Human Molecular Genetics Advance Access originally published online on September 12, 2006
Human Molecular Genetics 2006 15(20):3063-3081; doi:10.1093/hmg/ddl248
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Cytotoxicity of a mutant huntingtin fragment in yeast involves early alterations in mitochondrial OXPHOS complexes II and III
1 Department of Neurology, 2 Department of Biochemistry and Molecular Biology and 3 Neuroscience Program, The Dr John T. Macdonald Foundation Center for Medical Genetics, University of Miami, Miller School of Medicine, Miami, FL, USA
* To whom correspondence should be addressed at: Department of Neurology and Department of Biochemistry and Molecular Biology, The Dr John T. Macdonald Center for Medical Genetics, Universtiy of Miami, Miller School of Medicine, 1600 NW 10th Avenue, RMSB 2067, Miami, FL 33136, USA. Tel: +1 3052438683; Fax: +1 3052433914; Email: abarrientos{at}med.miami.edu
Received May 26, 2006; Accepted September 5, 2006
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Mitochondrial dysfunction may play an important role in the pathogenic mechanism of Huntington's disease (HD). However, the exact mechanism by which mutated huntingtin could cause bioenergetic dysfunction is still unknown. We have constructed a stable inducible yeast model of HD by expressing a human huntingtin fragment containing a mutant polyglutamine tract of 103Q fused to green fluorescent protein (GFP), and a control expressing a wild-type 25Q domain fused to GFP in a wild-type strain. We showed that in yeast cells expressing 103Q, cell respiration was progressively reduced after 4–6 h of induction with galactose, down to 50% of the control after 10 h of induction. The cell respiration defect results from an alteration in the function and amount of mitochondrial respiratory chain complex II+III, in congruency to data obtained from postmortem brain of HD patients and from toxin models. In our model, the production of reactive oxygen species (ROS) is significantly enhanced in cells expressing 103Q. Quenching of ROS with resveratrol partially prevents the cell respiration defect. Mitochondrial morphology and distribution were also altered in cells expressing 103Q, probably resulting from the interaction of aggregates with portions of the mitochondrial web and from a progressive disruption of the actin cytoskeleton. We propose a mechanism for mitochondrial dysfunction in our yeast model of HD in which the interactions of misfolded/aggregated polyglutamine domains with the mitochondrial and actin networks lead to disturbances in mitochondrial distribution and function and to increase in ROS production. Oxidative damage could preferentially affect the stability and function of enzymes containing iron–sulfur clusters such as complexes II and III. Our yeast model represents a very useful paradigm to study mitochondrial physiology alterations in the pathogenic mechanism of HD.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Mitochondrial dysfunction and oxidative damage have been proposed to play a critical role in both aging and neurodegenerative diseases, including Huntington's disease (HD). HD is a fatal, inheritance-based late-onset neurodegenerative disorder mainly characterized by a selective loss of neurons from the striatum and deep layers of the cerebral cortex (1). It is caused by the expansion of a CAG triplet-repeat domain in the disease gene, leading to an extended polyglutamine (PolyQ) domain (from 6–35 to more than 36 glutamines) in the N-terminal portion of huntingtin (htt), the expressed protein. Htt is a protein restricted to vertebrates (2), and its role is likely at a number of cellular levels (reviewed in 3–5). Neuronal degeneration in HD is probably the result of the combination of a loss of function of the wild-type protein along with gain-of-function mutation. Mutant htt polypeptides acquire an unusual conformation, which produces cell toxicity, and facilitate their aggregation into intracellular inclusion bodies (6–8). Whether non-aggregated abnormal proteins or aggregates cause toxicity and neurodegeneration is the focus of ongoing discussions in the field (4,9–15). Mutant htt is cleaved both in vivo and in vitro (16,17) to form N-terminal fragments containing PolyQ repeats. These fragments may be the toxic form of the protein and are believed to contribute to disease pathogenesis in many PolyQ diseases. Although the exact identity of these cleavage fragments is not known, they seem to include approximately the first 150 residues of htt. Taking this into account, many laboratories have modeled HD pathogenesis with exon 1 fragments, which cause toxicity and promote protein aggregation in vivo and in cell models (reviewed in 4,5).
Evidence of mitochondrial dysfunction associated to the pathogenesis of HD has been accumulated over the last 30 years. Indirect evidence from several sources indicates that a defect of energy metabolism and consequent excitotoxicity could be involved in HD (18–20). Toxin models of HD are induced by 3-nitropropionic acid or malonate, both inhibitors of succinate dehydrogenase, complex II of the mitochondrial respiratory chain (21–24). In HD patients, postmortem caudate and putamen have severe deficiencies in mitochondrial respiratory chain complexes II and III (18,25–29). Expression of two subunits of complex II are preferentially decreased in the striatum of HD patients compared with controls, affecting the dehydrogenase activity of the complex (30). Similar results were reported in cultured striatal neurons expressing the pathological length of 82Q. In this model, the overexpression of either Ip or Fp subunit restored complex II levels and blocked striatal cell death induced by 82Q, suggesting that complex II defects in HD may be instrumental in striatal neuronal death (30). Other reports have suggested that the mutant form of htt damages neurons by directly interfering with mitochondrial function. Early mitochondrial calcium homeostasis defects have been detected in HD patients and transgenic mice, as a direct effect of polyglutamines (31,32). By electron microscopy, Panov et al. identified N-terminal mutant htt on neuronal mitochondrial membranes (31). Recent data showed that mutant PolyQ domains implicate a dominant property of htt in mitochondrial energy metabolism (33) and that mitochondrial respiration and ATP production are significantly impaired in striatal cells expressing mutant htt (34). All these data support a model of HD pathogenesis involving alterations in aerobic energy production in particular and mitochondrial physiology in general.
Despite evidence in support of mitochondrial involvement in the pathogenesis of HD, the exact mechanism by which mutated htt could cause bioenergetic dysfunction is still unknown. One hypothesis sustains that expression of full-length mutant htt impairs vesicular and mitochondrial trafficking in mammalian neurons in vitro and in vivo, probably by disrupting the microtubule and actin cytoskeletal network (35). In transgenic mouse models, these defects occurred early in development prior to the onset of measurable neurological or mitochondrial abnormalities (35), suggesting a mechanism of neuronal dysfunction in HD in which misfolding/aggregation of the mutant PolyQ domain of htt directly or indirectly leads to disruption of the cytoskeleton, eventually causing alterations in mitochondrial morphology and distribution, and mitochondrial dysfunction and neuronal death.
Cell culture models of HD have proven to be very useful for the understanding of HD pathogenesis (36–39). Highly significant for the work presented in this paper is the creation and use of yeast models of HD. The unicellular yeast Saccharomyces cerevisiae has provided a useful tool for the dissection of individual pathophysiological pathways (40), screening genes involved in the formation of protein aggregates (41,42) and potential PolyQ-induced toxicity (43). Over the last few years, several groups have shown that normal PolyQ polypeptides are soluble in yeast, whereas long PolyQ polypeptides form aggregates (41,44,45). Recently, a new yeast model of PolyQ expansion diseases has been created, suggesting a direct link between PolyQ aggregation and toxicity. In this model, htt toxicity depended on PolyQ aggregation mediated by the prion-like protein Rnq1 (13). This is an useful model to study the individual pathways that contribute to the observed cytotoxic effect. Particularly, S. cerevisiae is a facultative anaerobic yeast, which allows us to use it to study mitochondrial biogenesis and the molecular basis of human disorders stemming from impaired mitochondrial metabolism (reviewed in 46). In yeast cells, ATP is produced through two mechanisms. When glucose is present, glycolysis is activated to make ATP, whereas gluconeogenesis and mitochondrial respiration are repressed. When fermentable carbon sources are not available, the cell has to resort to oxidative phosphorylation (OXPHOS) for the production of ATP. As a result, mutations affecting OXPHOS components are not lethal and the levels of expression of these components can be manipulated simply by changes in culture conditions, allowing for a convenient selection of respiratory defective mutants. All these properties make the yeast models of polyglutamine diseases, including HD, especially appropriate for the study of alterations in mitochondrial function involved in the pathogenesis of these disorders. In addition, since there are no yeast homologs of proteins such as htt, the use of yeast models has allowed us to study the effects on mitochondrial function caused by the misfolding/aggregation of mutant PolyQ domains with independence of the wild-type htt loss-of-function consequences.
Here, we used a yeast model of HD expressing an NH2-terminal fragment of htt with a mutant 103Q domain fused to green fluorescent protein (GFP) from an integrative plasmid under the control of a Gal1 promoter to study the role of mitochondrial function on the cytotoxic effect exerted by the misfolding/aggregation of mutant PolyQ domains. Cells expressing a non-pathogenic 25Q domain were used as controls. We show that early after induction of 103Q expression, cell respiration is impaired as the result of an alteration in the function of mitochondrial respiratory chain complex II+III activity. Mitochondrial morphology and distribution were also altered in cells expressing 103Q, probably resulting from the interaction of aggregates with portions of the mitochondrial web and from alterations in the actin cytoskeleton. Possible mechanisms for mitochondrial dysfunction in our yeast model of HD are discussed.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Establishing a yeast model of HD using integrative plasmids
Given the complexity of the pathogenic mechanism underlying HD, researchers have developed yeast, worm, fly and mouse models to answer specific questions about protein aggregation and toxicity. A direct link between aggregation of expanded PolyQ domain and its cytotoxicity has been recently reported in a yeast model (13). We have now established a refined model with the goal of using it to investigate the role of mitochondrial function in the cytotoxicity and PolyQ aggregation observed in HD. Two PolyQ constructs consisting of exon1 sequences containing the first 17 amino acids followed by 25 or 103 glutamines and fused in frame with a GFP tag at the C-terminus of each construct and a FLAG tag attached to their NH2 termini (resulting in 25Q or 103Q) were obtained from Professor Michael Y. Sherman (Boston University, Boston, MA) (13,19) and subcloned into the episomal plasmid YEp351 (47). The wild-type strain W303-1A carrying the prion-like protein Rnq1 was transformed with plasmids expressing either the wild-type 25Q domain or the expanded 103Q domain. In order to maintain the plasmids, the transformed cells needed to be cultured in synthetic media with selection for plasmid retention. However, with long incubations, a high proportion of the cells carrying the 103Q constructs lost the plasmid, probably as a defense mechanism against the toxic effect of the expressed protein (13). To avoid this problem, we subcloned the 25Q-GFP and 103Q-GFP constructs into the integrative plasmid YIp351, modified to express the proteins under the control of the Gal1 promoter. The plasmids were linearized and integrated into the LEU-2 gene in the W303-1A strain. As expected, in the presence of 2% galactose, 100% of the cells in these cultures expressed the corresponding PolyQ domain.
The model was further validated by assessing cytotoxicity and protein aggregation, upon induction of expression of the constructs. Toxicity was assayed with growth tests either in solid or liquid media containing galactose to induce the expression of the PolyQ domains. In contrast to 25Q, expression of 103Q in single copy was toxic to yeast wild-type cells, as colonies expressing 103Q ceased to grow within 1–2 days after induction on galactose medium (Fig. 1A and B). The defective growth was already manifested after 6 h of induction (Fig. 1B). Since the Gal1 promoter is repressed in the presence of glucose in yeast, each of our experimental cultures was pre-grown in media containing glycerol and ethanol before induction with galactose. No difference in colony size was observed on glucose-containing medium without PolyQ expression, indicating that accumulation of 103Q was responsible for the growth defect. Aggregation of the PolyQ-GFP domains expressed from chromosomally integrated plasmids was followed under a confocal laser scanning microscope, following GFP fluorescence. After 20 h of induction in the presence of galactose, the 25Q domain appeared diffuse in the cytoplasm, whereas we observed amorphous, mainly cytoplasmic aggregates in cells expressing the 103Q domain (Fig. 1B). Cells expressing the 103Q construct were of slightly smaller size, a characteristic of yeast cells with impaired mitochondrial function. The process of aggregation started after 2 h of induction of expression with the formation of small spherical inclusion bodies in more than 50% of the cell population which probably acted as the seed for the formation of bigger amorphous aggregates as previously reported (data not shown; 13). Subsequently, a modified western blot assay (43) was developed (Fig. 1C and D) to quantify the level of expression of the PolyQ domains and to follow the kinetics of aggregation of the extended domains. Following a time course, we observed increased levels of accumulation of both mutant and wild-type proteins. SDS-insoluble aggregates of the 103Q domain were detected after 4 h of induction. As expected, no aggregation was detected in the samples obtained from the culture expressing the wild-type 25Q domain (Fig. 1D).
|
Defective respiration in cells expressing the mutant PolyQ domain
To explore whether mitochondrial dysfunction accounts for part of the cytotoxicity exerted by the expression of the mutant 103Q domain, we polarographically measured the respiratory capacity of the cells as described (48). As shown in Fig. 2A, induction of the expression of the 103Q domain resulted in a progressive decline in the capacity of the cells to respire reaching a minimum of
50% of wild-type respiration after 10 h of induction. The respiratory capacity of the 103Q transformants did not decrease further after 20 h of induction. In contrast, the endogenous cell respiration rate of wild-type untransformed cells and of cells expressing the 25Q domain were indistinguishable (data not shown). The mitochondrial concentration of a, b and c type cytochromes is a reliable gauge of the activity of the mitochondrial respiratory chain enzymes for which they are prosthetic groups. Spectra of mitochondrial cytochromes indicated that wild-type cells and cells expressing the wild-type 25Q construct have 30% more cytochrome b and 20% more cytochrome c than the cells expressing the mutant 103Q construct during 6 h, a difference that was slightly increased over time (Fig. 2B and C).
|
These results showed that mitochondrial dysfunction is an early event in our yeast model of HD and it is measurable after 4 h of induction coinciding with the formation of the first SDS-insoluble aggregates.
The defective respiratory capacity of cells expressing the mutant PolyQ domain results from impaired OXPHOS complex II+III activities
The respiratory defect observed in cells expressing the 103Q domain was characterized polarographically and spectrophotometrically. Because the maximum decrease in respiratory capacity in the 103Q expressing cells was attained after 10 h of induction of expression, we opted to characterize the respiratory defect at that time point.
Respiratory substrate utilization was assayed in isolated mitochondria by measuring the rates of oxygen consumption with NADH and succinate as substrates. The rate of NADH oxidation was reduced by 60% of the wild-type, and succinate oxidation rate was reduced by almost 90% (Fig. 3A). Finally, oxygen consumption was measured in the presence of ascorbate and tetramethyl-p-phenylenediamine (TMPD), which deliver electrons to cytochrome c. This rate was reduced by 30% of wild-type (Fig. 3A). These results were consistent with an alteration of the complex II+III portion of the mitochondrial respiratory chain (succinate dehydrogenase+ubiquinol-cytochrome-c-reductase complex, the latter also called bc1 complex on the basis of its cytochrome composition). The reduction in the ascorbate+TMPD oxidation could be explained by slightly reduced levels of cytochrome c, cytochrome c oxidase (COX) or probably a reduced rate of electron transfer from one to the other.
|
The characterization of the respiratory defect in the 103Q expressing cells included spectrophotometric assays of some of the enzymes of the mitochondrial respiratory chain. The NADH cytochrome c reductase (NCCR) activity was significantly reduced in the 103Q expresing cells by
50% of wild-type, the succinate dehydrogenase activity (SDH) was reduced by 55%, the decylubiquinone cytochrome c reductase (QCCR) activity was reduced by 70%, whereas COX activity remained unaltered (Fig. 3B). Similar values were obtained when the respiratory chain enzymatic rates were normalized by the activity of citrate synthase (data not shown). To visualize OXPHOS complexes in a quantitative way, we exploited the fact that protein complexes retain enzymatic activity on blue native gels (BN-PAGE; 49,50). The activities of COX and ATPase were estimated by in-gel activity assays of mitochondrial enzymes separated by BN-PAGE (Fig. 3C). Under the mild extraction conditions used, the ATPase was extracted as a dimer, and the activity of this enzyme in 103Q cells was equivalent to the activity detected in wild-type and 25Q cells (Fig. 3C). Under the same extraction conditions and electrophoretic separation, COX activity was mostly detected in a band corresponding to the molecular size of the monomeric enzyme (Fig. 3C), although it was possible to detect the dimeric enzyme and a supercomplex formed by the dimeric enzyme plus a bc1 complex. The amount of the supercomplex appears to be reduced by more than 50% in mitochondria from 103Q cells (Fig. 3D), which is consistent with a decrease of complex III or bc1 complex in this strain.
Reduced steady-state levels of OXPHOS complex II+III account for the reduced activity measured in cells expressing a mutant PolyQ domain
To test whether the reduced complex II+III activity in 103Q cells is due to a reduction in the amount of enzyme, we tested the steady-state concentration of several subunits of these enzymes by western blot (Fig. 4A). The amount of several subunits of complex III, including core 1, core 2 and, particularly, cytochromes c1 and b, was significantly reduced by 30–40% of wild-type values (Fig. 4A and B). The amount of Sdh1p, a subunit of complex II, was reduced as well by more than 60% of control values (Fig. 4A and B). Consistent with the enzymatic assays, the steady-state levels of several COX and ATPase subunits were similar in mitochondria from 25Q and 103Q cells, as well as the levels of protein markers of the mitochondrial outer membrane and of the intermembrane space, including cytochrome c (Fig. 4A and B). Cytochrome spectra from the 103Q strain showed a reduction in cytochrome b and a decrease in the 
-absorption bands at 550 nm corresponding to cytochromes c (Fig. 2B and C). To distinguish between cytochromes c and c1, mitochondria were first treated with high salt to remove cytochrome c and were then extracted with deoxycholate to solubilize the remaining cytochromes. Spectra of the salt and detergent extracts (Fig. 4C) confirmed that the decrease in absorption at 550 nm in the 103Q mitochondria was due to the lower concentration of cytochrome c1, which was reduced by 30% of control values (Fig. 4D). The reduction in the amount of some of the subunits of the bc1 complex is interpreted as a reduction in the amount of the holoenzyme. This was confirmed by western blot analysis of a BN-PAGE gel. Probing the membrane with an antibody against the core 1 subunit of complex III, we detected both the complex III dimer and the supercomplex with COX (Fig. 4E), which were reduced by 40% of the control values (Fig. 4F).
|
Expression of the mutant PolyQ domain induces an increase in the production of free radicals
The mitochondrial respiratory chain is the major source of reactive oxygen species (ROS), and complex III plays a central role in its production (51,52). It is conceivable that in our yeast model of HD, the complex II+III deficiency measured could induce the production of ROS, which may significantly contribute to the cytotoxic effect exerted by the mutant PolyQ domain.
To assess intracellular ROS after 5, 10 and 20 h of induction of PolyQ expression with galactose, we used two independent oxidant sensitive probes, MitoSOX Red (Molecular Probes, Eugene, OR) and dihydroethidium (DHE). MitoSOX Red reagent is live-cell permeant and selectively targeted to the mitochondria, where it is oxidized selectively by superoxide and exhibits red fluorescence upon binding to nucleic acids. DHE has been reported to be more sensitive to superoxide than to other ROS but it can also react with species such as hydrogen peroxide (53). DHE is oxidized by ROS to ethidium and oxyethidium, intercalates with DNA in the nucleus and emits red fluorescence.
The fluorimetric measurement of ROS generation showed that after 5 h of induction with galactose, the amount of DHE-derived fluorescence measured in 103Q cells significantly increased by 2.6-fold with respect to wild-type value (Fig. 5A, upper panel). The level of increase was similar after 10 h of induction but increased further to 4.5-fold after 20 h of induction. The values obtained for W303-1A wild-type and 25Q cells were indistinguishable. As a positive control, DHE fluorescence was also measured in wild-type cells incubated in galactose-containing media supplemented with 100 nM antimycin A (AA), a bc1 complex inhibitor that is known to increase the amount of mitochondrial ROS production. AA-treated cells had 5-fold increase with respect to untreated cells, an increase that was similar after 5, 10 and 20 h of incubation.
|
A similar trend was measured by using the MitoSOX Red probe. The amount of MitoSOX fluorescence measured in 103Q cells was significantly increased by 1.9-fold with respect to wild-type values (Fig. 5A, lower panel) after 5 and 10 h of induction and was further increased to more than 6-fold after 20 h of incubation. The AA-treated cells showed an increase of 2.5-fold, which remained constant over the increasing times of incubation. No differences were measured between control and 25Q cells.
The large increase in the fluorescence measured in 103Q cells after 20 h of incubation prompted us to follow the time course of ROS production under a fluorescence microscope. In Figure 5B, a summary of the observations made by using MitoSOX is shown. After 5 and 10 h of incubation, mitochondria from 103Q cells were clearly more brilliant than mitochondria from 25Q cells, which displayed a minimal fluorescence. As expected, the AA-treated control cells were brighter than 103Q cells. In both cases, mitochondria appear fragmented. Interestingly, after 20 h of incubation, a relatively large number of 103Q cells accumulated the dye and fluoresced intensively red (Fig. 5B). After 10 h of incubation, the number of 103Q cells showing red brilliant fluorescence was of 1%, similar to 25Q and control cells. After 20 h of incubation, the percentage of 103Q cells accumulating the dye increased up to 15% (Fig. 5C). Similar values were obtained by using DHE (Fig. 5C) and even by using the non-cell permeant probe Amplex Red (Molecular Probes), commonly used to measure the release of hydrogen peroxide from cells and isolated mitochondria to the media. More than a half of these brilliant red cells were swollen and presented an abnormal morphology, suggesting that they are dead or committed to dye and have lost the integrity of the membrane.
Antioxidant treatment partially preserves cell respiration but does not alleviate the cell growth defect exerted by the mutant PolyQ domain
Antioxidant treatment has been used in nematode and mammalian models of HD to reduce the cytotoxicity associated to the expression of mutant PolyQ domains. For example, CoQ10 has proven to extend survival and delay the development of motor deficits and cerebral atrophy in a transgenic mouse model of HD (54). Treatment with the sirtuin activator and potent antioxidant resveratrol, a polyphenol found in red wine, specifically rescued early neuronal dysfunction phenotypes induced by mutant polyglutamines in transgenic Caenorhabditis elegans (55) and also rescued PolyQ-specific cell death in neuronal cells derived from a 111Q knock-in mice (55).
Oxidative stress impacts the function of the mitochondrial respiratory chain and affects cell respiration (56). Complexes containing iron–sulfur clusters seem to be more sensitive to oxidative damage (57,58). In our model, the enzymes contributing to the observed cell respiration defect are complexes II and II, both containing iron–sulfur clusters. To test whether the decrease in mitochondrial respiration in 103Q cells was the result of an increase in ROS, we measured the endogenous cell respiration of cells grown in the presence of several concentrations of resveratrol after 5 and 10 h of induction with galactose. Concentrations of 10–50 µM resveratrol were able to partially limit the decrease in 103Q cell respiration approximately from 20 to 10% after 5 h of induction, and from 45 to 25% after 10 h of induction (Fig. 5D). No differences were observed after 20 h of induction, time in which 103Q cell respiration was reduced by 50% in the presence and absence of the antioxidant (data not shown). Resveratrol concentrations higher than 50 µM were not used because they had adverse effects on cell growth, as previously described (59).
To test whether antioxidant supplementation to the growth media was able to palliate the cell growth inhibition observed in the yeast strain expressing the 103Q domain, cells expressing the PolyQ domains of wild-type and mutant length were grown in galactose media in the presence or absence of different species and concentrations of antioxidants. We tested 50 and 100 µM CoQ10, the form of ubiquinone synthesized in humans; 50 µM CoQ6, the natural form of ubiquinones in yeast and several concentrations (0.5–50 µM) of resveratrol. As mentioned earlier, resveratrol is a potent antioxidant that significantly extends replicative lifespan in the yeast S. cerevisiae and in higher metazoans like Caenorhabditis elegans and Drosophila melanogaster (59,60). The effects of resveratrol were studied in further detail. Supplementation of growth media containing galactose with resveratrol significantly reduced the amount of free radicals produced in the 103Q strain (Fig. 5A). The proportion of 103Q cells growing during 20 h in galactose-containing media that accumulated high amounts of the probes used for ROS determination was also significantly reduced (Fig. 5C). However, resveratrol neither changed the size and GFP fluorescence of the PolyQ aggregates nor suppressed the apparent mitochondrial fragmentation observed after 10 and 20 h of incubation in the presence of galactose (data not shown). More importantly, resveratrol (Fig. 5E) as well as any of the earlier-mentioned antioxidant compounds (data not shown) failed to produce any effect on the rate of growth of the strains expressing 25Q and 103Q when the cells were grown in solid or liquid media. Thus, although not relieving the detrimental effect of 103Q on cell growth, resveratrol probably inhibits or slows some of the final stages of cell death in the 103Q-expressing cells. These results suggested that oxidative damage does not play a primary role in the cytotoxicity exerted by the mutant PolyQ domain and that mutant PolyQ interferes with the normal functioning of the cells upstream of the ROS production.
Mitochondrial distribution is altered in cells expressing the mutant PolyQ domain
The fragmented mitochondrial pattern observed by using MitoSOX suggested that mitochondrial morphology and distribution within the cell could be affected in cells expressing 103Q. We have explored these parameters more carefully by using a deconvolution fluorescence microscope and staining mitochondria with the cationic dye Mitotracker Red-CM-H2XRos. Mitochondria in 103Q cells progressively appear more globular, fragmented and forming a less organized network than in 25Q cells (Fig. 6A). The morphology of the mitochondrial web was clearly altered in 65% of cells expressing 103Q after 4 h of induction, a phenotype that was uniform over the cell population after 6 h of induction (Fig. 6B). After 20 h of induction, 15% of the cells, probably dead cells, accumulated large amounts of Mitotracker, fluoresced intensively red, as described earlier for the dyes used to detect ROS production.
|
PolyQ aggregates interact with the mitochondrial network
The pattern of mitochondrial alterations described in the previous sections could be directly related to the misfolding/aggregation of the mutant 103Q domains, secondary to an earlier disturbance of another cell structure, or both.
Altered mitochondrial respiration could be the result of mutant PolyQ proteins either in a diffuse or aggregate state interacting with the mitochondrial membranes. We have explored the co-localization of mitochondria with PolyQ-containing aggregates by staining mitochondria with Mitotracker Red. As shown in Figure 6A and C, some portions of the mitochondrial network seem to co-localize with aggregates (yellow spots in the ‘overlap’ panels), suggesting possible interaction points, although it is clear that the number of interactions does not appear to be overwhelming at least up to 6 h of induction. After 10 h of induction, it is evident that PolyQ aggregates surround portions of the disrupted mitochondrial network (Fig. 6C). It is necessary to take into account that this method is powerful to obtain information about changes in mitochondrial distribution as explained earlier and about its interaction with big aggregates. However, it is evident that it cannot determine whether misfolded but non-aggregated proteins interact with the mitochondrial membranes.
Alterations in mitochondrial distribution and function could be a secondary effect of PolyQ misfolding/aggregation
The mitochondrial alterations described earlier could be the secondary result of a previous alteration in a related cellular function. It is well known that the alteration of some cellular structures significantly affects mitochondrial biogenesis and performance. The cytoskeletal network plays a crucial role in mitochondrial inheritance and distribution in the cell. In mammalian cells, disturbance of microtubule cytoskeleton produces abnormal accumulation of mitochondria and other organelles in perinuclear regions; in neurons, it arrests the migration of mitochondria along the axons, as shown in Charcot–Marie–Tooth disease (61) and HD (35). Mitochondria form a dynamic network of interconnected tubules. In S. cerevisiae, the actin cytoskeleton plays an important role in mitochondrial distribution and inheritance (62,63).
Using an HD yeast model, it was recently shown that expression of expanded PolyQ, followed by its aggregation, led to specific and acute inhibition of endocytosis, which preceded growth inhibition (64). Some components of the endocytic machinery were efficiently recruited into the PolyQ aggregates. Furthermore, in cells with PolyQ aggregates, cortical actin patches were delocalized and actin was recruited into the PolyQ aggregates (64). We have explored the integrity of the actin cytoskeleton in our yeast model of HD. As shown in Figure 7A, the actin network (cortical patches and cables) appears normal (polarized towards the bud) in 25Q expressing cells after different times of induction (Fig. 7A). In cells expressing 103Q, the actin patches were disorganized in a small percentage of cells (10%) after 2 h of induction (as in the bottom cell of the 2h panel in Fig. 7A). After 4 h, most cells contained 103Q aggregates and showed depolarized actin patches (Fig. 7A and B). After longer times of induction, cells with 103Q aggregates showed either disappearance or delocalization of actin patches and cables, consistent with a destabilized actin cytoskeleton (Fig. 7A and B). In more than 50% of the cells, at least one of the actin patches co-localized with the 103Q aggregates [yellow color in the overlap image (phalloidin+GFP) in Fig. 7C].
|
It is well known that actin cytoskeleton disturbances will affect the distribution of mitochondria and other organelles within the cell (63,65). In our model, mitochondrial distribution was altered in cells expressing 103Q after 4 h of induction, which could be, at least at some degree, a consequence of a destabilized actin cytoskeleton.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Although the cellular mechanisms by which PolyQ domains of pathological length cause neurodegeneration in HD are not fully understood, a large body of evidence from several sources indicates that a defect of mitochondrial energy metabolism could be involved (18–29,31,32). For example, it has been recently shown that early mitochondrial calcium homeostasis defects are detected in HD patients and transgenic mice, as a direct effect of polyglutamines (31,32). However, the concept of defective mitochondrial respiratory chain enzyme activities as the cause of at least some of the symptoms associated with HD is still controversial. There is no doubt that toxin models of HD are induced by 3-nitropropionic acid or malonate, both inhibitors of complex II of the mitochondrial respiratory chain (21–24). In addition, postmortem caudate and putamen in HD patients have been reported to present severe deficiencies in mitochondrial respiratory chain complexes II and III (18,25–29), and expression of complex II subunits is preferentially decreased in the striatum of HD patients compared with controls, as well as in striatal cell cultures (30). However, in other studies, mitochondrial respiration and ATP production in striatal cells expressing mutant htt have been found significantly impaired while the activity of the respiratory enzymes was found within the normal range (34). In addition, the mechanisms responsible for the respiratory defect are largely unknown. Here, we have created a yeast model of HD to investigate whether mitochondrial OXPHOS function is altered upon expression of mutant PolyQ domains and to identify the possible mechanisms involved.
Mitochondrial alterations are manifested in our model soon after large SDS-insoluble aggregates start to form; however, our results do not eliminate the possibility that misfolded PolyQ domains could actually be responsible for the mitochondrial toxicity. The mechanism of toxicity is obviously not related to a possible loss of the wild-type function of htt, which is not present in yeast, but to the misfolding/aggregation of an NH2-terminal fragment of htt containing a PolyQ domain of pathogenic length. The alteration in cell respiration could be explained by a disturbance of the mitochondrial membranes directly or indirectly interacting with the aggregates. We did observe some contact points of the mitochondrial web with amorphous aggregates which increased over time when they progressively populated the cytoplasm. In our model, the distribution and morphology within the cell of the mitochondrial network are also progressively disturbed, which could be either the cause or the result of the respiratory deficiency. The mitochondrial network disturbance is also contributed by a disruption of the actin cytoskeleton, observed early after expression of the mutant 103Q domain. This observation is in agreement with a recent report showing that PolyQ aggregation in yeast leads to specific and acute inhibition of endocytosis and that mutations affecting actin dynamics enhance mutant PolyQ toxicity (64). In S. cerevisiae, the integrity of the actin cytoskeleton is necessary for mitochondrial distribution and inheritance. Interestingly, decreased actin dynamics have been associated with cell death and aging in yeast, and it has been shown to cause depolarization of the mitochondrial membrane and an increase in ROS production (66). A preliminary analysis of the respiratory capacity of some actin mutants in our laboratory has shown that such a defect induces a variable but significant decrease of the respiratory capacity of the cells to 60–90% of control values and an increase in ROS production (data not shown), suggesting that actin cytoskeleton alterations could indirectly contribute to the respiratory defect observed in our yeast model of HD. As per discussion that follows, upon expression of 103Q in our yeast model of HD, mitochondrial respiration is decreased and the burden of ROS, including mitochondrial superoxide, is significantly enhanced, early before cells stop dividing and die, suggesting that interactions among the misfolded/aggregated PolyQ domains, actin cytoskeleton and mitochondria are not only important for distributing the organelle within cells, but also for events leading to mitochondria-dependent cell death. It is important to notice that these interactions could also be relevant in neurons, in which, although long-distance mitochondrial transport is assisted by the microtubules, the actin cytoskeleton is required for short-distance mitochondrial movements and for immobilization of the organelle at the cell cortex, which may be important for retaining mitochondria at sites of high ATP utilization (67,68). Alterations in the microtubule cytoskeletal network have been found in transgenic mouse models of HD, in which these defects occurred early in development prior to the onset of measurable neurological or mitochondrial abnormalities (35). In addition, mutant htt aggregates have been found to impair mitochondrial movement and trafficking in primary cortical neurons (69). Our results suggest that the interaction of PolyQ domains, actin cytoskeleton and mitochondria could play an important role in the HD pathogenesis and might help guide future efforts in cellular and animal models to elucidate whether this pathway contributes to neurodegeneration in HD.
The mitochondrial OXPHOS alteration caused by the presence of mutant PolyQ domains in our yeast model is intriguingly specific. Cell respiration is progressively compromised in cells expressing 103Q, a defect caused by a specific alteration in the mitochondrial respiratory chain segment including complex II+III. Several lines of evidence support this specificity. First, mitochondrial cytochrome spectra showed a progressive decrease in the amount of cytochromes b and c1, whereas the amounts of type A cytochromes, the prosthetic groups of COX, remained essentially unaltered. Secondly, oxygen consumption and spectrophotometric measurement of mitochondrial respiratory chain activities showed that complexes II and III activities were significantly decreased in 103Q cells, whereas COX activity was indistinguishable from the activity measured in cells expressing the non-pathogenic 25Q domain. Finally, the steady-state levels of complex II+III were decreased after 10 h of induction, suggesting not only a functional disturbance, but also a progressive interference with the balance between assembly and disassembly of these enzymes in 103Q cells. The specificity of these defects in the mitochondrial respiratory chain of our yeast model highly resembles the alterations found in the caudate and putamen of HD patients (18,25–29). These results are suggestive of a conserved mechanism of polyglutamine-induced mitochondrial toxicity from yeast to humans.
How might mutant PolyQ domains induce mitochondrial OXPHOS toxicity affecting specifically the function and steady-state levels of complex II+III while the other complexes remain mostly unaffected? The mechanism probably involves some specific characteristics of these two enzymes in terms of composition, structure and function. Complexes II and III are both multimeric proteins, as other respiratory enzymes, and contain iron–sulfur clusters. Complex II or succinate:ubiquinone reductase consists of four nuclear DNA-encoded polypeptides, among them, an iron–sulfur protein and a flavoprotein. Electrons coming from the oxidation of succinate to fumarate are channeled through complex II to ubiquinone. Complex III, bc1 complex or ubiquinol:cytochrome-c oxidoreductase is formed by three catalytic subunits (cytochrome c1, Rieske iron–sulfur protein and the single mitochondrial DNA-encoded subunit cytochrome b) and eight non-catalytic subunits. The enzyme works through a modified Q-cycle mechanism (reviewed in 70) and catalyzes the transfer of two electrons from ubiquinol to ferrycytochrome c, using the free-energy change to transport protons across the inner membrane, which contribute to the electrochemical gradient that drives aerobic synthesis of ATP. Several concurrent mechanisms could contribute to the specific complex II+III defect. A first mechanism involves defective assembly of these enzymes when key subunits are available in limited amounts. Because mutant PolyQ is readily accumulated in daughter cells, the misfolded and aggregated form could partially prevent mitochondrial complex II+III assembly in this set of cells. Misfolded/aggregated PolyQ domains interacting with mitochondrial membranes directly or indirectly could alter the expression and import of particular subunits of these complexes, as it has been proposed for complex II subunits in neuronal models of the disease (30). These putative subunits could be preferentially trapped by misfolded PolyQ domains and be included as part of the amorphous aggregates or, simply, be targeted for degradation. In the presence of limiting amounts of some of the component subunits, the assembly of these enzymes would be partially compromised. As an alternative but not mutually exclusive hypothetical mechanism, supported by the results presented in this paper, the direct and indirect interactions of misfolded/aggregated mutant PolyQ domains with the mitochondrial membranes could also promote alterations in the function and/or stability of complex II+III, which could be disassembled. In this mechanism, our results highlight the role of the significant increase in ROS generation measured upon expression of 103Q. It is fair to postulate that the disturbance in the actin and mitochondrial networks produces an increase in ROS production, measured early after induction of 103Q expresion. ROS could alter protein enzymes, preferentially those containing iron–sulfur clusters such as complexes II and III, which are highly susceptible to oxidation, resulting in further respiratory deficiency. Altered mitochondrial complex II+III will perpetuate the build-up of ROS because they are important sites for its production within mitochondria (reviewed in 71,72). In support of this possibility, ROS quenching with resveratrol partially alleviates the cell respiration defect measured in our model in the early stages after expression of 103Q. Further studies with our model will focus on clarifying these questions and the contribution of each of these mechanisms.
To what extent do the respiratory defect and mitochondrial complex II+III alteration measured in our yeast model contribute to the cytotoxicity induced by the expression of mutant PolyQ domains? Respiratory-deficient yeast strains grow at rates lower than wild-type in media containing galactose, a carbon source that can be used for fermentation but it does not induce repression of the expression of OXPHOS genes. However, the growth defect observed in cells expressing 103Q is significantly more stringent than the alteration observed in respiratory-deficient strains with mutations in components of the OXPHOS system (data not shown), suggesting that the respiratory dysfunction in 103Q cells is just part of the cytotoxicity. In our model, we could hypothesize that the contribution of mitochondrial toxicity to the cell growth defect is mediated by the reduced respiratory capacity and by the generation of excessive amounts of ROS measured in our model. However, quenching of ROS with resveratrol and other antioxidants failed to alleviate the growth defect of yeast cells expressing 103Q, although cell respiration was partially protected as mentioned in the previous paragraph. Interestingly, in the presence of resveratrol, cell death seems to be partially inhibited, in agreement with the previously reported observation that supplementation of the media with the antioxidant -tocopherol had no effect on the rate of growth of 103Q cells but inhibited the final stages of cell death including DNA cleavage (73). Increase in ROS generation probably contributes to the 103Q cytotoxicity, although inhibition of cell growth seems to be independent of ROS.
In conclusion, alterations in cell respiration resulting from reduced complex II+III activities contribute to the cytotoxic effect produced by the misfolding/aggregation of a mutant PolyQ fragment in a yeast model of HD. Taken together, our results support a mechanism for mitochondrial dysfunction in our yeast model of HD involving at least two different pathways (Fig. 8). Misfolded/aggregated mutant PolyQ domains could directly interact with the mitochondrial membranes, leading to the increased ROS production and the observed functional and physiological mitochondrial alterations. As a concurrent mechanism, misfolded/aggregated mutant PolyQ could alter other structures in the cell, such as the cytoskeleton, which will lead to disturbances in mitochondrial distribution, increase in ROS production and decline in mitochondrial function. Given the complexity of PolyQ-induced neuronal death and formation of intracellular aggregates, development of adequate cellular models is critical to dissect cellular mechanisms of these processes, and our yeast model of HD has proven to be a useful tool. Future studies with yeast models should help us establish the exact mechanism by which these mitochondrial alterations are produced and devise strategies for therapeutic interventions.
|
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Yeast strains and media
The genotypes and sources of the S. cerevisiae strains used are listed in Table 1. The compositions of the growth media have been described elsewhere (74). The following media were used routinely to grow yeast: YPD (2% glucose, 1% yeast extract, 2% peptone), YPGal (2% galactose, 1% yeast extract, 2% peptone), YPEG (2% ethanol, 3% glycerol, 1% yeast extract, 2% peptone). Prior to induction of PolyQ expression with 2% galactose, all cultures were grown on media containing non-fermentable carbon sources (YPEG). All our experimentation was done by using freshly transformed cells.
|
Cloning of 25Q-GFP and 103Q-GFP in integrative plasmids
Two PolyQ constructs (25Q and 103Q in the plasmid pYES2) were obtained from Professor Michael Y. Sherman (13,19). In brief, exon1 sequences containing the first 17 amino acids followed by 25 or 103 glutamines were fused in frame with a GFP tag at the C-terminus of each construct, and a FLAG tag was attached to their NH2 terminus (resulting in 25Q or 103Q). These constructs were originally prepared taking into account that homopolymeric tracts of CAG, which is the naturally occurring glutamine codon in htt, are inherently unstable (75). To minimize instability problems, yeast models prepared in several laboratories have taken advantage of the fact that glutamine is encoded by both CAG and CAA and that mixed-codon repeats are considerably more stable (19). We obtained and used DNA constructs for the expression of htt with alternating CAG/CAA repeats of different lengths (25Q and 103Q) previously described by Kazantsev et al. (19).
In each case, a KpnI/XbaI fragment containing the PolyQ domain fused to GFP was excised and inserted in either an episomal Yep351 plasmid or an integrative YIp351 plasmid (47) previously modified to express the proteins under the control of the Gal1 promoter.
Mitochondrial preparation
Mitochondria were prepared from wild-type strains expressing 25Q-GFP and 103Q-GFP domains grown in media containing ethanol and glycerol as the carbon sources, transferred to media containing 2% galactose to induce expression of the PolyQ domains by times ranging from 2 to 20 h. Mitochondria were prepared by the method of Faye et al. (76), except that zymolyase 20T (ICN Biochemicals Inc., Aurora, OH) instead of Glusulase was used for the conversion of cells to spheroplasts.
Oxygen consumption measurements in isolated mitochondria and in whole cells
Mitochondria prepared from the different strains after 10 h of induction in galactose were assayed polarographically for NADH oxidase and succinate oxidase using a Clark-type polarographic oxygen electrode from Hansatech Instruments (Norfolk, UK) at 25°C as described (48). Briefly, oxygen utilization was measured polarographically in 1 ml of standard medium containing 20 mM KH2PO4, pH 7.4, 2 mM ethilenediaminetetraacetic acid (EDTA). In each experiment, 100 µg of mitochondrial protein was used. The oxidation of NADH (2 mM) was measured. The reaction was inhibited with KCN (700 µM). The oxidation of succinate (10 mM) was also performed and inhibited with malonate (10 mM), a complex II inhibitor. In another experiment, the oxidation of ascorbate (10 mM) plus N,N,N',N'-TMPD (0.2 mM) through COX was also measured in isolated mitochondria and inhibited with 700 mM KCN. The specific activities reported were corrected for the inhibitor-insensitive oxygen consumption.
Respiration was also assayed in whole cells in the presence of galactose (cell respiration) after 2, 4, 6 8, 10, 12 and 20 h of induction of the expression of the 25Q and 103Q domains from integrative plasmids. The specific activities reported were corrected for KCN-insensitive respiration.
Mitochondrial respiratory chain enzyme spectrophotometric measurements
Mitochondria prepared from the different strains after 10 h of induction in galactose were used for spectrophotometric assays carried at 24°C.
KCN-sensitive COX activity was assayed in 50 µg of mitochondria permeabilized with potassium deoxycholate (KDOC), as described, (48) by measuring at 550 nm the oxidation of 50 µM reduced cytochrome c in a medium containing 20 mM KH2PO4 (pH 7.4). The addition of 0.3 mM KCN inhibited the reaction.
Antimycin A-sensitive NCCR activity was assayed in 25 µg of mitochondria permeabilized with KDOC as described (48) by measuring at 550 nm the reduction of oxidized 50 µM cytochrome c using 0.4 mM NADH as the electron donor in a medium containing 20 mM KH2PO4 (pH 7.4) and 2 mM EDTA. The addition of 0.4 µM antimycin A inhibited the reaction.
Malonate-sensitive SDH was measured at 600 nm in 50 µg mitochondria following the phenazine methosulfate-mediated (1 mM) reduction of 80 µM 2,6-dichlorophenolindophenol (DCPIP) using 10 mM succinate as the electron donor in a medium containing 20 mM KH2PO4 (pH 7.4) and 2 mM EDTA in the presence of 0.2 mM ATP as previously described (77). The addition of 10 mM malonate inhibited the oxidation of succinate.
Antimycin A-sensitive QCCR activity was measured in 50 µg of mitochondria as described (78). Basically, the assay was performed at 550 nm using 50 µM reduced cytochrome c as the electron acceptor and 50 mM duroquinol as the donor in a medium containing 20 mM KH2PO4 (pH 7.4) and 2 mM EDTA in the presence of 0.3 mM KCN. The addition of 0.4 µM antimycin A allowed us to distinguish between the reduction in cytochrome c catalyzed by the complex III and the non-enzymatic reduction of cytochrome c by the reduced quinone.
Citrate synthase activity was measured in 25 µg of mitochondria at 412 nm as described (77) following the reduction of 0.1 mM 5,5'-dithio-bis(2-nitrobenzoic acid) in the presence of 0.2 mM acetyl-CoA and 0.5 mM oxalacetic acid in a medium containing 10 mM Tris–HCl, pH 7.5, and 0.2% Triton X-100.
Mitochondrial cytochromes spectra
Mitochondria were prepared from the wild-type strain W303-1A, and from the same strain transformed with a chromosomally integrative plasmid expressing either the 25Q (W303-1A+YIp/25Q) or 103Q (W303-1A+YIp/25Q) PolyQ domains, growing in the presence of galactose for the indicated times. Isolated mitochondria were extracted at a protein concentration of 5 mg/ml in 20 mM Tris–HCl, pH 7.5, 1 M KCl, 1% KDOC, conditions that quantitatively solubilize all the cytochromes (79). These conditions quantitatively solubilize mitochondrial cytochromes of yeast. Samples of the extract were either oxidized with ferricyanide or reduced with sodium dithionite and the difference spectra were measured at room temperature using a UV-2401PC Shimadzu spectrophotometer.
In a second experiment, we wanted to obtain an estimation of cytochromes c and c1, both have a maxima of absorbance at 550 nm. Cytochrome c was extracted from the mitochondrial membranes with 1 M KCl. Subsequently, the salt-extracted membranes were resuspended in 20 mM Tris–HCl, pH 7.5, 1 M KCl, 1% KDOC at a protein concentration corresponding to 5 mg/ml starting mitochondrial protein. The mixture was centrifuged at 250 000gav, and difference spectra of the reduced versus oxidize extracts were recorded as described earlier.
Blue native polyacrylamide gel electrophoresis
Blue native polyacrylamide gel electrophoresis (BN-PAGE) for the identification of individual mitochondrial respiratory chain complexes and supercomplexes was performed as described (50,80). With digitonine at a protein:detergent ratio of 1:2, 200 µg of mitochondrial proteins were extracted. The extracts were supplemented with a solution of 5% Coomassie blue G250 to a detergent:G250 ratio of 4:1. The amount of extract equivalent to 100 µg of mitochondria was loaded on 5–13% blue native gradient gels as reported (50,80). High-molecular-weight markers for native electrophoresis (Amersham Biosciences, Piscataway, NJ) were treated the same as the samples (1.5 M aminocaproic acid, 50 mM Bis–Tris, pH 7.0, lauryl maltoside and Coomassie brilliant blue G). After native electrophoresis, proteins were transferred to polyvinylidene difluoride membrane for immunoblot analyses as described previously (50,80).
To perform in-gel activity staining, the blue cathode buffer was replaced by colorless buffer after the gel was run for 3 h at 30 V. The gel was run until the dye front left the gel. Native gels were histochemically stained for ATP hydrolysis activity or COX activity, as previously reported (50).