Adenine nucleotide translocase (Ant) is primarily involved in ATP/ADP exchange across the mitochondrial inner membrane. Recently, the A114P missense mutation in the human Ant1 protein was found to be associated with autosomal dominant progressive external ophthalmoplegia (adPEO). Ant1A114P was proposed to cause an imbalance of the mitochondrial deoxynucleotide pool that subsequently affects the accuracy of mtDNA replication, thereby leading to accumulation of mutant mtDNA. In the present study, it has been shown that the A128P mutation of the Saccharomyces cerevisiae Aac2 protein, equivalent to A114P in human Ant1p, does not always affect respiratory growth. However, expression of aac2 A128P results in depolarization, structural swelling and disintegration of mitochondria, and ultimately an arrest of cell growth in a dominant-negative manner. The aac2 A128P mutation likely induces an unregulated channel allowing free passage of solutes across the inner membrane. These data raise the possibility that the formation of an unregulated channel, rather than a defect in ATP/ADP exchange, is a direct pathogenic factor in human adPEO. The accumulation of mtDNA mutations might be a consequence of mitochondrial dysfunction.
Adenine nucleotide translocase (Ant) is primarily involved in ATP/ADP exchange across the mitochondrial inner membrane (1–3). Under respiring conditions, ATP produced within mitochondria is exported through Ant to support cellular activities. As exchange, ADP is imported to fuel the conversion of ADP to ATP by ATP synthase. Ant is the most abundant protein in the mitochondrial inner membrane. It belongs to the mitochondrial carrier family that supports a variety of transport activities across the mitochondrial inner membrane (3,4). A typical Ant comprises about 300–320 residues that form six transmembrane helices. The functional unit is likely to be a homodimer acting as a gated pore that channels only one molecule of ADP and ATP (5).
Ant has also been proposed to participate in the formation of the mitochondrial permeability transition pore (PTP), a high-conductance channel of low selectivity that occurs on mitochondrial membranes (6–9). The opening of PTP allows the free passage of solutes with a molecular mass below 1.5 kDa, thereby leading to membrane depolarization, substrate depletion, equilibration of solutes across the inner membrane, swelling of mitochondrial matrix and rupture of the outer membrane. Although it is not known whether PTP operates in cells under physiological conditions, activation of PTP is believed to be involved in many forms of cell injury such as ischemia–reperfusion and in programmed cell death through its action in permeabilizing the mitochondrial membranes (10–12). Based on the regulation of PTP by Ant inhibitors such as bongkrakate and atractylate, it has been proposed that Ant is the core pore-forming element of PTP (13) (reviewed in 9). This proposal was supported by the observation that, in reconstituted lipid bilayers, purified Ant forms a large Ca2+-dependent channel resembling PTP (14,15). However, the contribution of Ant to PTP remained a matter of active debate because of the lack of direct in vivo evidence to support Ant's pore formation ability. Also, the complexity of PTP modulation in isolated mitochondria does not tend to support the Ant-based single channel model (reviewed in 9,16).
Recently, mutations in human adenine nucleotide translocase, Ant1p, were found to be associated with autosomal dominant progressive external ophthalmoplegia (adPEO) (17). adPEO is an adult-onset mitochondrial disorder characterized by mildly reduced activities of the respiratory-chain enzymes and multiple deletions of the mitochondrial genome in skeletal muscle. How mutations in the ANT1 gene lead to mitochondrial genome disintegration remains unclear. However, it has been proposed that mutant Ant may have a reduced activity in ATP/ADP translocation across the mitochondrial inner membrane that subsequently causes an imbalance of the mitochondrial deoxynucleotide pool. Consequently, it would affect the accuracy of mtDNA replication, thereby leading to accumulation of mutant mitochondrial DNA (mtDNA).
The present study used the genetically amenable yeast system to understand the pathogenic mechanism of adPEO caused by mutations in ANT1. It was found that the A128P mutation in the yeast Aac2 protein, equivalent to the A114P mutation in human Ant1p identified in adPEO patients, did not always affect respiratory growth. Genetic and cell biological data suggested that the aac2A128P mutation induces an uncontroled channel on the mitochondrial inner membrane. It was proposed that the opening of the ant1A114P-induced unregulated channel, rather than a defect in ATP/ADP translocation, could be the primary pathogenic factor in human adPEO. Accumulation of mtDNA mutations may be a consequence of mitochondrial damage following membrane permeabilization.
The yeast aac2A128P allele is respiratory-competent but affects cell viability in a dominant-negative manner
Saccharomyces cerevisiae has three homologous genes encoding Ant, namely AAC1, AAC2 and AAC3. In respiring cells, AAC2 encodes the bulk of the ATP/ADP translocase (18,19) (reviewed in 20). To mimic the A114P mutation in human Ant1p responsible for adPEO, the evolutionarily conserved Ala128 codon of AAC2 was changed by in vitro mutagenesis to that coding for proline. The constructed aac2A128P allele was integrated into the ura3 locus of yeast strains to replace wild-type AAC2. In several laboratory strains examined, replacement of AAC2 with aac2A128P did not result in a respiratory-deficient phenotype. As exemplified in Figure 1A, a haploid strain containing the aac2A128P allele in place of the wild type does not show significant growth defect in comparison with a wild-type strain on medium containing glycerol as a carbon source (Fig. 1A). However, in some laboratory strains, a decrease in respiratory growth on glycerol medium was noticeable (not illustrated), as previously reported (17). As can be seen in Figure 1A, an AAC2/aac2A128P heteroallelic haploid strain also displays normal growth on glycerol.
Despite a slight variation in their respiratory growth, strains carrying solely aac2A128P or at a heteroallelic state form sectoring colonies on complete glucose medium in comparison with an isogenic AAC2 strain (Fig. 1B). The colony-sectoring phenotype suggested that a subfraction of cells in the population are unable to continue cell division. Indeed, when single cells from the aac2A128P and AAC2/aac2A128P heteroallelic haploid strains were transferred to fresh complete glucose medium by microdissection, 22–66% of the cells formed non-viable microcolonies of variable sizes, as exemplified in Figure 1C. Microscopic inspection revealed that the microcolonies contain 4–4000 cells, indicating that these cells can undergo 2–12 cell divisions before ceasing to grow.
The aac2A128P-induced cell death phenotype appeared to occur in a typical dominant-negative manner. The diploid strain CS439 heterozygous for AAC2/aac2A128P formed non-viable colonies on glucose medium at a rate of 6.7%, compared with 0% in the control wild-type strain CS5. A128P is therefore a gain-of-function mutation for the induction of cell death in the population.
Mitochondrial depolarization by aac2A128P
To gain insight into structural and functional changes to mitochondria associated with the expression of aac2A128P, the mutant allele was placed under the control of the galactose-inducible GAL10 promoter. As shown in Figure 2A, a wild-type haploid expressing the GAL10–aac2A128P fusion failed to form visible colonies on plates containing raffinose and galactose, indicating that an elevated expression of aac2A128P from the GAL10 pro-moter is a lethal event. In liquid raffinose plus galactose medium, these cells were able to undergo up to five divisions before ceasing to divide. The cell growth arrest was found to be partially reversible. After propagation for four generations in galactose medium, 60% of the cells were capable of forming colonies when returned to the repressing glucose medium (Fig. 2A). Among the surviving cells, a significantly increased formation of respiratory-deficient petite mutants was detected, with up to 20% of viable cells being ρ−. This observation is reminiscent of the accumulation of mtDNA mutations in human cells expressing ANT1A114P (17). As revealed by Southern-blot analysis, the overall mtDNA level in yeast is not conspicuously changed following the induction of GAL10–aac2A128P (data not shown).
The detrimental effect on cell growth suggested that expression of the aac2A128P allele interferes with a vital mitochondrial function other than ATP/ADP transport. Flow-cytometric analysis of GAL10–aac2A128P-bearing cells five divisions after galactose induction revealed that the mitochondrial membrane potential (Δψ)-dependent uptake of the fluorescent dye DiOC6 was dramatically decreased (Fig. 2C). The mean fluorescence intensity in these cells was only 2.19, compared with 57.98 in an isogenic strain expressing the wild-type AAC2 from the GAL10 promoter. Interestingly, DiOC6 accumulation in the aac2A128P-overexpressing cells was found to be even lower than that in the wild type incubated with the chemical uncoupler CCCP, which has a mean fluorescence intensity of 5.74. Confocal microscopy confirmed that most aac2A128P-overexpressing cells can no longer accumulate DiOC6 (not illustrated). A small proportion of the cells retained the ability to take up the dye as a result of an excision of the chromosomally integrated GAL10–aac2A128P cassette (data not shown). A 26.5-fold decrease in dye accumulation strongly suggests that overexpression of aac2A128P depolarizes mitochondria, although a fraction (5.4-fold) of DiOC6 signal reduction may be attributed to loss of mitochondrial structures (see below). It is interesting to note that the aac2A128P-overexpressing cells have a lower Δψ than the wild type treated with CCCP. The in vivo depolarizing activity associated with Aac2A128P seems to be more potent than with CCCP.
Structural swelling and disintegration of mitochondria induced by the overexpression of aac2A128P
When the expression of GAL10–aac2A128P was induced in galactose-plus-raffinose medium for only 1.5 generations before the loading of DiOC6, a release of the dye into non-mitochondrial compartments was observed, in contrast to an exclusive accumulation in the wild-type mitochondrial structures (Fig. 3A). Also noticeable was the presence of spherical and swollen mitochondria. Overexpression of aac2A128P therefore induces membrane depolarization accompanied by mitochondrial swelling.
The effect of long-period aac2A128P overexpression on mitochondrial structure was examined by electron microscopy of cells induced for five generations in galactose-plus-raffinose medium. Sixty-eight percent of cell sections were found to lack mitochondrial profiles. The average number of mitochondrial profiles per section was only 0.67, compared with a value of 3.6 in the control (Fig. 3B). One-third of the mitochondria-containing sections from the aac2A128P-overexpressing cells displayed grossly swollen mitochondria with few visible cristae (Fig. 3C). From these data, it can be concluded that overexpression of aac2A128P leads to matrix swelling followed by structural disintegration of mitochondria.
An unregulated channel induced by the A128P mutation
Based on the mitochondrial-depolarizing property, the gain-of-function nature and the structural changes to mitochondria associated with the expression of aac2A128P, it can be proposed that aac2A128P induces an unregulated channel on the mitochondrial inner membrane. Opening of the channel allows the free passage of protons and other solutes, as manifested by the collapse of the mitochondrial membrane potential and the swelling of the mitochondrial matrix.
Further genetic studies suggested that the aac2A128P-induced pore might result from deregulation of a pre-existing channel. As described above, a haploid strain heteroallelic for aac2A128P/AAC2 is viable with a subfraction of cells in the population failing to continue cell division. However, although the presence of two tandemly integrated copies of aac2A128P, driven by the native AAC2 promoter, did not lead to a lethal phenotype in the diploid strain CS509/20 (AAC2/AAC2, ura3::aac2A128P::aac2A128P/ura3), meiotic progeny of CS509/20 receiving the two copies of aac2A128P segregated into non-viable spores (Fig. 4A). Furthermore, introduction of a plasmid carrying the wild-type AAC2 into CS509/20 failed to rescue the non-viable meiotic spores. From these observations, it can therefore be concluded that expression of two tandem copies of aac2A128P can be tolerated by diploid but not by haploid cells. An elevated gene dosage for the wild-type AAC2 could not help cells to escape the aac2A128P-induced lethality. As summarized in Figure 4B, the lethal phenotype correlates with the aac2A128P/ ploidy (n) ratio rather than with the relative dosage of the mutant allele versus the wild-type AAC2. Clearly, cells cannot survive when aac2A128P/n is 2 or more, irrespective of the presence of excessive wild-type Aac2p. It appears that, in cells with an aac2A128P/n ratio of 1 or less, the aac2A128P-induced channel remains regulated by a hypothetical inhibitory factor. When the ratio is 2 or more, the inhibitory factor is titrated out and the excessive Aac2pA128P induces an uncontrolled pore capable of depolarizing the inner membrane and causing structural swelling of mitochondria.
Functional implications of aac2 A128P expression
The opening of an unregulated channel on the mitochondrial inner membrane would be expected to have detrimental consequences for mitochondrial biogenesis. Indeed, it was found that the mitochondrial targeting of nuclear-encoded proteins is severely affected in cells overexpressing aac2A128P. When the targeting of the nuclear-encoded and green fluorescent protein (GFP)-labeled mitochondrial matrix protein, Mcx1p (21), was monitored by confocal microscopy of cells co-expressing MCX1–GFP and GAL10–aac2A128P, green fluorescence was no longer restricted to a recognizable reticulum structure as seen in the wild-type control upon induction of aac2A128P in galactose (Fig. 5A). Instead, GFP remained in the cytoplasm or/and was released from mitochondria following structural swelling. The failure of the targeting of Mcx1–GFP into mitochondria could result from the collapse of Δψ, a global loss of mitochondrial ATP synthesis and/or a loss of mitochondrial membrane integrity.
Effect of aac2A128P overexpression on mitochondrial biogenesis was also evaluated by monitoring the export/proteolytic processing of the mtDNA-encoded Cox2 protein (cytochrome c oxidase subunit II). Processing of the Cox2p precursor ( pCox2) into its mature form (mCox2) occurs on the external surface of the inner membrane, after the insertion of pCox2 into the membrane from the matrix side in a Δψ-dependent step (22). Western-blot analysis showed that, upon induction of aac2A128P expression from the GAL10 promoter, Cox2p was detected only in its precursor form ( pCox2). The steady-state levels of Cox2p and other mitochondrial proteins such as porin and the β-subunit of F1-ATPase (F1-β) were not significantly altered (Fig. 5B). The failure in pCox2 processing reflects a defect in protein export and/or a loss of the cleavage activity following overexpression of aac2A128P.
This study has described a phenomenon of significant biological and medical interest. The expression of a mutant ATP/ADP translocase in yeast was found to sequentially induce structural and functional changes of mitochondria. These changes include membrane depolarization, structural swelling and, in a more advanced stage, defects in the import and export/processing of mitochondrial proteins, disintegration of the organelle and ultimately a total arrest of cell growth. Since a haploid carrying solely the aac2A128P allele is respiratory-competent, the mutant Aac2A128P protein should have been correctly targeted into the mitochondrial inner membrane and remained functionally active. The detrimental effect of aac2A128P must have been caused by gain of a novel function associated with Aac2A128P. Together with the dominant-negative nature of Aac2A128P, all the available data support the proposal that the A128P mutation in AAC2 induces an unselective channel on the inner membrane. The opening of such a channel allows free passage of protons and possibly other solutes. When the accumulation of the opened channel reaches a critical threshold, membrane depolarization and structural swelling take place, which subsequently affect mitochondrial biogenesis and cause loss of mitochondrial structures and cell viability.
It is unclear whether Aac2A128P participates directly or indirectly in the formation of the proposed unselective channel. Adenine nucleotide translocase has long been suspected to be the pore-forming element of the mitochondrial permeability transition pore (PTP). In vitro reconstituted translocase is able to form a poorly selective channel, as determined by electrophysiological studies (14). The presence of PTP activity on yeast mitochondrial membranes has been well documented by investigating the permeability of isolated mitochondria in vitro (23–27). Yeast PTP is cyclosporin-insensitive and Ca2+-independent, and is induced by respiratory substrates and ATP (26). Because mitochondrial depolarization and matrix swelling following aac2A128P expression are the events expected to occur in the case of PTP opening, the possibility was raised that the aac2A128P allele may sensitize a PTP component to endogenous pore inducers. To test this possibility, PTP activity was measured by monitoring the swelling of mitochondria isolated from cells expressing solely aac2A128P in mannitol/sucrose-based medium. In response to ethanol and ATP, no acceleration of mitochondrial swelling was found in the mutant mitochondria in comparison with those from wild-type cells (data not shown). These results imply either that ethanol and ATP are not among the endogenous PTP-inducers or that the aac2A128P-induced channel is unrelated to PTP. Early investigations have shown that disruption of the Ant-encoding genes in yeast does not abolish PTP as measured by patch-clamp techniques. Ant inhibitors such as carboxyatractyloside have no effect on PTP activity (24,26). Likely, PTP activity either is independent of Ant or is composed of multiple pathways with Ant being one of the contributing components. In view of these complications, caution needs to be taken when superposing the aac2A128P-induced channel on an unregulated PTP.
Ant is intrinsically a channel-type transporter for the translocation of adenine nucleotides. So one can propose that the A128P mutation may simply converts the adenine nucleotide channel into an unselective pore allowing the passage of solutes of anionic nature. Also, A128P could alter the sensitivity to endogenous inducers for the proton transport activity that is associated with Ant. For instance, Ant has been reported to be involved in fatty acid-mediated proton transport across the mitochondrial inner membrane (28,29) (reviewed in 30). Fatty acids can cross the membrane in undissociated form and return through the membrane via Ant as an anion, thereby causing mitochondrial depolarization. In yeast, it has been shown that fatty acid-induced uncoupling requires an intact Ant (31). Therefore, it is conceivable that the A128P mutation in the yeast Aac2 protein may sensitize mitochondria to fatty acid-mediated depolarization.
The opening of an uncontroled pore in aac2A128P-expressing cells would imply that the aac2A128P-induced channel is subject to tight regulation. In line with the above-mentioned hypotheses that support direct participation of Aac2A128P in channel formation, the A128P mutation may affect interactions with proteinaceous or non-proteinaceous regulatory factors. Structurally, the mutated Ala128 residue in Aac2p, or its equivalent in the human Ant1p, Ala114, is predicted to be located either in the third transmembrane domain or in the loop joining the second and third transmembrane domains in the intermembrane space (32,33). Interestingly, this location coincides with the binding site of the HIV-1 derived Vpr peptide (34). The binding of Vpr has been shown to specifically modulate the opening of PTP in the mammalian system. Also, this subdomain of human Ant1p is essential for its apoptotic effect on overexpression (35). Furthermore, the equivalent region in the human Ant2 isoenzyme appears to interact directly with the proapoptotic protein Bax (36). Therefore, it seems that the sequence flanking the alanine residue plays an important role in interacting with pore regulators in both yeast and humans. Finally, when drawing possible analogies between the mammalian and yeast Ants, it may also be relevant to mention that overexpression of the mammalian ANT1 gene can dominantly induce apoptosis – supposedly via the sequestration of potential PTP repressors (35). ANT1-induced cell death appears to be suppressed by overexpression of the gene encoding cyclophilin D, a component of PTP that is believed to directly interact with Ant1 from the matrix side. In yeast, when the CPR3 gene, encoding the mitochondrial isoform of cyclophilin, was overexpressed from a multicopy vector, aac2A128P-induced lethality appeared to be poorly suppressed in the CS490/3 background (data not shown; see Materials and Methods for details). Whether the Ant–cyclophilin interaction is conserved in yeast is being investigated.
Finally, as an alternative, the aac2A128P allele could also induce an unselective channel in an indirect manner. For instance, Aac2A128P could induce the opening of a membrane channel defined by other proteins, either by changing the physiological conditions within mitochondria, or simply by interfering with the general import pathway for the mitochondrial carrier proteins in the inner membrane (37).
In summary, the data presented in this study could provide important insight into the pathogenesis of human adPEO caused by mutations in Ant1p. The dominant-negative feature, commonly shared by the A114P mutation in humans and A128P in yeast, suggests a similar pathogenic mechanism in the two systems. Like the mutant protein in yeast, the human ant1A114P allele could depolarize mitochondria by inducing an unregulated channel on the inner membrane. The most immediate consequence would be the collapse of the cell's energy transduction system and, possibly, the activation of mitochondria-dependent cell death. The accumulation of mutant mtDNA, as observed in both yeast and humans, could result from loss of mtDNA replication precursors (e.g. nucleotides) following channel opening. Therefore, from a therapeutic perspective, efforts may need to be focused on blocking the ant1A114P-induced channel rather than on re-establishing the nucleotide translocation process.
MATERIALS AND METHODS
Construction of plasmids and yeast strains
The Saccharomyces cerevisiae AAC2 gene was mutagenized using the QuickChange site-directed mutagenesis kit (Stratagene) to change the codon Ala128 from GCC to CCC coding for proline. The open reading frames of the wild-type AAC2 and the aac2A128P allele were amplified by PCR and placed under the control of the GAL10 promoter in the URA3-based integrative vector pUC–URA3/4 to generate pURA–GAL–AAC2 and pURA–GAL–aac2A128P. Insertion of the GAL10–aac2A128P allele into the centromeric vector pCXJ24 (38) gave rise to pCXJ24–GAL–aac2A128P/5. Cloning of the aac2A128P allele under the control of the native AAC2 promoter into pUC–URA3/4 resulted in pURA–aac2A128P/1. To target GFP into mitochondria, the coding sequence of MCX1, a nuclear gene encoding a putative mitochondrial chaperone (21), was fused on its C terminus to the S65T variant of GFP. MCX1–GFP was cloned into the yeast vector pCXJ15 (38) to generate pScCLPX–GFP/3. The AAC2- and aac2A128P-overexpressing strains CS488/1 and CS490/3 were constructed by integration of pURA–GAL–AAC2 and pURA–GAL–aac2A128P into the ura3 locus of M2915-6A (MATa, ade2, leu2, ura3) after linearization of the plasmids by StuI in URA3. The diploid strain used was derived from CS5 (MATa/MATα, ade2/+, +/his4, ura3/ura3, leu2/leu2). CS439 was constructed by disruption of one copy of AAC2 and integration of pURA–aac2A128P/1 into the ura3 locus of CS5. A strain with two tandemly integrated copies of pURA–aac2A128P/1 was designated CS509/20. To express MCX1–GFP, M2915-6A was transformed with pScCLPX–GFP/3 to generate CS251/1 or co-transformed with pScCLPX–GFP/3 (URA3) and pCXJ24–GAL–aac2A128P/5 (LEU2) to produce CS521.
The mitochondrial cyclophilin gene CPR3 was amplified from yeast genomic DNA using the Expand High Fidelity PCR system (Roche). The PCR product, encompassing 561 and 198 nucleotides upstream and downstream of the CPR3 coding region, respectively (39), was cloned into the 2 µm-based yeast multicopy vector YEp13M4 (LEU2). To test whether overexpression of CPR3 rescues aac2A128P-induced lethality, the YEp13–CPR3 constructs were introduced into CS490/3 (MATa, ade2, leu2, ura3::pURA–GAL–aac2A128P) by selecting for Leu+ transformants. The resulting transformants were grown in liquid selective medium, diluted in sterile water and spotted onto galactose complete medium for inducing the expression of aac2A128P (Fig. 2A). Transformants containing only the vector YEp13M4 were used as a control. The formation of colonies on galactose medium by the YEp13–CPR3 transformants indicates that overexpression of CPR3 can rescue aac2A128P-induced lethality.
Flow cytometry, microscopy and permeability transition monitoring
For flow-cytometric analysis, CS488/1 and CS490/3 were grown in complete glucose medium to late exponential phase, washed with water and transferred to induction medium containing 2% raffinose and 2% galactose. After incubation for five cell divisions at 30°C, cells were loaded with the mitochondrial membrane potential-sensitive fluorescent probe DiOC6 (3,3′-dihexyloxacarbocyanine iodine, Molecular Probes) at 0.1 µm (40), and intracellular accumulation of the dye was measured with a Becton Dickinson FACscan flow cytometer. For confocal microscopy, a Leica TCS-SP2-UV microscope was used to detect DiOC6 and GFP using an excitation wavelength of 488 nm. For electron microscopy, CS488/1 and CS490/3 cells were grown in raffinose-plus-galactose medium for approximately five generations, prefixed with 3% of glutaraldehyde and fixed with 2% potassium permanganate before being dehydrated, embedded and sectioned for examination with a JEOL 1200EX electron microscope. Mitochondrial isolation and permeability transition monitoring were carried out as previously described (26).
The author thanks G.D. Clark-Walker and D.R. Pfeiffer for critical comments, A. Janssen, X.-M. Zuo, L.-J. Ouyang and H. Liszczynsky for technical assistance, S. Grüninger for advice with FACS analysis, D. Webb for help with confocal microscopy, C.X. Huang, L. Shen and S. Stowe for electron microscopy, and J. Velours for providing the antibody against F1-β.
To whom correspondence should be addressed at: School of Biological Sciences, Nanyang Technological University, 1 Nanyang Walk, Block 5 Level 3, Singapore 637616, Republic of Singapore. Fax: +65 67953261; Email: firstname.lastname@example.org
Present address: School of Biological Sciences, Nanyang Technological University, 1 Nanyang Walk, Block 5 Level 3, Singapore 637616, Republic of Singapore.