Main

The filamentous ascomycete, P. anserina, is extensively studied as an experimental model system to elucidate the molecular mechanisms of ageing3. Characteristic symptoms of ageing are a decrease of growth rate and fertility, a reduction in the formation of aerial hyphae and an increase in pigmentation. At the microscopic level, peripheral hyphae show abnormal branching and swelling, and finally burst. Ageing of P. anserina has a clear mitochondrial etiology4 with an age-related systematic reorganization of the mitochondrial DNA (mtDNA)5,6,7 — a hallmark of ageing in all wild-type strains. Various long-lived mutants with stabilized mtDNA have been characterized8,9 and one of these mutants is characterized by a pleiotropic phenotype resulting from mutation of Grisea, a gene encoding a copper-regulated transcription factor10,11,12,13.

To obtain a more detailed view of the differences in gene expression between wild-type 's' strain (hereafter referred to as wild type) and those mutant Grisea, a suppressive subtractive hybridization (SSH) analysis, using mRNA from juvenile cultures of the two P. anserina strains, was performed. After verification of the SSH data by northern blot analysis, the individual genes identified as being differentially expressed were cloned and sequenced. As ageing of P. anserina cultures has a strong mitochondrial component4, a putative gene affecting the morphology of these organelles was selected for further analysis. This gene, which is transcribed in juvenile cultures of the grisea mutant but not of the wild type, encodes a protein with significant homology to the dynamin-related protein 1 (Dnm1p) from various other organisms (see Supplementary Information, Fig. S1). The protein represents a large GTPase involved in mitochondrial fission14,15. The P. anserina gene was consequently termed PaDnm1. In senescent cultures of the wild-type strain, transcript levels of this gene were found to be much higher than in juvenile cultures (Fig. 1a). Staining of mycelia with Mitotracker Red (Fig. 1b) showed that juvenile (4 day) and middle-aged (15 day) wild-type cultures mainly contained filamentous mitochondria. In contrast, in the senescent phase (26 day) mitochondria were spherical, indicating that fragmentation increases during ageing, which correlates with the observed age-related increase in PaDnm1 expression.

Figure 1: Analysis of mitochondrial morphology in wild-type (WT) P. anserina and PaDnm1 deletion strain (PaDnm1::ble).
figure 1

(a) Transcription of PaDnm1 is induced in the senescent wild-type strain. (b) Fragmentation of the filamentous mitochondria in the senescent (sen; 26 d) culture. The scale bars represent 2 μm. (c) In the juvenile (juv) and middle-aged (ma) state, the PaDnm1::ble strain displays extremely elongated mitochondria (arrows point to a single mitochondrion in a hypha of the indicated strain). In the senescent phase of PaDnm1 deletion strains, mitochondrial fragmentation is observed as in wild-type cultures. The scale bars represent 2 μm. In b and c, the age of the isolate is indicated in the upper right corner.

Full size image

Although a clear correlation between age-induced PaDnm1 transcript levels and mitochondrial morphology was found in the wild-type strain, this was not the case in mutant grisea. Although PaDnm1 transcription is constitutive in this mutant, its mitochondrial morphology was similar to the wild type (data not shown). At first glance, these data seem to be contradictory. However, it must be emphasized that grisea is a pleiotropic mutant which, due to the loss-of-function of a transcription factor involved in regulating cellular copper levels, is depleted for copper. Consequently, various molecular changes occur and pathways involved in the control of mitochondrial dynamics may be affected.

To determine the function of PaDnm1p more clearly, experiments modulating PaDnm1 expression were performed. In wild type, overexpression of PaDnm1 under control of the P. anserina glycerolaldehyde-3-phosphate dehydrogenase promoter resulted in increased mitochondrial fragmentation without any other effects on phenotypic properties (see Supplementary Information, Fig. S2). In contrast, deletion of PaDnm1 (see Supplementary Information, Fig. S3) lead to an extreme elongation of mitochondria in the juvenile and middle-aged state, verifying the proposed function of PaDnm1 in mitochondrial fission (Fig. 1c). However, in senescent cultures of the PaDnm1 deletion strain (PaDnm1::ble), mitochondria were fragmented, as in senescent wild-type cultures.

Remarkably, PaDnm1::ble strains display an 11-fold increase in the mean life span compared with wild type (244 d versus 22 d; Fig. 2a). Re-introduction of a wild-type copy of PaDnm1 into PaDnm1::ble led to a reversion of the mutant phenotype (see Supplementary Information, Fig. S3). Further phenotypic characterization demonstrated that mycelial morphology, growth rate and male and female fertility of PaDnm1::ble are identical to wild type grown on standard cornmeal agar (BMM; Fig. 2b,c). However, both the germination frequency of PaDnm1::ble ascospores and the growth rate of PaDnm1::ble isolates were severely reduced on ammonium acetate supplemented BMM — the medium routinely used to induce germination of ascospores in wild-type strains (Fig. 2c, d). In contrast, PaDnm1::ble ascospores are able to germinate more efficiently on unsupplemented BMM than wild-type ascospores. At present, the reasons for these effects are unclear.

Figure 2: Phenotypic comparison between independent wild-type and PaDnm1::ble strains.
figure 2

(a) The mean life span of wild-type isolates is 22 d (n = 12). In contrast, PaDnm1::ble isolates (n = 43) display a mean life span of 244 d (U test, two-tailed, P <0.001 versus wild-type control). The mean life span of meiotic offspring from grPaDnm1::ble (105 d, n = 10) is also extended compared with mutant grisea isolates (65 d, n = 10; U test, two-tailed, P <0.001 versus mutant grisea control). (b) The morphology of middle-aged wild type and PaDnm1 deletion strains is identical on BMM. The distance between two blue arcs indicates the mycelial growth over two days incubation. (c) The PaDnm1::ble strain displays wild-type like growth rate (100% = 0.6 cm per day) and fertility on BMM (n = 10). The growth rate on BMM + AmAc (ammonium acetate), is reduced to approximately 75% (U test, two-tailed, P <0.001 versus wild-type control). Data represent the mean ± s.d. (d) PaDnm1::ble ascospores (n = 50) fail to germinate on BMM supplemented with 60 mM ammonium acetate (BMM + AmAc), however the germination rate is raised to around 50% on unsupplemented BMM. In contrast, germination of wild-type ascospores (n = 50) is facilitated on BMM + AmAc. Data represent the mean ± s.d.

Full size image

A molecular hallmark of ageing in wild-type strains of P. anserina is the reorganization of the mtDNA5,6,7. In marked contrast to these strains, in which a progressive accumulation of a circular DNA molecule (termed plasmid-like DNA, plDNA) was observed, the mtDNA of PaDnm1::ble isolates remained stable during the whole life span (Fig. 3a). Only in some very old PaDnm1::ble cultures was plDNA identified. However, old cultures of these isolates also contained mtDNA molecules with the integrated pl-intron, suggesting that significant amounts of intact mtDNA are also available in this strain in the senescent phase.

Figure 3: Analysis of age-related parameters in wild-type and PaDnm1 deletion strains of P. anserina.
figure 3

(a) In the wild-type strain, the amplification of a fragment of 2.5 kb corresponding to plDNA and the disappearance of the two mtDNA fragments of 4.5 and 1.9 kbp containing the integrated plDNA (pl-intron) is observed, which is a hallmark of the age-related reorganization of mtDNA in P. anserina. In contrast, the PaDnm1 deletion strains contain intact mtDNA molecules throughout the life span. A strong amplification of plDNA is not detectable. The age-related reorganization of mtDNA in wild-type strains seems to be suppressed. Identifiers (for example, F33610) classify the ascospore the mycelium originated from. (b) Release of the ROS species H2O2 from wild-type and PaDnm1::ble strains. In the senescent wild-type (26 d) and PaDnm1::ble (120 d) strains, a high amount of H2O2 is visualized as marked diamino benzidine (DAB) precipitation. The PaDnm1::ble isolate does not show any signs of elevated H2O2 release after 100 d growth. (c) The growth rate of PaDnm1:ble isolates on BMM + 0.03% H2O2 (1 mm per day) is significantly lower than wild-type isolates (1.8 mm per day; U test, two-tailed, P <0.05). Data represent the mean ± s.d. (n = 10, each strain). (d) Treatment of wild-type and PaDnm1::ble isolates with 0.03% H2O2 leads to mitochondrial fragmentation. The scale bars represent 2 μm. (e) PaDnm1 deletion strains (PaDnm1::ble and grPaDnm1::ble) show resistance against the apoptosis inducer, etoposide. Controls indicate growth rates on unsupplemented medium. Data represent the mean ± s.d. (n = 3 for each strain).

Full size image

Next, we analysed the release of the reactive oxygen species (ROS) hydrogen peroxide (H2O2). As shown in Fig. 3b, mycelial staining using diaminobenzidine demonstrated that senescent wild-type and PaDnm1::ble cultures released large quantities of H2O2. However, the release of H2O2 was much delayed in the PaDnm1::ble mutant (Fig. 3b). Eventually, there was also a concomitant increase in mitochondrial fragmentation and H2O2 production in the PaDnm1 deletion strain. The fact that mitochondrial fragmentation was not completely impaired in the PaDnm1 deletion strain suggests that other factors can compensate for the lack of PaDnm1p, but only do so in the terminal stages of the life cycle.

To evaluate the significance of mtDNA stabilization and the observed delay in ROS generation for life-span extension in the PaDnm1::ble strain, PaDnm1 was deleted in long-lived mutant grisea. In the resulting grPaDnm1::ble strain we observed that the formation of distinctly elongated mitochondria was similar to those found in the PaDnm1::ble mutant. The grPaDnm1::ble strain and mutant grisea displayed a similar mycelial morphology (see Supplementary Information, Fig. S3), growth rate and fertility (data not shown). Interestingly, mean life span of the double mutant was further extended, although not as pronounced as in the PaDnm1::ble single mutant (Fig. 2a). It seems that in this mutant, in which gene expression is greatly altered in comparison with wild type, additional factors affect life span. Among others, the reduction of functional superoxide dismutases (SOD1 and SOD2)11 may have a pivotal role in masking the full potential of the PaDnm1 deletion.

To gain insight into the mechanistic basis for the observed effects on mitochondrial dynamics, we asked how increased filamentation of mitochondria may affect ageing in P. anserina. Recently, we identified an apoptotic machinery in this ascomycete and assessed the impact of a gene coding for a metacaspase on life span16 (A.H., Brust, D. & H.D.O., unpublished observations). From mammalian cultures, it is known that downregulation of the PaDnm1 homologue Drp1 (by short interfering RNA (siRNA) or the expression of a dominant–negative form of the protein) leads to a marked resistance to different apoptotic elicitors17,18,19. As ROS are important in both ageing and the induction of apoptosis, we surmised that there may be a link between mitochondrial morphology, ROS and apoptosis. We addressed this possibility by growing wild type andPaDnm1::ble strains on H2O2-containing medium (Fig. 3c, d). From other systems, it is known that H2O2 initiates apoptosis through the induction of pro-apoptotic Ca2+ waves within mitochondria20. Interestingly, the PaDnm1::ble mutant displayed increased H2O2 sensitivity in comparison with wild type (Fig. 3c). These results are in agreement with data from HeLa cells containing filamentous mitochondria, which were found to be more sensitive to Ca2+-mediated apoptosis than their Drp1 overexpressing counterparts with fragmented mitochondria20. It is thus possible that the increased H2O2 sensitivity of PaDnm1::ble is due to the efficient propagation of ROS-induced Ca2+ waves along filamentous mitochondria. Growth of both wild-type and PaDnm1::ble strains on medium supplemented with 0.03% H2O2 eventually lead to fragmentation of mitochondria (Fig. 3d). Moreover, in such cultures, mitochondria stained with Mitotracker Red were only very rarely observed, suggesting that large regions of the mycelium are devoid of functional mitochondria.

Although the PaDnm1 deletion strain is more sensitive to external H2O2, the internal generation of this ROS seems to be delayed in comparison with wild type. These findings are in good agreement with the 'intracellular electric-cable hypothesis'21, which suggests that the delivery of energy and the diffusion of metabolites is more efficient in long filamentous mitochondria. The higher efficiency of these mitochondria was suggested to lead to lowered ROS generation21. In the PaDnm1 deletion strain, this seems to hold true for an extended period of time, but eventually H2O2 levels reached a threshold, leading to PaDnm1-independent fragmentation of mitochondria and, most likely, to the induction of an apoptotic programme.

To further verify such a link, wild type and various mutant strains were grown on medium supplemented with etoposide, frequently used to elicit apoptosis independent of pro-apoptotic mitochondrial Ca2+ waves19,20 (Fig. 3e). Significantly, the PaDnm1 deletion strains (PaDnm1::ble and grPaDnm1::ble), and two mutants in which genes of the apoptotic machinery of P. anserina are deleted (PaMca1::ble, encoding a metacaspase16 and PaAmid1, encoding an apoptosis inducing factor (AIF)-homologous mitochondrion-associated inducer of death16), were able to grow well on etoposide-supplemented medium (Fig. 3e). In contrast, growth of wild-type, mutant grisea and two PaDnm1 overexpressing strains was highly affected. These data strongly support the hypotheis that life-span extension in PaDnm1 deletion mutants is because of an increased resistance to Ca2+-wave independent apoptosis. Such a role has been demonstrated for the PaDnm1p homologue Drp1p in mammalian cells. If the function of this factor is affected, the release of the pro-apoptotic factor cytochrome c from mitochondria, and the induction of apoptosis, was delayed17,18,19.

The connection between the mitochondrial dynamics machinery and sensitivity to apoptotic stimuli has previously also been investigated in yeast, where it is known that Dnm1p promotes cell death on external apoptotic stimulation22. To determine the relevance of this process to ageing and whether the new pathway affecting ageing identified in P. anserina is conserved among organisms, we investigated the impact of the mitochondrial dynamics on ageing in S. cerevisiae. First, we investigated whether the relative abundance of different mitochondrial morphotypes15 changed during ageing and found that mitochondrial fragmentation increased in wild-type cells in both genetic backgrounds analysed (Fig. 4a). Similarly to P. anserina, deletion of the Dnm1 gene counteracted age-related fragmentation of mitochondria (Fig. 4a, b). Instead, the mitochondria of both young and old dnm1 cells predominantly formed nets, which, in many cells, occurred as one or more regions of densely packed mitochondrial tubules15 (Fig. 4a, b). The increase in the levels of H2O2 typically observed on ageing of cells was not as pronounced in the dnm1 mutant (Fig. 4c). This was not because of a low activity of the electron transport chain in dnm1 mitochondria, as the membrane potential of dnm1 mitochondria was similar to that of wild-type cells (Fig. 4d). Thus, reduced mitochondrial fission seems to allow a high mitochondrial activity without a concomitant increased generation of H2O2 (Fig. 4e). Strikingly, as in P. anserina, Dnm1p deficiency retarded ageing of the yeast FY10 cells (Fig. 5a). It also caused a pronounced effect on the fitness of old mother cells, such that the generation time of progressively older dnm1 cells remained constant, whereas the generation time of ageing wild-type cells increased significantly (Fig. 5b). We also examined the effect of a dnm1 deletion in a more long-lived strain of S. cerevisiae (BY4741). Both FY10 and BY4741 are derivatives of S288C, but they differ in a number of phenotypes, including their replicative life span. In BY4741, the dnm1 mutation increased the mean replicative life span of cells more modestly than in the FY10 strain (Fig. 5c). However, the dnm1 deletion also caused a pronounced effect on the fitness of old BY4741 dnm1Δ cells; the mutant maintained its 'virgin' fitness characteristics even after producing 30–40 daughter cells (Fig. 5d). Next, we examined whether the effects on life span and fitness are gene-specific or if other means of reducing mitochondrial fission caused a similar effect. The effects of deleting Fis1, another gene required for proper mitochondrial fission23, were examined and, similarly to dnm1, the fis1 mutation prolonged the life span of cells and improved the fitness of old mother cells (Fig. 5d). Age-specific mortality trajectories, calculated as previously described24, suggest that it is the rate of ageing, rather than the onset of ageing, that is affected by the deletion of Dnm1 and Fis1, respectively (see Supplementary Information, Fig. S4). Finally, introducing a mutation in fzo1, a gene required for mitochondrial fusion25,26, resulted in a life span and fitness similar to wild-type cells (data not shown). The observation that PaDnm1 overexpression in P. anserina and fzo1 deletion in yeast both caused mitochondrial fragmentation but did not reduce life span is interesting and can be explained by the observation that such fragmentation constitutively affected young cells and young mitochondria, whereas the fragmentation counteracted by the dnm1 and fis1 deletions is a process that occured progressively when the cells got older. Possibly, fragmentation of mitochondria in young cells can be compensated for by appropriate responses (for example, retrograde signalling) and that the ability to mount such a response is reduced when cells get older.

Figure 4: Analysis of mitochondrial morphology, ROS production and membrane potential in young and old wild-type and dnm1Δ S. cerevisiae.
figure 4

(a) Mitochondrial morphologies in the cell populations expressed as a percentage of total. A schematic representation is included to indicate the different morphologies as previously described15. (b) Representative images of mitochondria (Mitotracker) in young (left) and old (right) wild-type and dnm1Δ cells. The scale bar represents 5 μm. (c) H2O2 levels in young and old (10 generations) wild-type and dnm1Δ cells (n = 53 for each strain). (d) Mean membrane potential in young and old (10 generations) wild-type and dnm1Δ cells (n = 53 for each strain). (e) H2O2 levels (HP) normalized to membrane potential (MP) in young and old (10 generations) wild-type and dnm1Δ cells (n = 53 for each strain). Data represent the mean ± s.d. in c–e. The young cells are cells that have never budded or budded only once (checked with bud scar staining), whereas the old cells had produced 10 generations. Young and old cells were isolated by elutriation. The mitochondria were analysed directly after isolating old and young cells.

Full size image
Figure 5: Effect of dnm1 deletion on S. cerevisiae life span and fitness.
figure 5

(a) Replicative life span of the FY10 wild type (17.1 ± 0.1; n = 77) is decreased in comparison with dnm1Δ cells (25.5 ± 0.9; n = 80; Wilcoxon test, two-tailed, P = 1.04E-9). (b) Replicative fitness, measured as generation time, of progressively older FY10 wild-type (n = 77) and dnm1Δ cells (n = 80). (c) Replicative life span of span of the BY4741 wild-type (27.4 ± 0.1, n = 80), and otherwise isogenic dnm1Δ (31.5 ± 1.0; n = 66) and fis Δ (30.9 ± 0.1; n = 76) cells. Life span is modestly increased in the BY4741 strain background by dnm1Δ (Wilcoxon test, two-tailed, P = 0.01553) and fis1Δ (Wilcoxon test, two-tailed, P = 0.002602). (d) Replicative fitness, measured as generation time, of progressively older BY4741 wild-type (n = 80), dnm1Δ cells (n = 66) and fis1Δ (n = 76) cells. Data represent the mean ± s.d. in all panels.

Full size image

In summary, we identified the pathway controlling mitochondrial dynamics as a new element in the complex network control of ageing in two fungal ageing models. A fundamental finding of our study is that the increased life span achieved by a PaDnm1 deletion is not accompanied by an impaired fitness (for example, growth rate, fertility), as is the case for many longevity strains of P. anserina. Remarkedly, deletion of Dnm1 in both P. anserina and S. cerevisiae (and Fis1 in yeast) results in an extension of the 'health span' — the healthy period of time within the life cyle of organisms. It will be interesting to determine whether altered mitochondrial dynamics influence the life span of other model systems, and whether the corresponding pathways are part of a conserved network involved in the control of biological ageing.

Methods

P. anserina strains and cultivation.

Wild-type P. anserina27, long-lived mutant grisea8, the PaDnm1 deletion strains PaDnm1::ble and grPaDnm1::ble (genetic background of mutant grisea), as well as PaDnm1 overexpressing strains PaDnm1_Ex1 and PaDnm1_Ex2 were used in this study. '::ble' indicates that the gene of interest has been exchanged with a phleomycin resistance cassette. Strains were grown on standard cornmeal agar (BMM) at 27 °C.

S. cerevisiae strains and cultivation.

Yeast strains used in this study were BY4741 (EUROSCARF, Frankfurt, Germany) or FY10 (kindly provided by J. M. Shaw, University of Utah School of Medicine, Salt Lake City, UT) and were grown in YPD, following standard protocols.

Deletion of PaDnm1 in P. anserina.

Deletion of PaDnm1 in wild type was performed according to a previously described method16. Briefly, small flanking regions of the PaDnm1 gene were amplified using the 5′-flank oligonucleotides KoDnm1 (5′-TTGGTACCACGACTGACGATTTGCCC-3′) and KoDnm2 (5′-CCAAGCTTGATTTTCTGTGCAGGGCC-3′), and the 3′-flank oligonucleotides KoDnm3 (5′-AAACTAGTATATTCGGGGGTCAAGGG-3′) and KoDnm4 (5′-AAGCGGCCGCACATGATGGCGGAGAGGC-3′) for amplification. Recognition sites of restriction endonucleases are underlined. The 5′ fragment was digested with KpnI and HindIII and ligated into pKO3. The resulting plasmid was named pDnm1Ko1. The 3′ fragment was digested with BcuI and NotI and ligated into pDnm1Ko1. The resulting plasmid was termed pDnm1Ko2 and contained both fragments flanking a resistance cassette bearing phleomycin and blasticidin marker genes for fungal and bacterial selection, respectively. The resistance cassette with the flanking regions was excised by restriction with NotI and KpnI and used to transform Escherichia coli strain KS272 bearing the plasmid pKOBEG28, which contains the 'red' region of bacteriophage lambda, and the cosmid 42B8 containing the PaDnm1 locus. Homologous recombination between the flanks of the resistance cassette and cosmid 42B8 lead to generation of cosmid ΔDnm1#42B8, which contains the phleomycin–blasticidin cassette with large flanking genomic regions. The cosmid was isolated and used to transform wild-type P. anserina. Positive transformants (that is, PaDnm1 deletion strains) were selected by growth on phleomycin-containing medium. The cosmid also bears a hygromycin B resistance cassette, which is integrated on its ectopic integration into the genome. Positive transformants should be unable to grow on medium supplemented with hygromycin B. Reversion of the PaDnm1::ble strain was achieved by transformation of cosmid 42B8 into the mutant genome. Positive revertants were selected by growth on hygromycin B containing medium.

To construct a grisea PaDnm1::ble double mutant (grPaDnm1::ble), a homokaryotic PaDnm1::ble strain was crossed with a homokaryotic mutant grisea isolate. In contrast with heterokaryotic strains, homokaryotic ones cannnot self-fertilize and remain in a vegetative state until crossed with a strain of the opposite mating type. Meiotic offspring bearing the double mutation were identified after crossing the single-mutant strains by testing the growth of mycelia germinated from grey ascospores on phleomycin-containing medium. The deletion of PaDnm1 in the grPaDnm1::ble mutants was verified by Southern blot analysis as described below.

P. anserina life-span analysis.

Ascospores from the wild-type strain were isolated and germinated on BMM supplemented with 60 mM ammonium acetate. Ammonium acetate was omitted from the medium for efficient germination of PaDnm1::ble and grPaDnm1::ble spores. After three days of incubation, a small piece cut from the juvenile mycelium was placed at the fringe of a Petri dish containing 30 ml BMM. Growth was measured until the isolate stopped growing. Such cultures are defined as dead. The elapsed time to this point was recorded as the life span of the corresponding isolate.

S. cerevisiae life-span and fitness analysis.

Life-span analysis was performed by placing 70–80 exponentially growing cells on a Petri dish using a micromanipulator (Singer Instruments, Roadwater, UK) and allowing them to bud once. Mother cells were discarded and buds were used as starting virgin cells. Thereafter, buds were removed after every division and the number of times each mother cell divided, together with the time it took to do so, was recorded.

P. anserina fertility analysis.

To assess male fertility, mycelia from wild-type and PaDnm1::ble strains of opposite mating type were used to overgrow the surface of BMM plates. Incubation was in daylight at 27 °C. Spermatia (male gametes) were isolated by flooding the wild-type plate with 2 ml of sterile water. The suspension was collected in a 2 ml centrifugation tube and centrifuged for 10 min at 14,000g (5415 C, Eppendorf, Hamburg, Germany). The pellet of spermatia was resuspended in 0.2 ml sterile water. The number of spermatia was microscopically determined using a Thoma chamber.

To assess female fertility, mycelia from wild-type and PaDnm1::ble strains of opposite mating type were used to overgrow the surface of BMM plates. Incubation was in daylight at 27 °C. Spermatia were harvested by flooding the wild-type plate with 3 ml of sterile water and 2 ml of the suspension was collected in a 2 ml centrifugation tube. Drops (400 μl) from this suspension were pipetted onto mycelia of opposite mating type from wild-type and PaDnm1::ble strains. After 1 min, the drops were removed carefully. After 5–6 days incubation in daylight at 27 °C, perithecia were counted. The resulting values were divided by the area of the drop, which was approximately circular.

Determination of P. anserina ascospore germination.

Ascospores (approximately 50) were isolated from the following crosses: wild type (mat −) × wild type (mat +) and PaDnm1::ble (mat −) × PaDnm1::ble (mat +), respectively. Mat specifies the mating type of the mycelia. Single isolated spores were placed on either BMM medium or BMM medium supplemented with 60 mM ammonium acetate. The plates were incubated for 2 days at 27 °C in the dark, after which the percentage of germinated mycelia was determined.

SSH assay.

Total RNA was isolated from juvenile wild-type and mutant grisea isolates. The corresponding mRNA was purified using an mRNA purification kit (Amersham Pharmacia Biotech, Little Chalfont, UK). SSH was used to identify differentially transcribed genes between the juvenile wild-type and mutant grisea, using the Clontech-PCR-Select cDNA Subtraction Kit (Clontech, Mountain View, CA) according to the manufacturer's instructions.

Southern blot analysis.

Total DNA was isolated and digested with restriction endonucleases according to the manufacturer's instructions. The resulting fragments were separated on 0.8–1% standard TAE agarose gels. DNA was transferred onto nylon membranes and hybridized with DIG-labelled probes for 16 h at 65 °C in high-SDS hybridization buffer. For the detection of mtDNA rearrangements, a 2.5 kb SalI fragment of plasmid pSP17 containing the sequence of the first intron of the mitochondrial CoI gene was used. For the verification of PaDnm1 deletion strains, a 1.7 kb genomic NcoI–EcoRI PaDnm1 fragment and a 0.4 kb EcoRV–SmaI fragment of plasmid pKO3 containing a part of the phleomycin-resistance gene were used. Southern blots were washed twice after hybridization for 5 min at room temperature in 2× SSC 0.5% SDS and twice for 15 min at 68 °C in 0.1× SSC 0.1% SDS.

Northern blot analysis.

Total RNA was isolated and separated on a 1.5% formaldehyde agarose gel. RNA was transfered onto nylon membranes and hybridized with 32P-labelled probes for 16 h at 37 °C in standard formamide-containing buffer. For the detection of PaDnm1 transcript, a 1.7 kb genomic NcoI-EcoRI PaDnm1 fragment was used. As a RNA loading control, a 5.7 kb HindIII fragment of plasmid pMy60 containing rDNA sequence from Saccharomyces carlsbergensis was applied. After hybridization, blots were washed twice for 10 min at 37 °C in 2× SSC 0.5% SDS and once for 10 min at 37 °C in 0.1× SSC 0.1% SDS.

Staining techniques using fluorescent dyes.

P. anserina mycelia were grown on a glass slide which central depression containing a 1:1 mixture of BMM and 2% agarose for one day in a wet chamber at 27 °C. The mycelium was covered with 1 μM Mitotracker CMXRos (Invitrogen, Carlsbad, CA). After 2 min incubation, mitochondria were visualized using a fluorescence microscope (DM LB, Leica, Wetzlar, Germany) equipped with the appropriate excitation and emission filters. A digital camera system (Canon, Tokio, Japan) connected to the microscope was used for documentation.

For visualization of yeast mitochondria, H2O2 and of mitochondrial membrane potential, cells were stained with Mitotracker Green FM, dihydrorhodamine and DiOC6 respectively (Invitrogen, Carlsbad, CA). At least 100 cells were analysed for every strain or mutant. Old cells (10–12 generations) were obtained by centrifugal elutriation and assessed for the number of bud scars using Calcofluor white (Sigma, St Louis, MO).

P. anserina H2O2-release assay.

The basic procedure was adapted from a previously published protocol29. Briefly, mycelia were grown on synthetic medium for 2–3 days in the dark. The plates were flooded with a solution containing 2.5 mM diaminobenzidine and 0.02 mg ml−1 horseradish peroxidase (Sigma) and incubated for 1 h at 27 °C in the dark. The solution was subsequently poured off and the plates incubated again for 2 h at 27 °C in the dark. A scanner (Hewlett Packard, Palo Alto, CA) was used for documentation.

Apoptosis induction experiments.

The central cavity of object slides was filled with a 1:1 mixture of BMM and 2% agarose containing the apoptosis inducer etoposide (200 μg ml−1, dissolved in DMSO; Sigma). Pieces of the corresponding strains were placed onto the medium and incubated in a wet chamber in the dark at 27 °C for two days before growth rates were determined. As a control, experiments in which the growth of isolates was examined on medium containing DMSO without etoposide were performed.

Accession number for PaDnm1.

The genomic PaDnm1 sequence has been submitted to the EMBL nucleotide sequence database. The accession number is AJ972664.

Note: Supplementary Information is available on the Nature Cell Biology website.