Main

Mitochondria are dynamic organelles in morphology and turnover. They form a branched, tubular network located below the cell cortex in budding yeast. In growing cells this network is a highly motile structure. Equilibrium between fission and fusion of tubular branches is essential for cellular adaptation to varying physiological conditions and to ensure inheritance of mitochondria by daughter cells.1, 2 These processes involve a series of proteins that have to deal with joining or separating both the outer and inner mitochondrial membranes in a coordinated fashion. Maintenance and inheritance of mitochondria also involve active movement along cytoskeletal elements.3 Mutants with defects in the fusion process contain numerous spherical fragmented mitochondria. In contrast, fission defects result in net-like agglomerations of mitochondria.1 There is growing support for the role of ion homeostasis in determining mitochondrial morphology. Recent work points to a crucial role of mitochondrial Ca2+ homeostasis in shaping mitochondria. Changes of intramitochondrial Ca2+ concentration and in Ca2+ signaling between ER and mitochondria trigger mitochondrial fragmentation at an early stage of cell death.4

Systematic analysis of a yeast deletion mutant collection has led to the identification of a large series of genes as being directly or indirectly involved in shaping mitochondrial morphology.5 One of those genes, MDM38/YOL027c, also showed up in a screen for genes involved in mitochondrial cation homeostasis. Its product is an integral protein of the inner mitochondrial membrane, and is well conserved among all eukaryotes.6 Hemizygous deletion in humans has been invoked as causing a major part of the Wolf–Hirschhorn disease phenotype.7 The presence of a putative EF Ca2+ binding hand domain within almost all Mdm38p homologues suggested a role for Mdm38p in mitochondrial Ca2+ homeostasis and morphology.5, 8 We have recently characterized Mdm38p/Mkh1p as an essential component of the mitochondrial K+/H+ exchange system. Mutants deleted for this gene have functional defects of mitochondria, reduced growth of cells on non-fermentable substrate (petite phenotype), a nearly complete loss of mitochondrial K+(Na+)/H+ exchange activity, high K+ matrix content, low membrane potential (Δψ), and dramatically increased volume.6, 9 Recently, Frazier et al. (2006) noted reduced amounts of certain mitochondrially encoded proteins in the inner membrane despite their translation efficiency being unchanged. They found that protein A-tagged Mdm38p formed complexes with mitochondrial ribosomes, and suggested that the reduced amount of some of the mitochondrially encoded proteins of complexes III, IV, and V resulted from the lack of Mdm38p interacting with mitochondrial ribosomes to mediate membrane insertion of several proteins of nuclear and mitochondrial origin.10 These data led Frazier et al. to propose that Mdm38p is acting as a component of the mitochondrial protein export machinery.

Making use of a yeast strain conditionally expressing Mdm38p, we address the following questions. Which is the primary phenotype arising from the deletion of MDM38? Is the reduction of mitochondrial-encoded proteins a direct effect of the lack of Mdm38p, or a secondary consequence resulting from the disturbance of mitochondrial K+ homeostasis, swelling and/or reduction in Δψ? Since we observed mitochondrial fragmentation upon Mdm38p shut-off, and the association of mitochondrial fragments with vacuoles, we asked whether breakdown of the mitochondrial reticulum due to osmotic swelling led to mitochondria-selective autophagy, which has been termed mitophagy.11, 12

Results

Disturbance of mitochondrial translation and membrane insertion is not a primary effect of mdm38Δ

Yeast strains disrupted for the MDM38 gene (mdm38Δ) exhibit a growth defect on non-fermentable substrate (petite phenotype) and altered mitochondrial morphology,5 reduced amounts of certain mitochondrially encoded proteins, reduced Δψ6, 13 and a near total loss of mitochondrial K+/H+ exchange activity associated with increased K+ content and osmotic swelling.6, 9 Figure 1 presents the pattern of mitochondrially translated proteins obtained by incubation of wild type and mdm38Δ cells with labeled methionine in the presence of cycloheximide. Consistent with previous observations,13 we observe a reduction in abundance of Cox2, Cob, and Cox3, but no apparent changes in ATP synthase proteins. Cytochrome spectra also revealed reductions in the bc1 and a.a3 spectral bands (data not shown). Similar results were obtained when we compared the levels of mitochondrial membrane proteins as detected by BN gel electrophoresis and immunoblotting (Supplementary Figure S1). Addition of nigericin, a K+/H+ ionophore, to mdm38Δ cells efficiently restored both growth of these cells on non-fermentable substrate (Figure 2) and the pattern of mitochondrially synthesized proteins (Figure 1). These results are difficult to reconcile with the suggestion that the primary role of Mdm38p is in the insertion of proteins from the mitochondrial matrix into the inner membrane.13 Rather, they support our proposal that a lack of K+/H+ activity causes the mdm38Δ growth defect6 and that effects on the pattern of mitochondrially synthesized proteins are a result thereof.

Figure 1
figure 1

Effects of the mdm38Δ mutation on the expression pattern of mitochondrially encoded proteins. Upper panel: Cultures of cox2Δ, cox3Δ, and mdm38Δ mutant and wild-type cells were grown and pulse-labeled with 35S-methionine in the presence of cycloheximide as described under ‘Materials and Methods’. mdm38Δ cells were grown in the presence or the absence of nigericin as indicated. Total cell protein (15 μl loaded per fraction) was separated by 16% SDS-PAGE and subsequently analyzed by autoradiography. The mit− cox2 and cox3Δ strains serve as reference to identify cox2p and cox3p bands. Lower panel: immunoblots of the same protein extractions as shown in the upper panel (5 μl loaded per fraction) were probed in parallel with an Mdm38p antiserum and with hexokinase (Hxk1p) antiserum serving as a loading control

Full size image
Figure 2
figure 2

Suppression of the mdm38Δ growth phenotype by nigericin. Serial dilutions of wild type and mdm38Δ strains were spotted onto fermentable (YPD) and non-fermentable (YPG) media containing or lacking nigericin (2 μM) as indicated. YPD plates were incubated at 28°C for 3 days, YPG plates for 5 days

Full size image

To differentiate further between primary and secondary phenotypic effects of the mdm38Δ mutation, we ectopically expressed the MDM38 gene under the control of a doxycycline (dox) regulated promoter in an mdm38Δ deletion strain (designated MDM38dox). As shown in Figure 3 (upper panel), the amount of Mdm38p decreased below the level of detection within 25 h (t25) following downregulation of its expression. The pattern of mitochondrially synthesized proteins observed with pulse labeling assays, however, remained essentially unchanged up to 50 h following downregulation (Figure 3, lower panel). Equally, steady-state levels of mitochondrially encoded Cox subunits remained constant for up to 50 h following MDM38 downregulation (Supplementary Figure S2). In addition, cytochrome bc1 and a.a3 spectral bands were comparable to those of wild-type cells up to 50 h following Mdm38p downregulation (not shown).

Figure 3
figure 3

Synthesis of mitochondrial proteins continues during Mdm38p depletion. Mutant mdm38Δ cells expressing Mdm38p under the control of dox from a CEN plasmid pCM189 (MDM38dox) and wild-type cells carrying the same plasmid without the MDM38 gene (WT) were grown in SGal medium in the presence or the absence of dox for up to 50 h. Cells were kept in logarithmic growth phase by daily re-dilutions. Cells were harvested at indicated times following 35S-methionine labeling in the presence of cycloheximinde (as indicated in Figure 1, except that the duration of the chase was either 10 or 120 min), proteins were separated using 16% SDS-PAGE. Upper panel: samples (5 μl loaded per fraction) were immunodecorated with an Mdm38p antiserum to observe dox-induced depletion or with GFP or hexokinase (Hxk1p) antisera (loading control). Lower panel: autoradiography of same samples (15 μl loaded per fraction)

Full size image

Primary effects of mdm38Δ mutation on mitochondrial K+/H+ exchange activity

Having established that reduction in the amount of mitochondrial proteins is not an immediate effect of the MDM38 gene deletion, we sought to determine its primary effect. For this purpose, we carried out further series of time-course experiments, making use of the MDM38dox strain.

Hypoosmotic potassium acetate (KOAc) treatment of wild-type mitochondria results in rapid swelling, which reflects K+/H+ exchange activity.14 It can be observed as a decrease in light absorbance of mitochondrial suspensions. As described in our previous studies,6 mutant mdm38Δ mitochondria exhibit severely reduced KOAc-induced swelling in contrast to wild-type mitochondria. Mitochondria isolated from MDM38dox cells initially (up to t16 of Mdm38p depletion) displayed a high swelling amplitude (Figure 4a). At later times, following dox-mediated shut-off of Mdm38p expression the swelling amplitude reached first intermediate levels (t20), then became marginal (t25 to t50). This correlates with the level of Mdm38 protein at these time points (Figure 4b). Mitochondria of wild-type cells treated similarly with dox for a time period up to t50 did not show any significant change of their swelling amplitude (data not shown). These results are consistent with a primary role of Mdm38p in K+/H+ exchange.

Figure 4
figure 4

Loss of K+/H+ exchange activity and Δψ correlate with Mdm38p depletion. MDM38dox cells and wild-type cells expressing the tetO promoter from the empty plasmid pCM189 (WT) were transformed with pYX232-mtGFP and grown in SGal-medium in the presence of dox for up to 50 h. (a) KOAc induced swelling of mitochondria isolated from MDM38dox cells at various times of dox treatment was measured as a decrease of OD540. Duration of dox treatment was –⋄– 5 h, –▴– 20 h, –▪– 25 h, –□– 50 h. (b) Depletion of Mdm38p was monitored by Western blotting. Hxk1p and mitochondrially targeted GFP served as loading controls. (c) Mitochondrial Δψ was determined by JC-1 staining. Shown are mean values of three measurements expressed individually as percentage of the maximum and the minimum obtained by hyperpolarization with nigericin and depolarization with FCCP, respectively. Black bars WT, grey MDM38dox. Error bars represent S.D.

Full size image

As a further parameter, we determined the mitochondrial Δψ from the same preparations used for swelling and immunoblotting experiments (Figure 4a and b, respectively). Mdm38p depletion resulted in a decrease of Δψ starting at t25 (Figure 4c) while dox treatment of wild-type cells did not show any reduction in Δψ.

Ultrastructural changes of cell organelles in Mdm38-depleted cells

As shown previously by electron microscopy, one hallmark of mdm38Δ cells is drastically enlarged mitochondria that are nearly devoid of cristae structures and have reduced electron density.6 This phenotype is in accordance with a defect in K+/H+ exchange activity, increased K+ content and osmotic swelling.6, 9 Here, we observed changes in mitochondrial and vacuolar morphology resulting from gradual depletion of Mdm38p in MDM38dox cells. As soon as K+/H+ exchange activity became low (beginning with t25 Mdm38p depletion), we observed spherical mitochondria with low electron density and few remnants of internal membrane structures (cristae) (Figure 5a, middle panel). Most interestingly, mitochondria appeared to be closely associated with the vacuole. The electron density of the vacuolar membrane was strongly increased and exhibited concave indentations all over (Figure 5b, middle panel and Figure 5c, middle panel). These vacuolar membrane formations have previously been described to occur in mitophagy.11, 15, 16 In fact, at t25 of Mdm38p depletion vacuoles appeared to protrude towards mitochondria and the electron density of both interacting organelles was strongly reduced (Figure 5b, left panel). Ongoing proteolysis could be a possible explanation for this phenomenon. Mitochondrial particles exhibiting low electron density were eventually engulfed by the vacuoles together with or without surrounding cytoplasm (Figure 5b, right panel). Cells exhibiting these particular features of mitochondria and vacuoles were abundant at t25 to t29 of the Mdm38p depletion experiment. At t50 they were still observed, but less frequently. Instead, numerous spherical mitochondria appeared distributed in unconnected manner throughout the cell (not shown). Selective degradation of organelles has been described to occur via microautophagy or macroautophagy.12, 17 Evidence of direct engulfment of mitochondrial material surrounded with cytosolic material within the concave membranes of vacuoles (Figure 5b, right panel) strongly suggests that microautophagy is occurring here. Macroautophagy cannot be excluded, but in contrast to microautophagy, a complete sequence of autophagosome formation events was not observed. None of the changes in vacuolar and mitochondrial morphology were observed in wild-type cells treated with dox (Figure 5c, left panel).

Figure 5
figure 5

Changes in mitochondrial morphology of MDM38dox cells after downregulation of Mdm38p and following addition of nigericin. Electron micrographs were prepared from cells grown to early logarithmic phase (OD600=0.5) in SGal-medium in the presence of dox when indicated. (a) Comparison of mitochondria from a wild-type strain expressing the tetO promoter from the empty plasmid pCM189 (WT, left) cell, a MDM38dox cell grown for 29 h in the presence of dox (middle), a MDM38dox cell grown for 29 h in the presence of dox and nigericin added for the final 5 h of the growth period (right) scale bars: 200 nm. (b) Interactions between vacuoles and mitochondria in MDM38dox. Vacuolar extension towards mitochondria dox treatment for 25 h (left); Mitochondrial particles with low electron density in proximity of vacuole membrane indentations (middle); mitochondrial particles engulfed by vacuoles, dox treatment for 29 h (right). Scale bar: 200 nm. (c) Structure of mitochondria (M) and vacuole (V) in wild-type (WT, left) and MDM38dox cells grown in presence of dox and in absence (middle) or presence (right) of nigericin. Scale bar (left) 0.5 μm, (middle and right) 1 μM

Full size image

To test whether nigericin can rescue the mutant phenotypes of mitochondria and vacuoles, this K+/H+ ionophore was added to MDM38dox cells grown for 25 h in presence of dox and then incubated in the presence of both drugs for further 4 h. As shown in Figure 5a (right panel), mitochondria appeared as elongated electron dense structures with abundant cristae, although these were swollen in comparison to those of mitochondria in wild-type cells. Moreover, the morphology of vacuoles was also restored, since the vacuolar membrane showed low electron density as shown by wild-type cells, no indentations and there were no larger inclusions visible in the vacuole (Figure 5c, left and right panels, respectively).

Taken together, these data reveal that Mdm38p depletion resulted in an early loss of K+/H+ exchange-mediated swelling capacity of mitochondria, loss of Δψ and in the extensive interaction of mitochondria with vacuoles, a phenomenon previously named mitophagy.11, 12 Reversion of these phenotypic features was observed following addition of the ionophore nigericin to growing cells. This strongly supports the notion that the loss of K+/H+ exchange activity is the primary cause for morphological changes of mitochondria, which in turn trigger the process of mitophagy.

Confocal microscopy of cells undergoing mitophagy induced by Mdm38p depletion

Figure 6 depicts mitochondrial structures observed by confocal microscopy in wild-type cells (Figures 6a–c) and in mdm38Δ mutant cells (Figures 6d–f) expressing GFP targeted to the mitochondrial matrix. In mutant cells (d–f) a large number of individual, spherical units were present instead of the highly branched mitochondrial network of wild-type cells (a–c). Similar structures were seen when GFP was directed to the outer mitochondrial membrane (data not shown). Accordingly, the distinct spherical units in mdm38Δ cells did not result from cristae constriction, but from a full breakdown of the mitochondrial network.

Figure 6
figure 6

Disturbed mitochondrial morphology in mdm38Δ cells. Wild type (a–c) and mdm38Δ cells (d–f) were transformed with pVT100U-mtGFP,18 grown on SGal-ura medium to early exponential phase, stained with 0.5 μM rhodamine B hexyl ester and visualized by laser confocal microscopy. (a) and (d) GFP fluorescence, (b) and (e) rhodamine fluorescence, (c) and (f) overlay of GFP and rhodamine fluorescence

Full size image

During Mdm38p depletion from MDM38dox cells we observed that mitochondrial tubules initially became shorter and the regular distribution of elongated tubular networks throughout the cell cortex changed into crown-like networks of thick, but short branches surrounding the vacuole and were eventually fragmented into distinct spherical organelles (not shown). Table 1 indicates the percentage of cells exhibiting fragmented mitochondria observed over the period of Mdm38p downregulation in comparison to a strain constitutively expressing Mdm38p.

Table 1 Quantification of mitochondrial fragmentation upon doxycycline dependent repression of Mdm38p
Full size table

Simultaneous counterstaining of the vacuole membrane with the dye FM4-64 illustrated vacuolar membranes with indentations and pointed to some green and red overlaps after Mdm38p depletion, consistent with ongoing mitophagy. Parallel cultures of MDM38dox cells, grown in the presence of dox and nigericin, did not show any indications of close proximity of vacuoles and mitochondria (not shown).

To monitor further the fate of mitochondria upon Mdm38p depletion, we made use of a biosensor targeted specifically to mitochondria by means of the citrate synthase targeting sequence. This biosensor consists of two fluorescent proteins, a pH-sensitive variant of GFP and a pH-insensitive variant of the RFP, DsRed. In the acidic environment of the vacuole (pH ∼5.5), the pH-sensitive GFP loses its fluorescence, whereas the red fluorescence expressed from the same construct is retained. Thus, this biosensor reports mitochondrial uptake by accumulation of red fluorescence within yeast vacuoles. For wild-type cells grown under non-starvation conditioned few cells show mitochondrial uptake; however, under starvation conditions mitochondrial uptake is dramatically enhanced (C Rosado, D Mijaljica, M Prescott and RJ Devenish, unpublished observations).

We co-transformed MDM38dox with the biosensor construct pCS-G/RFP (this strain was denoted MDM38dox-G/RFP) and monitored the changes in green and red fluorescence using confocal microscopy (Figure 7B). Time dependent depletion of Mdm38p was revealed in parallel by Western blotting (Figure 7A). The cells were grown to the early logarithmic stage for periods of 4, 10, 20, 25, 29, or 50 h in presence of dox, or dox and nigericin as indicated. Green fluorescence clearly documents the normal branched reticular network of mitochondria of wild-type cells. Red fluorescence colocalizes with green fluorescence as expected in wild-type cells, and there is little evidence for mitochondrial uptake into the vacuole since no red fluorescence is seen there (Figure 7B, panel a). Growth of wild-type cells in presence of dox for 50 h had no effect on the mitochondrial structure as can be seen in Figure 7B, panel b. In contrast, at early time points of growth of MDM38dox-G/RFP cells in the presence of dox, the gradual alteration in mitochondrial morphology, from elongated to rounded forms, became evident. Similar to wild-type cells (Figure 7B, panel a), red fluorescence was first fully colocalized with green fluorescence as shown in the overlay image (Figure 7B, panels c–e). However, at a later time point (t25), correlating to the loss of mitochondrial K+/H+ exchange activity and ongoing swelling, mitochondrial fragmentation became apparent and some fragmented organelles were found lacking green fluorescence but retaining red fluorescence (Figure 7B, panel f). Thereafter (t29 and t50) significant accumulation of red fluorescence in vacuoles was detected, reflecting turnover of mitochondria in vacuoles (Figure 7B, panels g and i). In line with the EM data presented above, the appearance of some mitochondria showing red fluorescence only and thereafter of red fluorescence in vacuoles suggests that some mitochondria become acidic before being ingested. In a parallel experiment, MDM38dox-G/RFP cells were grown for 29 and 50 h in the presence of dox with nigericin also present for the last 4 h. Remarkably, the mitochondria were fragmented to a much lesser extent and more strikingly, these cells did not display any accumulation of red fluorescence in the vacuoles (Figure 7B, panels h and j), suggesting little if any turnover of mitochondria was occurring. These results confirm that in the presence of the K+/H+ ionophore nigericin mitochondrial tubular networks have been re-established either by fusion of persisting spherical organelles, or from newly formed organelles, or both. Presumably, such mitochondria are no longer marked for vacuolar degradation even in the absence of Mdm38p.

Figure 7
figure 7

Mdm38p dependent mitochondrial turnover. (A) Western blot detection of Mdm38p during dox-dependent growth in wild type expressing the tetO promoter from the empty plasmid pCM189 and G/RFP from the plasmid pCS-G/RFP (WT) and in MDM38dox–G/RFP cells. Whole cell extracts corresponding to 1 ml (OD600=4) were separated by 12.5% SDS-PAGE and probed with antiserum against Mdm38p. Blot probed with antiserum against Hxk1p served as a loading control. (B) Confocal microscopy. Control wild-type strain expressing the empty pCM189 and pCS-G/RFP in absence (a) or in presence of dox (b). MDM38dox cells transformed with the plasmid pCS-G/RFP and grown for 4, 10, 20, 25, 29, or 50 h under Mdm38p depletion conditions (c–g, and i). Cells from the same pre–culture as in (g and i), but grown for 29 or 50 h in the presence of both dox and nigericin added for the last 5 h of the growth, respectively (h and j)

Full size image

To learn whether mitophagy induced by dysfunctional mitochondria contributes to cell survival, aliquots of logarithmic growth phase cells taken from each time point of Mdm38p depletion were stained with the vital stain Trypan blue and the percentage of dead cells was evaluated. Both, wild type and MDM38dox-G/RFP cells exhibited a cell death rate of 0.5–3% per generation. Thus, induction of mitophagy (clearly observed from t25 on) has no significant effect on cell viability. Interestingly, we noticed that Mdm38p depletion had a strong negative effect on survival of cells in stationary growth phase. Cells grown for 50 h in presence of dox were kept in the original galactose-containing growth medium and incubated with shaking at 28°C for a week. Every day samples were taken and stained with Trypan blue. Counting of Trypan blue-positive cells indicated a significant decrease in the survival rate of Mdm38p depleted cells in comparison to wild-type cells (Figure 8).

Figure 8
figure 8

Cell survival in stationary phase. Wild type (WT) and Mdm38p-depleted cells (MDM38dox) were kept in stationary phase for 7 days in presence of 5 μg/ml dox. Cell viability was determined each day using Trypan blue staining and counting of 103 cells under light microscopy. % survival is the percentage of Trypan-blue-negative (unstained) cells. The results present the average of three different experiments

Full size image

Fragmentation and autophagic degradation of mdm38Δ mitochondria require the presence of Dnm1p

As described previously, the dynamin-related GTPase Dnm1p mediates scission of the outer mitochondrial membrane. Mitochondrial membranes in the dnm1Δ strain have been shown to be collapsed into one or more long tubular structures near the cell cortex and often a planar ‘net’ of interconnected tubules.19, 20 We addressed the question whether fragmentation of mdm38Δ mitochondria could be restored by changing the balance towards reduced fission or enhanced fusion. We first constructed a dnm1Δmdm38Δ strain expressing the mitochondrial targeted GFP and then examined the mitochondrial network. Confocal microscopy of dnm1Δmdm38Δ cells expressing mitochondrially targeted GFP revealed the presence of giant mitochondrial spheres without any obvious tubulation. The dnm1Δ mutation thus prevented the fragmentation of mitochondria into a larger number of spherical units typical of mdm38Δ single mutant cells. Simultaneous counterstaining of vacuolar membranes with the dye FM4-64 (not shown), or expression of pCS-G/RFP in dnm1Δmdm38Δ cells revealed that dnm1Δ suppressed mdm38Δ induced mitophagy (Figure 9a). Transmission electron microscopy of dnm1Δmdm38Δ cells (Figure 9b and c) confirmed the presence of giant mitochondrial spheres. Formation of these giant spheres is assumed to reflect osmotic swelling of mitochondria due to the lack of Mdm38p combined with a fission defect due to the dnm1Δ mutation. The lack of association between mitochondria and vacuoles in the double mutant suggests that mitophagy is prevented by the fission defect and that fragmentation of mitochondria may be a prerequisite for the degradation process. Interestingly, growth of dnm1Δmdm38Δ cells was much reduced compared to mdm38Δ cells, both on fermentable and non-fermentable substrates (not shown). Western blotting revealed only traces of Cox2 protein and reduced amounts of ATPase F1β subunit in dnm1Δmdm38Δ mitochondria (Supplementary Figure S1A and B). This is consistent with the near total absence of cristae structures.

Figure 9
figure 9

Structure of dnm1Δ mdm38Δ cells. (a) Nomarski and fluorescence microscopy of dnm1Δmdm38Δ cells expressing pCS-G/RFP. (b) EM of dnm1Δmdm38Δ cells grown to early logarithmic phase in YPGal. V, vacuole; M, mitochondrion; N, nucleus. Scale bar, 2 μM. (c) Details of (b)

Full size image

Finally, we asked if mitochondrial fragmentation and mitophagy could be compensated by enhancing fusion activity. Therefore, we overexpressed Fzo1p or Mgm1p, both proteins mediating mitochondrial fusion, in mdm38Δ cells. Neither protein could counteract the process of mitochondrial fragmentation and mitophagy or rescue the mdm38Δ petite phenotype (data not shown), indicating that fragmentation of mdm38Δ mitochondria is not caused by low activity of proteins that promote mitochondrial fusion.

Discussion

Conditional shutdown of MDM38 gene expression has been used here to discriminate between primary and secondary phenotypic effects resulting from the loss of Mdm38p in yeast cells. While the pattern of mitochondrially synthesized proteins as well as of cytochrome spectra remained unaffected even after prolonged Mdm38p depletion, mitochondrial K+/H+ exchange activity diminished with the loss of Mdm38p. Accordingly, we conclude that this is the predominant and only known phenotype immediately resulting from Mdm38p depletion. Strong support for this conclusion comes from our observation that nigericin, a K+/H+ ionophore, restores growth of Mdm38p depleted cells on non-fermentable media. Loss of mitochondrially synthesized proteins,13 therefore, is not a primary phenotypic feature of Mdm38 depletion.

Loss of K+/H+ exchange activity in mdm38Δ cells resulted in mitochondrial K+ overload and osmotic swelling.6, 9 As shown here by EM and confocal microscopy, these events were paralleled by the conversion of the tubular mitochondrial network into many distinct spheres and their association with the yeast vacuole. Moreover, the vacuolar membrane was found to be considerably thickened and to exhibit indentations all over. This pathway has features similar to those recently described, where the vacuolar membrane exhibited concave indentations associated with microautophagy of mitochondria.11, 16 It is unclear at present how selective this microautophagy might be. However, recent studies implicate mitochondrial membrane depolarization and swelling of the matrix compartment in the onset of selective macroautophagic degradation processes.21, 22

Here, we made use of electron microscopy and a novel biosensor composed of a pH-sensitive GFP and a pH-insensitive RFP, targeted to mitochondria. Upon Mdm38p depletion in logartithmically growing cells, we found red fluorescence associated with vacuoles indicative of the uptake of mitochondria into vacuoles. Addition of nigericin, a K+/H+ ionophore, to growing mdm38Δ cells depleted for Mdm38p essentially reverted the appearance of mitochondria as well as of vacuoles within a few hours. There were no longer any indications of ongoing mitophagy. Thus, autophagic degradation of mitochondria in Mdm38p-depleted cells clearly correlates with the loss of K+/H+ exchange activity.

For the first time our data show that a mutation-induced dysfunction of mitochondria, without any externally added drugs or physiological stresses, can trigger extensive mitophagy. In previous experiments, the onset of mitophagy had been attributed to the impairing of the mitochondrial electrochemical transmembrane potential.11 Mdm38p-depleted cells showed a reduction in Δψ, but two observations suggested that this was not sufficient to trigger fragmentation of the mitochondrial network and mitophagy. First, growth of mdm38Δ cells in the presence of oligomycin restored Δψ, but not the tubular reticulum of mitochondria (not shown). Second, mutants lacking Cox4 (cox4Δ) or having very low amounts of Cox2 only (dnmΔmdm38Δ, Supplementary Figure S1), show a similar reduction in Δψ as mdm38Δ cells and show no signs of mitophagy (data not shown). Accordingly, we conclude that osmotic swelling of Mdm38p-depleted cells is sufficient to massively induce mitophagy, provided that fragmentation of the mitochondrial reticulum is not inhibited.

Induction of mitophagy by Mdm38p depletion had no significant effect on cell viability under logarithmic growth conditions. Ultrastructure analysis of cells by electron microscopy revealed enhanced abundance of spherical mitochondria after prolonged (t50) depletion of Mdm38p. Presumably, in logarithmic growth, ongoing mitochondrial proliferation counterbalances degradation. In stationary phase, however, Mdm38p depleted cells had a significantly reduced survival rate as compared to wild-type cells, suggesting a synthetic effect of mitophagy and stationary phase induced autophagy. Thus, Mdm38p is required for enhanced survival of cells in stationary phase, and mitophagy in stationary phase is likely involved in a cell death program.

Overexpression of proteins of the fusion machinery (Mgm1p, Fzo1p), had no significant effect on mdm38Δ mitochondrial morphology, or mitophagy. By contrast, deletion of DNM1, a gene with a central role in mitochondrial fission,23 generated a mitochondrial phenotype combining the features of the dnm1Δ mutant and the mdm38Δ mutant. These mdm38Δdnm1Δ double mutant cells contained one or few individual giant mitochondrial spheres which we interpret as resulting from osmotic swelling of mitochondria which continue to fuse but lack fission activity. Interestingly, a block of fission by the dnm1Δ mutation prevented mitochondrial fragmentation and also mitophagy. Thus, it appears that induction of mitophagy in mdm38Δ cells requires the mitochondrial reticulum to be capable of undergoing fission. Our experimental data provide the first evidence in support of the idea that formation of discrete mitochondrial units allows the cell to rid itself of damaged parts of the mitochondrial network rather than the whole network.12 Support for this idea is also provided by the studies of Lemasters,22 concerning selective mitophagy in hepatocytes, where mitochondria are generally found as discrete spheres rather than as a reticulum.

Materials and Methods

Yeast strains, plasmids and media

The Saccharomyces cerevisiae strain W303 (ATCC no. 201239) served as wild-type strain. The null mutant mdm38Δ was constructed from W303 by replacement of MDM38 with HIS3 as described previously in Nowikovsky et al.6 To generate the mdm38Δdnm1Δ double deletion strain, mdm38Δ was crossed with the deletion strain dnm1Δ obtained from sporulation of W303 mgm1Δdnm1Δ generously provided by Sesaki et al.24

Culturing of yeast was performed according to standard protocols at 30°C on yeast media, including YPD (2% glucose), YPG 2% (glycerol), YPGal (2% galactose), YPRaf (2% raffinose), SRaf (2%Raf) and SGal (2% galactose), prepared as described previously,25 except for SGal liquid medium which was supplemented with 2 mM adenine. Yeast transformations were performed as described previously.26 For all experiments except the one shown in Figure 2, all strains (but not the cox2/3Δ mutants) were grown in galactose containing-media, since galactose is a fermentative carbon source that does not repress mitochondria like glucose. Dox was used when indicated at a concentration of 5 μg/ml. All experiments described in this study were carried out with cultures of cells kept in the logarithmic phase of growth unless otherwise stated. Exponential cultures were diluted as appropriate if grown over periods longer than overnight.

Mitochondrial DNA mutants PTY cox2Δ, PTY cox3Δ (G. Wiesenberger) fail to express the Cox2, or Cox3 protein, respectively and were grown on YPRaf. To express MDM38 under a regulated promoter, we constructed an mdm38Δ strain expressing Mdm38p from a dox repressible (tetO) promoter. For this purpose, MDM38 was amplified from YCp-MDM38-HA6 with the primers 5′-TAATATGGATCCATGTTGAATTTCGCATCAAGAGCG-3′ and 5′-ATATATGGCCCTAGCGGCCTCAGCACTGAGCAGCGTAATCTGG-3′ and the BamH1–Sfi1 fragment was inserted into the respective sites of the vector pCM189,27 yielding pCM189-MDM38-HA.

The biosensor construct pCS-G/RFP used to follow mitochondrial uptake into vacuoles was constructed by sequential addition of three DNA sequence cassettes into the yeast multi-copy expression vector pAS1NB,28 having PGK1 transcription control sequences and LEU2 selectable marker. The sequence cassettes encoded the following components: (1) the RFP component retrieved by PCR from pQE31-DsRed.T3;29 (2) the GFP component retrieved by PCR from pGEX-2 T carrying the ecliptic pHluorin;30 and (3) the first 55 amino acids of the yeast citrate synthase precursor polypeptide amplified by PCR from yeast genomic DNA.31

Visualization of mitochondria with mitochondrial matrix targeted GFP expressed from the constitutive TPI (triosephosphate isomerase) promoter was achieved by use of the plasmid pYX232-mtGFP.18

In vivo labeling of mitochondrial translation products

In vivo pulse labeling was performed according to Fox et al.32 Briefly, wild-type expressing the empty plasmid pCM189 and MDM38dox cells were pre-grown in SGal medium containing dox (5 μg/ml), wild type and mdm38Δ in YPGal. Nigericin (2 μM) was added as indicated. Control mit− strains (cox2Δ, cox3Δ) were grown in YPRaf. Cells taken from precultures were transferred to synthetic medium lacking methionine and incubated for 3 h in presence or absence of dox as indicated. After addition of cycloheximide (150 μg/ml) cells were labeled with 35S-methionine (8 μl/ml of a 5 mMCi stock, Amersham) and incubated for 30 min with agitation. Reactions were chased with unlabeled methionine (4 mM) for 10 and 120 min. Whole cell proteins were precipitated with trichloroacetic according to Westermann,33 separated by 16% SDS-PAGE and visualized by autoradiography.

Isolation of mitochondria

Yeast mitochondria from cultures grown exponentially in galactose-containing media were isolated as described in Nowikovsky et al.6

Western blotting

Yeast cells were grown in SGal in absence or presence of 5 μg/ml dox as indicated. Cultures were re-diluted as required to keep them in logarithmic growth phase. Total cell extracts were analyzed using 12.5% polyacrylamide gels unless indicated otherwise. Proteins were transferred to a nitrocellulose membrane and analyzed using primary antibodies to HA (1 : 200;34), Hexokinase-1 (1 : 15 000; Biotrend), GFP (1 : 2000; Roche), Mdm38p (1 : 200; provided by P. Rehling) in PBS-T plus 5% dry milk. Proteins were visualized by incubation with the mouse or rabbit secondary antibody conjugated to horseradish peroxidase (1 : 10 000; Molecular Probes, or 1 : 15 000; Sigma, respectively) in PBS-T plus 5% dry milk, followed by ECL (Pierce).

Measurement of K+/H+ exchange activities

Light scattering experiments were performed as previously described in Nowikovsky et al.6

Determination of mitochondrial Δψ

Isolated yeast mitochondria (0.5 mg protein) were incubated in the presence of 0.5 mM ATP, 0.2% succinate and 0.01% pyruvate with 0.5 μM JC-1 (5,5′,6,6′-tetrachloro-1,1′,3,3′-tetraethylbenzimidazolocarbocynine iodide, Molecular Probes) for 7 min at room temperature in the dark. The intensity changes of the monomeric form (low energy, 540 nm) and of the multimeric form (high energy, 590 nm) were recorded with the Scan mode of the Perkin Elmer LS55 luminescence photometer. Data collection was carried out with FL WinLab4.0. For calibration, aliquots of the same preparation were hyperpolarized with 1 μM nigericin (Sigma) and depolarized with 1 μM carbonylcyanide-p-trifluoromethoxyphenylhydrazone (FCCP, Sigma). A calibration curve was calculated for every single measurement and applied to correlate the Δψ expressed as a percentage of the maximum and the minimum obtained by hyperpolarization and depolarization, respectively.

Microscopy

pVT100U-mtGFP, pYX232-mtGFP,18 pCS-G/RFP (C Rosado, M Prescott, and RJ Devenish, unpublished) or 5 μM rhodamine B hexyl ester (Molecular Probes) served to label mitochondria. Vacuoles were stained with FM4-64 in a final concentration of 10 μM (Molecular Probes). Classification of the morphotype of cells in logarithmic growth phase was performed with hidden identity by standard fluorescence microscopy on Axioplan 2 with a 63 oil immersion objective (Zeiss Plan Apochromat). Confocal images were captured with a Zeiss Axiowert LSM 510 microscope. The excitation wavelength for GFP and FM4-64/dsRed was 488 and 543 nm, respectively. Images were acquired in multitrack Z stack and projected with LSM5 Image Browser.

Electron microscopy

Cells were harvested at early logarithmic growth phase (OD600=0.5). For cryofixation, cell pellets were introduced to flat sample holders of an EMPACT high-pressure freezer (LEICA Microssystems, Austria). With the help of a loading station, the flat specimen holders were placed in a sample holder pod and tightly sealed. The samples were frozen at a pressure of about 2000 bar, as described previously for specimens of mammalian tissues.35 Following freezing, the flat specimen holders were separated from the pods under liquid nitrogen and transferred to a LEICA AFS freeze substitution system (LEICA Microsystems, Austria). Freeze-substitution was as described elsewhere.36 In short, samples were substituted in acetone containing 2% OsO4 for 4 days at −90°C, followed by warming and embedding in epoxy resin (Agar 100). Thin sections were cut with an Ultracut S ultramicrotome (LEICA Microsystems, Austria), mounted on copper grids with Formvar support film, counterstained with uranyl acetate and lead citrate and examined at 80 kV in a JEOL JEM-1210 electron microscope. Images were acquired using a digital camera Morada for the wide-angle port of the TEM and analySIS FIVE software (Soft Image System).

Trypan blue staining

Cells were collected and stained with Trypan blue (Gibco BRL) for 15 min at room temperature. Unstained cells (alive) and blue cells (dead) were visualized and counted under light microscopy.