μs pixel dwell time, 12-bit) using a 405 nm laser (BP420-480) for Hoechst staining, 488 nm laser (BP500-550) for GFP, 561 nm laser (LP585) for mCherry, and 633 nm laser (Meta 585-625) for mitotracker-IR. Cells were randomly chosen, and detector gain and amplifier offset were adjusted to avoid clipping.
Live-cell imaging was performed with an Olympus CellR unit (widefield) using appropriate filter sets for GFP (BP457-487 excitation; BP503-538 emission) and mCherry (BP510-550 excitation; LP590 emission).
FRAP experiments were performed on an LSM5Live DuoScan (Zeiss; Plan-Apochromat 63×1.4, pixel size 120×120 nm, 12-bit) using 489 nm laser (BP500-525) for GFP. Two pre-bleach images were recorded to ensure stable imaging conditions. Bleaching was performed with a 488 nm point laser (100 mW, 30%, pixel dwell time 6.25 μs). Post-bleach-images were recorded until the mean fluorescent intensity of the bleached region of interest (ROI) reached saturation.
Images were processed using ImageJ software (NIH, Bethesda, MD). Usually, images were filtered using a 3×3 Median Filter, stacks were projected along the z-axis (maximum intensity), and brightness and contrast were adjusted for each channel. Deconvolution (QMLE algorithm) and surface rendering was performed with Huygens Professional using an experimentally derived PSF. Figures were finally composed in CorelDrawX4.
For statistical analysis, established counting techniques (Kim et al., 2006) were used and expanded. Briefly, for each PEX11 expression, images were collected of at least 50 cells randomly chosen at 24 hours post-transfection. All images were taken in the widest focal plane of a cell. Images were filtered, converted to 8-bit, and a threshold was applied to highlight the peroxisomal fluorescence. Peroxisomes were counted using the Particle Analysis package of ImageJ. Herein, peroxisomes were separated into four categories according to their diameter in μm, I 0–0.35, II 0.36–0.66, III 0.67–0.95, IV 0.96–10.
For plant cells, confocal images were acquired on a Leica TCS SP equipped with a KrAr laser using the following settings. For YFP-tagged (green channel) and for cherry–SKL (red channel), the laser emission was 476/568 nm and detection bandwidth was 500–535/600–635 nm, respectively. Chloroplast fluorescence (blue channel) was detected at 665–795 nm. The detector gain and amplifier offset were adjusted to avoid clipping, and the sequential imaging mode was used to ensure separated excitation and detection of the fluorescent proteins. To allow high resolution imaging of peroxisomes, the tissues were incubated in 500 μl 100 μM F-actin depolymerizing cytochalasin D (Sigma) for 30 minutes. This treatment led to immobile but otherwise normal peroxisomes (Mathur et al., 2002).
For ultrastructural analysis, cells grown on 12 mm Aclar discs (EMS, Hatfield, PA) were fixed in 3% glutaraldehyde (electron microscopy grade; Serva) in PBS for 60 minutes, osmicated in 2% veronal-acetate-buffered OsO4 for 60 minutes, dehydrated in a sequential series of ethanol, and embedded in Epon. Sections of 80 nm were stained with uranyl acetate and lead citrate and examined in a Tecnai-20 electron microscope (FEI, Eindhoven, The Netherlands) at 80 kV; images were acquired with a slow-scan CCD camera (Gatan, MSC 794).
The authors wish to thank Tom Rapoport (Harvard Medical School, Boston, MA) for the pAcGFP-Sec61β plasmid, Aviva M. Tolkovsky (University of Cambridge, Cambridge, UK) for the pmRFP-LC3 plasmid, Michael Schrader (University of Aveiro, Aveiro, Portugal) for the plasmid expressing CFP-DRP1. We are grateful to Josef Gotzmann (MFPL, Vienna) and Pavel Pasierbeck (IMP, Vienna) for technical assistance with CLSM. This study was supported by a joint grant to C.B., F.K. and A.H. focused on “Symbiosis and Molecular Principles of Recognition”, University of Vienna and grants from the Austrian Science Fund (FWF) to A.H. P19753 and C.B. P20803. F.K. is supported by the FWF, P19682-B03. C.B. is supported by the Elise-Richter Program of the FWF, B39-V09.
Note added in proof
During the reviewing process of this manuscript, the Schrader group (University of Aveiro, Portugal) reported a study on ectopic expression of a PEX11β–YFP fusion protein illustrating the formation of tubular peroxisomal accumulations (TPAs), in which matrix proteins are sequestered at the end of the peroxisomal tubules (Delille et al., 2010). The authors suggest that PEX11β–YFP affects the assembly of a functional fission complex, thereby inhibiting peroxisomal division, which explains the rapid kinetics of TPA formation as opposed to the slow process of JEP formation described in our study.