Saccharomyces cerevisiae contains three dynamin-related-proteins, Vps1p, Dnm1p and Mgm1p. Previous data from glucose-grown VPS1 and DNM1 null mutants suggested that Vps1p, but not Dnm1p, plays a role in regulating peroxisome abundance. Here we show that deletion of DNM1 also results in reduction of peroxisome numbers. This was not observed in glucose-grown dnm1 cells, but was evident in cells grown in the presence of oleate. Similar observations were made in cells lacking Fis1p, a protein involved in Dnm1p function. Fluorescence microscopy of cells producing Dnm1-GFP or GFP-Fis1p demonstrated that both proteins had a dual localization on mitochondria and peroxisomes. Quantitative analysis revealed a greater reduction in peroxisome number in oleate-induced vps1 cells relative to dnm1 or fis1 cells. A significant fraction of oleate-induced vps1 cells still contained two or more peroxisomes. Conversely, almost all cells of a dnm1 vps1 double-deletion strain contained only one, enlarged peroxisome. This suggests that deletion of DNM1 reinforces the vps1 peroxisome phenotype. Time-lapse imaging indicated that during budding of dnm1 vps1 cells, the single peroxisome present in the mother cell formed long protrusions into the developing bud. This organelle divided at a very late stage of the budding process, possibly during cytokinesis.
Peroxisomes are single membrane-bound organelles that are ubiquitously present in eukaryotic cells. These organelles are involved in a wide range of metabolic functions, which vary with the organism in which they occur and with environmental conditions. However, two general functions are recognized, namely H2O2 metabolism and β-oxidation of fatty acids (reviewed by Van den Bosch et al., 1992; Purdue and Lazarow, 2001). In humans, peroxisomes are essential. Malfunctioning of these organelles as a result of inherited disorders results in severe abnormalities and might even be lethal (reviewed by Wanders and Waterham, 2005).
Peroxisomes have long been considered to be autonomous organelles that form by growth and division of pre-existing ones (Lazarow and Fujiki, 1985). Various morphological observations support the view that peroxisomes are indeed capable of dividing. This is reinforced by the identification of proteins that function at the peroxisomal membrane in organelle elongation (e.g. Pex11p, Pex25p, and Pex27p) or fission (Vps1p, Dnm1/Fis1p, Rho1p) (reviewed by Yan et al., 2005; Thoms and Erdmann, 2005).
Recent observations, however, suggest that peroxisomes may also form from the endoplasmic reticulum (ER). The first reports on the formation of peroxisomes from the ER came from elegant molecular and biochemical studies in the yeast Yarrowia lipolytica (Titorenko et al., 2000) and detailed electron microscopy and tomography on mouse dendrite cells (Geuze et al., 2003). More recently, detailed fluorescence microscopy studies in Saccharomyces cerevisiae and Hansenula polymorpha convincingly demonstrate that peroxisomes can originate from specialized regions of the ER (Hoepfner et al., 2005; Tam et al., 2005; Haan et al., 2006; Otzen et al., 2006). It is possible that the processes of peroxisome formation from the ER and fission of pre-existing organelles might both occur in cells of most organisms.
In yeast (Hoepfner et al., 2001), plant (Mano et al., 2004) and mammalian cells (Koch et al., 2003; Li and Gould, 2003), dynamin-related proteins (DRPs) have been shown to play a role in regulating peroxisome numbers. DRPs are GTPases that play important roles in membrane fission and fusion events. Originally, these proteins were proposed to act as mechanochemical fission factors (Hinshaw and Schmid, 1995; Marks et al., 2001), but a function as a classic signalling GTPase has also been suggested (Sever et al., 1999; Newmyer et al., 2003; Peters et al., 2004).
The S. cerevisiae genome encodes three DRPs, Vps1p, Dnm1p and Mgm1p. Of these, only Vps1p has been shown to be involved in peroxisome fission and proliferation (Hoepfner et al., 2001). Vps1p was initially identified as a protein required for vacuolar protein sorting (the Vps pathway) and localized to the Golgi apparatus (Vater et al., 1992). Now it is clear that Vsp1p also functions at other subcellular membranes, including vacuoles (Peters et al., 2004) and peroxisomes (Hoepfner et al., 2001; Marelli et al., 2004).
So far, in yeast, Dnm1p has only been localized to mitochondria, where it mediates organelle fission. In mammals, this task is performed by Dlp1, which also is involved in peroxisome fission (Pitts et al., 1999; Koch et al., 2003; Li and Gould, 2003). Mammalian Dlp1 and yeast Dnm1p are recruited to mitochondria by the outer membrane localized protein, Fis1p (Mozdy et al., 2000; Yoon et al., 2003). Recent data presented by the Schrader group revealed that a portion of the mammalian Fis1 protein is localized to peroxisomes, where it associates Dlp1p to the organelle (Koch et al., 2005).
These observations prompted us to re-examine the roles of S. cerevisiae Dnm1p and Fis1p in regulating peroxisome abundance. Details of these studies are presented in this paper.
Peroxisome numbers in oleate-induced S. cerevisiae dnm1 and fis1 cells
We analyzed peroxisome numbers in cells of S. cerevisiae DNM1 and FIS1 deletion strains, grown on glucose or in the presence of oleate, using wild-type (WT) cells and VPS1 deletion cells as controls. All strains produced GFP-SKL to label the peroxisomes. First, we performed a quantitative analysis of peroxisome abundance in the various strains and found a large variation in peroxisome numbers in oleate-grown WT cells (Fig. 1 and Table 1). In these cells, up to 12 fluorescent spots could be detected per cell. Most cells contained 2-7 fluorescent spots, with an average of ∼4.2 per cell (Fig. 1B, Table 1). The organelle numbers were reduced in oleate-induced dnm1 and fis1 cells, which showed comparable average numbers and frequency distributions (Fig. 1D,F, Table 1).
A greater reduction in organelle numbers was observed in vps1 cells grown in the presence of oleate. On average, vps1 cells contained a single fluorescent spot, although a significant fraction of the cells still harbored two or more spots (Fig. 1H, Fig. 2, Table 1). The lowest peroxisome abundance was invariably observed in cells of the dnm1 vps1 double-deletion strain, of which the cells contained a single peroxisome with only rare exceptions (Fig. 1J, Fig. 2). Similar average peroxisome numbers were observed in dnm1 vps1 cells when the peroxisomal membrane marker GFP-Ant1p was used instead of GFP-SKL, indicating that all peroxisomes were labelled by GFP-SKL (Fig. 2C, Table 1). The difference in peroxisome abundance between glucose or oleate-induced vps1 and dnm1 vps1 cells is reflected in a small, but significant difference in the average number of spots per cell (Table 1).
The reduction in organelle numbers in dnm1 and fis1 cells compared with WT controls was not observed when cells were grown on glucose (Fig. 1A,C,E, Table 1). Under these conditions, a reduction in organelle number was only observed in vps1 cells and was further pronounced in dnm1 vps1 cells (Fig. 1G,I, Table 1). The average peroxisome number of fis1 vps1 cells (0.90±0.03) was identical to that of dnm1 vps1 cells. Based on these data, we conclude that the peroxisome phenotype of dnm1 and fis1 cells was not evident in glucose-grown cells, in line with the earlier observations by Hoepfner et al. (Hoepfner et al., 2001), but was evident in oleate-induced cells and further reinforced in dnm1 vps1 double-deletion cells.
Representative fluorescence images of oleate-induced cells of WT and the various mutant strains are shown in Fig. 2A,C. In the dnm1 and fis1 cells, the number of organelles is clearly reduced, but no strong alterations in peroxisome morphology were observed relative to WT cells. By contrast, in vps1 and dnm1 vps1 cells, the enlarged peroxisomes often showed constrictions (Fig. 2A). Mitochondrial staining (Fig. 2B) revealed the expected alterations in mitochondrial morphology in dnm1, fis1 and dnm1 vps1 cells (one long tubular structure instead of several, branched mitochondria) (Bleazard et al., 1999). In vps1 cells the mitochondrial morphology was similar to that observed in WT cells.
To study whether deletion of VPS1, DNM1 or FIS1 affected organelle position in budding cells, we quantitatively analyzed the distribution of peroxisomes over mother cells and buds using fluorescence microscopy. In WT controls, grown on glucose or oleate, the expected peroxisome distribution pattern was observed: organelles accumulated in the neck region between the mother cell and the bud (Fig. 3A,B region 3) and were also abundant in the buds (Fig. 3A,B region 4). Comparable peroxisome distribution patterns were observed in each of the mutant strains (vps1, dnm1, fis1, dnm1 vps1), indicating that deletion of either of the genes encoding a DRP, although influencing total numbers, did not affect the patterns of peroxisome positioning in budding S. cerevisiae cells (Fig. 3C-J).
Organelle dynamics in dnm1 vps1 cells
Time-lapse videos were recorded by confocal laser-scanning microscopy (CLSM) to relate the process of peroxisome fission and inheritance in dividing WT, vps1 and dnm1 vps1 cells producing GFP-SKL. The data summarized in Fig. 4 are extracted from the videos of oleate-induced cells (Movies 1-5 in supplementary material). The time-lapse series presented in Fig. 4A shows that in WT cells, peroxisomes migrate into the buds at very early stages of their development (see also Movie 1 in supplementary material). Comparable patterns were observed in oleate-induced dnm1 and fis1 cells (data not shown). The remaining organelles are retained in the mother cells. The series of Fig. 4B shows that this process differs in oleate-induced vps1 cells. In these cells, elongated peroxisomes were often observed located in the neck between the mother cell and the bud. This morphology and position is similar to earlier observations by Hoepfner et al. (Hoepfner et al., 2001) in glucose-grown S. cerevisiae vps1 cells. However, more than one organelle was often also present in the mother cell before the onset of bud formation or at the initial stages of bud development. In such cells, one of these organelles migrates into the developing bud, in a similar manner to that observed in the WT control (see also Movie 2 in supplementary material).
The process of peroxisome segregation was different in dnm1 vps1 cells (Fig. 4C). Time-lapse videos revealed that the single elongated peroxisome protruded into the developing bud. This structure was maintained in this position until the very late stages of the cell division process (see also Movie 3 in supplementary material and Fig. 5). A detailed image of an elongated peroxisomal structure at this stage of yeast budding is shown in Fig. 5 (see also Movie 4 in supplementary material). This image, obtained by 3D CLSM, illustrates that at a very late stage of yeast budding, a single, elongated peroxisome protrudes from the mother cell into the bud.
Dnm1p and Fis1p are localized to mitochondria and peroxisomes
Our observation that Dnm1p plays a role in peroxisome abundance in yeast implies that the protein can be localized to these organelles. To study this, we analyzed WT S. cerevisiae, producing Dnm1-GFP and incubated with MitoTracker Orange to visualize mitochondria. As shown in Fig. 6A, most Dnm1-GFP fluorescence is observed as distinct spots at elongated mitochondrial structures, in line with earlier reports on Dnm1p localization (Bleazard et al., 1999). However, GFP fluorescent spots were regularly observed that did not co-localize with MitoTracker (Fig. 6A). To examine whether these spots were associated with peroxisomes, a strain was analyzed that co-produced the red fluorescent protein DsRed fused to a PTS1 (DsRed-SKL). In these cells, few of the Dnm1p-related green fluorescent spots co-localized with red fluorescence, in either glucose-grown cells (Fig. 6B, Movie 5 in supplementary material) or oleate-induced cells (data not shown). Association of Dnm1p-GFP with peroxisomes was not increased in oleate-induced cells relative to glucose-grown cells. These observations indicate that yeast Dnm1p is predominantly localized to mitochondria, but may also be present at peroxisomes.
Localization of Dnm1p-GFP in a fis1 deletion strain (Fig. 6C) revealed a strong reduction in the number of fluorescent spots, which is in line with earlier observations by Mozdy et al. (Mozdy et al., 2000). In these cells, no peroxisome-localized Dnm1-GFP was detected.
Finally, we analyzed the localization of Fis1p. As shown in Fig. 6D, a fusion protein consisting of GFP fused to full-length Fis1p (GFP-Fis1p) is mainly localized at large structures, which represent mitochondria. However, a portion of the protein is present in smaller spots, which co-localize with the peroxisomal marker protein DsRed-SKL (Fig. 6D). These findings indicate that in S. cerevisiae, both Dnm1p and Fis1p have a dual location on mitochondria and peroxisomes.
We show that in the yeast Saccharomyces cerevisiae, two dynamin-related-proteins (DRPs), Vps1p and Dnm1p, are involved in peroxisome proliferation. A function for Vps1p in regulating peroxisome abundance was reported before, both at peroxisome-inducing (oleate) (Li and Gould, 2003) and non-inducing (glucose) growth conditions (Hoepfner et al., 2001). Here we demonstrate that Dnm1p also plays a distinct role in regulating peroxisome abundance, especially in cells placed in peroxisome-inducing conditions.
Why two different dynamin-related proteins, Dnm1p and Vps1p, play a role in peroxisome abundance in S. cerevisiae is not known. In the absence of Vps1p, peroxisome numbers are strongly reduced, both in induced and non-induced growth conditions. However, when both dynamin-related proteins are absent (in dnm1 vps1 cells) the peroxisome number further decreases, indicating that Dnm1p is also involved in regulating peroxisome numbers in both glucose and oleate conditions.
So far, Dnm1p was shown to function with Fis1p at the outer mitochondrial membrane in mitochondrial fission (Otsuga et al., 1998; Mozdy et al., 2000). We demonstrate that the absence of either Dnm1p or Fis1p results in a comparable reduction of peroxisome abundance in cells grown in the presence of oleate. Moreover, we show that a portion of the Dnm1p and Fis1p is localized to peroxisomes. Our data suggest that peroxisome-localized Fis1p recruits Dnm1p to peroxisomes. In line with this suggestion is our observation that the average peroxisome number is the same when either DNM1 or FIS1 is deleted in glucose-grown vps1 cells.
The lowest peroxisome numbers were observed in cells of the dnm1 vps1 double deletion strain, which - both in inducing and non-inducing conditions - contained a single, enlarged peroxisome per cell. Interestingly, this organelle remained intact and positioned in the neck between mother and daughter cells during bud development. The time-lapse recordings indicated that the peroxisome fission event takes place at a very late stage of budding, possibly during cytokinesis. The actomyosin ring in the bud neck might exert sufficient force onto the elongated peroxisome, positioned in the bud neck, to cause its scission.
Inspection of time-lapse videos of S. cerevisiae vps1 cells indicated that in vps1 cells, organelle fission regularly occurred in non-budding cells or in mother cells at the initial stage of bud formation. However, single, elongated peroxisomes were also frequently observed, which extended from the mother cells into the bud before fission. Apparently two different modes of partition may occur in budding vps1 cells, namely (1) fission of the organelle in the mother cell before trafficking of an organelle to the bud or (2) protrusion of a single, elongated organelle into the bud and late fission.
The observed occurrence of peroxisome partitioning in budding dnm1 vps1 cells suggests that the mechanisms that control organelle transport and retention function normally, but all are directed to one and the same organelle. The motor protein Myo2p and the actin cytoskeleton are involved in trafficking of peroxisomes to the bud (Hoepfner et al., 2001), whereas the peroxisomal membrane proteins Inp1 and Inp2 are required for association of the organelle to cortical anchors and Myo2p, respectively (Fagarasanu et al., 2005; Fagarasanu et al., 2006). In line with this, one possibility to explain the excessive elongation of peroxisomes in dnm1 vps1 cells is that one side of the organelle is anchored in the mother with a concurrent pulling at another site of the same organelle to the bud. This could also explain the strong movements of the organelle in the bud. However, other explanations cannot be excluded.
Peroxisome transport requires proper assembly and polarity of the actin cytoskeleton. Yu and Cai (Yu and Cai, 2004) reported that deletion of VPS1 results in depolarization and aggregation of actin. However, Hoepfner et al. (Hoepfner et al., 2001) observed that the actin cytoskeleton is not disturbed in vps1 cells. Also, we did not observe any alteration in the actin cytoskeleton in the S. cerevisiae vps1 strains studied (data not shown).
Although peroxisomes have long been considered to be autonomous organelles that proliferate by growth and division of pre-existing peroxisomes, recent reports indicated that peroxisomes might also be formed from the ER (Hoepfner et al., 2005; Tam et al., 2005; Kragt et al., 2005; Haan et al., 2006; Otzen et al., 2006). However, several results are consistent with the view that peroxisome fission events also occur. In mammalian cells different stages of the fission process have been identified, namely organelle elongation, followed by constriction and subsequent fission (Koch et al., 2004). Our time-lapse videos of S. cerevisiae dnm1 vps1 cells revealed that in these cells, peroxisome proliferation is almost completely blocked. Hence, an intriguing question remains: what is the contribution of peroxisome fission and ER-dependent peroxisome formation to the total numbers of peroxisomes per cell?
Materials and Methods
Micro-organisms and growth conditions
Yeast strains used in this study are listed in supplementary material Table S1. Saccharomyces cerevisiae cells were grown in (1) selective media containing 0.67% yeast nitrogen base without amino acids (DIFCO) (YNB) supplemented with 1% glucose and 0.25% ammonium sulfate as a nitrogen source or (2) oleate induction media containing 0.67% yeast nitrogen base without amino acids, 0.1% glucose, 0.1% oleate, 0.05% Tween 40, and 0.1% yeast extract, pH 6.0 (Erdmann et al., 1989). Whenever necessary, media were supplemented with leucine (30 mg/l), histidine (20 mg/l) or lysine (30 mg/l). For growth on plates, 2% agar was added to the media.
For cloning purposes, Escherichia coli DH5α (Gibco-BRL, Gaithersburg, MD) was used and grown at 37°C in LB (1% trypton, 0.5% yeast extract, 0.5% NaCl), supplemented with 100 μg/ml ampicillin, when required.
Construction of strains
S. cerevisiae BY4742 vps1 (supplementary material Table S1) was modified by exchanging the kan-MX marker by a HIS3 marker using plasmid MBA-72 (M4754) (Voth et al., 2003). The resulting S. cerevisiae vps1 strain was used to delete DNM1 using the loxP-flanked kan-MX4 cassette obtained by PCR using primer pair RE1395/96 and pUG6 as a template. Correct gene replacement in G418-resistant transformants was verified by PCR using primers RE1386 and K3.
A strain producing the Dnm1-GFP fusion protein was obtained by transforming BY4742 WT with DNA fragments that were obtained by PCR using primer pair RE1384/85 and plasmid pYM12 as a template. Proper integration was checked by PCR using primers RE1386 and K3.
S. cerevisiae BY4742 fis1 Dnm1-GFP was constructed as follows: the FIS1 gene was deleted by using the loxP-flanked kan-MX4 cassette that was obtained by PCR using primers RE1695/96 and pUG6 as a template. Correct gene replacement in G418-resistant transformants was verified by PCR using primers RE1697/98. Subsequently the resistance marker was removed by the action of Cre recombinase using the plasmid pSH47 (Güldener et al., 1996). The resulting strain was designated fis1::lox entf. Subsequently, DNM1.GFP was integrated into this strain by the cassette obtained by PCR using primers RE1384/1386 and plasmid pYM12 as a template. Proper integration was confirmed by PCR using primers RE1386 and K3. The resulting strain was designated BY4742 fis1 Dnm1-GFP.
Plasmid pUG34DsRed.SKL enables PMET25-driven expression of DsRed.SKL gene (Monastyrska et al., 2005). For the construction of pUG34DsRed.SKL, a 720 bp BamHI (blunt ended by Klenow treatment)-SalI fragment containing the DsRed.SKL gene, was isolated from plasmid pHIPZ4-DsRed-T1.SKL and inserted between the XbaI (after Klenow treatment) and SalI sites of pUG34, thereby replacing the GFP gene.
Plasmid pMD23, which enables PMET25-driven expression of GFP.ANT1 (Palmieri et al., 2001) was constructed as follows: a fragment of 921bp (comprising the entire ANT1 ORF without the start codon) was obtained by PCR using primers RE1664/RE1665 and S. cerevisiae genomic DNA as a template. Subsequently this fragment was digested by EcoR1-Xho1, and cloned into EcoR1-Xho1-digested pUG36. The resulting plasmid was designated pMD23.
S. cerevisiae BY4742 WT, vps1, fis1, dnm1 and dnm1 vps1 were transformed with plasmid pRS6. S. cerevisiae BY4742 Dnm1-GFP and BY4742 fis1 Dnm1-GFP were transformed with pUG34DsRed.SKL (see supplementary material Table S2). The strain dnm1 vps1 GFP-Ant1p was obtained by transforming dnm1 vps1 with plasmid pMD23 (see supplementary material Table S2). S. cerevisiae WT BY4742 was transformed with plasmids pRS415 GFP-FIS1aa1-155 and pUG34DsRed.SKL.
Miscellaneous DNA techniques
Plasmids and primers used in this study are listed in supplementary material Tables S2 and S3. All DNA manipulations were carried out according to standard methods (Sambrook et al., 1989). S. cerevisiae cells were transformed using the lithium acetate (LiAc) method (Gietz and Sugino, 1988). DNA-modifying enzymes were used as recommended by the supplier (Roche, Almere, The Netherlands). Pwo polymerase was used for preparative polymerase chain reactions (PCR). Oligonucleotides were synthesized by Eurogentec (Seraing, Belgium). For DNA sequence analysis, the Clone Manager 5 program (Scientific and Educational Software, Durham, USA) was used.
For quantitative determination of the number of fluorescent spots per cell, images were prepared using a Zeiss Axioskop fluorescence microscope (Zeiss Netherlands B.v. Weesp, The Netherlands). All cells were included including those that did not display distinct fluorescence owing to an out-of-focus effect. All other fluorescence microscopy experiments (time-lapse imaging, co-localization) were performed using a Zeiss LSM510 confocal laser-scanning microscope (CLSM).
For quantitative experiments, cells were fixed in 4% formaldehyde in 10 mM potassium phosphate buffer pH 7.5 for 2 hours on ice. Fluorescent spots were counted in single cells or, for determining organelle position, in budding cells, which are defined as cells containing buds with a diameter of at least one-third of that of the mother cell. In each quantification experiment, ∼300 cells were counted (150 cells each from two independent cultures). Statistical differences in average numbers were determined using a Z-test.
Mitochondria were stained by incubation of intact cells for 30 minutes at 30°C with 0.5 μg/ml MitoTracker Orange CMTMRos (Invitrogen) followed by extensive washing with cultivation medium.
For time-lapse recordings, the temperature of the objective and objective slide were kept at 30°C. Eight z-axis planes were acquired for each interval. The laser power (Ar-ion laser; 30 mW; 488 nm) was set at 50% of its maximum value, the AOTF was tuned down to 0.5%. Images were prepared using ImageJ software (http://rsb.info.nih.gov/nih-image/) and videos were prepared in Animation Shop3 using Mpeg4 compression.
For preparation of 3D images, CLSM images were deconvoluted using Huygens professional software (Scientific Volume Imaging) via the Quick Maximum Likelihood method using a calculated PSF prior to construction of the surface rendered model using Amira 3.1 software (TGS). The cell wall was reconstructed by manual tracing the contours in the bright field images, fluorescent structures by using the isosurface function of Amira.
I.J.V.D.K. holds a PIONIER fellowship (ALW/NWO). M.D. and R.E. are supported by the Deutsche Forschungsgemeinschaft (SFB642 and ER178/2-4). We thank Janet M. Shaw, University of Utah, USA, for plasmid pRS415 GFP-FIS1aa1-155.
Supplementary material available online at http://jcs.biologists.org/cgi/content/full/119/19/3994/DC1
↵* These authors contributed equally to this work
- Accepted July 13, 2006.
- © The Company of Biologists Limited 2006