The yeast protein Ubx4p contributes to mitochondrial respiration and lithium–galactose–mediated activation of the unfolded protein response

In the presence of galactose, lithium ions activate the unfolded protein response (UPR) by inhibiting phosphoglucomutase activity and causing the accumulation of galactose-related metabolites, including galactose-1-phosphate. These metabolites also accumulate in humans who have the disease classic galactosemia. Here, we demonstrate that Saccharomyces cerevisiae yeast strains harboring a deletion of UBX4, a gene encoding a partner of Cdc48p in the endoplasmic reticulum–associated degradation (ERAD) pathway, exhibit delayed UPR activation after lithium and galactose exposure because the deletion decreases galactose-1-phosphate levels. The delay in UPR activation did not occur in yeast strains in which key ERAD or proteasomal pathway genes had been disrupted, indicating that the ubx4Δ phenotype is ERAD-independent. We also observed that the ubx4Δ strain displays decreased oxygen consumption. The inhibition of mitochondrial respiration was sufficient to diminish galactose-1-phosphate levels and, consequently, affects UPR activation. Finally, we show that the deletion of the AMP-activated protein kinase ortholog–encoding gene SNF1 can restore the oxygen consumption rate in ubx4Δ strain, thereby reestablishing galactose metabolism, UPR activation, and cellular adaption to lithium–galactose challenge. Our results indicate a role for Ubx4p in yeast mitochondrial function and highlight that mitochondrial and endoplasmic reticulum functions are intertwined through galactose metabolism. These findings also shed new light on the mechanisms of lithium action and on the pathophysiology of galactosemia.

In the presence of galactose, lithium ions activate the unfolded protein response (UPR) by inhibiting phosphoglucomutase activity and causing the accumulation of galactoserelated metabolites, including galactose-1-phosphate. These metabolites also accumulate in humans who have the disease classic galactosemia. Here, we demonstrate that Saccharomyces cerevisiae yeast strains harboring a deletion of UBX4, a gene encoding a partner of Cdc48p in the endoplasmic reticulumassociated degradation (ERAD) pathway, exhibit delayed UPR activation after lithium and galactose exposure because the deletion decreases galactose-1-phosphate levels. The delay in UPR activation did not occur in yeast strains in which key ERAD or proteasomal pathway genes had been disrupted, indicating that the ubx4⌬ phenotype is ERAD-independent. We also observed that the ubx4⌬ strain displays decreased oxygen consumption. The inhibition of mitochondrial respiration was sufficient to diminish galactose-1-phosphate levels and, consequently, affects UPR activation. Finally, we show that the deletion of the AMP-activated protein kinase orthologencoding gene SNF1 can restore the oxygen consumption rate in ubx4⌬ strain, thereby reestablishing galactose metabolism, UPR activation, and cellular adaption to lithiumgalactose challenge. Our results indicate a role for Ubx4p in yeast mitochondrial function and highlight that mitochondrial and endoplasmic reticulum functions are intertwined through galactose metabolism. These findings also shed new light on the mechanisms of lithium action and on the pathophysiology of galactosemia.
Lithium is one of the most common drugs used in the treatment of bipolar disorder patients. The molecular mechanisms of lithium action are not fully elucidated, but evidence points to a role for lithium as a GSK-3␤ inhibitor, as an inducer of inositol depletion, among others hypotheses (1). Lithium is also capable of inhibiting the activity of the enzyme phosphoglucomutase because it competes with its cofactor magnesium (2). Because of this inhibition, cells treated with lithium in the presence of galactose accumulate toxic metabolites of the Leloir pathway in yeast (2)(3)(4) and human cells (5). Accumulation of intermediate metabolites of galactose metabolism such as galactose-1-phosphate also occurs in patients with classic galactosemia, an inborn error of metabolism caused by mutations in the GALT gene (6,7). Yeast is a eukaryotic cell model commonly employed in studies to better understand human diseases because of its similarities with human metabolism and its vast toolkit for genetic manipulation (8). Yeast cells treated with lithium in the presence of galactose exhibit modifications in cellular homeostasis such as changes in glycogen, calcium, and phosphate levels and a metabolic shift from fermentation to respiration (3,(9)(10)(11). Another important feature of lithium/galactose treatment is the increase in endoplasmic reticulum (ER) 3 stress and unfolded protein response (UPR), both in yeast and in human cells (3,5). This UPR activation depends on the synthesis of galactose-1-phosphate and also occurs in genetic models of classic galactosemia (3,12,13).
The ER is an organelle in which protein folding, glycosylation, and lipid synthesis take place. In yeast, during ER stress conditions, the communication between ER and nucleus is mediated by the conserved kinase/endoribonuclease protein Ire1p. This transmembrane protein, together with the tRNA ligase Trl1p, promotes the cytosolic splicing of the HAC1 mRNA, which in turn codes for the transcription factor Hac1p (14,15). Hac1p induces the activation of genes responsible for restoring ER homeostasis such as ER-resident chaperones (16). The Ire1/Hac1 branch of UPR is the only one present in yeast and is conserved from yeast to mammals (Ire1␣/Xbp1), being relevant to different diseases (17)(18)(19)(20). A second way that cells can cope with ER stress is by alleviating proteotoxicity through the ER-associated protein degradation (ERAD) pathway. ER-localized misfolded proteins retrotranslocate to the cytosol, where they are ubiquitinated and degraded by the proteasome (21). In this work, we demonstrate the function of Ubx4p (Cui1p), a Cdc48p protein adaptor in the ERAD pathway (22,23), on the cellular adaption to lithium and galactose toxicity. Ubx4p contains a ubiquitin regulatory X (UBX) domain that mediates physical interaction with Cdc48p (24) and an ubiquitin-like domain that is commonly found in proteins that can interact and recruit proteasome (25). In fact, both UBX and ubiquitin-like domains are necessary for Ubx4p interaction with the proteasome (23). Ubx4p homolog in humans is called ASPL/TUG, a protein that interacts with p97/VCP (homolog of Cdc48 in mammals) (26). Importantly, the expression of ASPL/ TUG in an ubx4⌬ strain is capable of recovering the proteasome localization defect of this strain. ASPL/TUG in mammals can also bind to the GLUT4 (glucose transporter type 4) and control its translocation to the plasma membrane upon insulin stimulation (27)(28)(29).
Here we show a novel and ERAD-independent role for Ubx4p in controlling mitochondrial and galactose metabolism. Our data demonstrate that UBX4 deletion impairs mitochondrial respiration, which in turn affects galactose metabolism and activation of the UPR. For these reasons, ubx4⌬ mutants are less tolerant to lithium/galactose.

Results
Based on a genetic screen to identify new regulators of UPR activation induced by lithium and galactose treatment, we identified that the deletion of the UBX4/YMR067C gene delays UPR activation (Fig. 1A). In this context, the inhibition of the UPR pathway (e.g. deletion of IRE1 or HAC1 genes) causes loss of cell viability (3). We did not observe the same for the ubx4⌬ mutant, which remained viable after 24 h of treatment with lithium in the presence of galactose (Fig. 1B), probably because this strain can still fully activate UPR (as measured by HAC1 mRNA splicing) after 4 h of treatment. These results suggest that Ubx4p is not interfering directly with Ire1p activity. Accordingly, the deletion of UBX4 did not affect UPR activation by tunicamycin treatment (Fig. S1).

Figure 1. UBX4 is necessary for proper UPR activation and galactose metabolism in lithium-galactose-treated cells.
A, the levels of HAC1 mRNA splicing were monitored at the indicated time points by RT-PCR in control and ubx4⌬ cells grown in YPGal and treated with 30 mM LiCl. The data are presented as means Ϯ S.E. Two-way ANOVA was performed with Sidak's multiple comparisons test. **, p Ͻ 0.01. B, the indicated strains, control (n ϭ 3), ubx4⌬ (n ϭ 3), and ire1⌬ (n ϭ 1) were grown in the presence of 30 mM LiCl for 24 h in YPGal and, after this incubation period, plated on YPD for colony-forming unit measurement. Normalization was performed by the number of colony-forming units in strains grown in YPGal without LiCl. C, relative expression levels of GAL1 and GAL2 mRNAs, normalized by TFC1 gene, were determined by quantitative RT-PCR from control or ubx4⌬ cells grown in YPGal (n ϭ 5-7 biological replicates). Student's t test was used at this analysis to compare the control versus ubx4⌬ groups, and no statistically significant difference between groups was observed. We obtained similar results by normalizing the data with TAF10 gene (data not shown). D and E, levels of intracellular galactose (D) or intracellular galactose-1-phosphate (E) were measured on the indicated strains grown in YPGal for 1 h in the presence or absence of 30 mM LiCl (n ϭ 6 -7 biological replicates). ****, p Ͻ 0.0001. A 600 nm , absorbance at 600 nm.

Ubx4p regulates yeast mitochondrial respiration
This raised the hypothesis that the deletion of UBX4 is delaying ER stress occurrence in a lithium-galactose-specific manner. As previously described by our group, UPR activation in models of galactosemia can be prevented by the deletion of yeast galactokinase GAL1 (3), suggesting that the synthesis of galactose-1-phosphate is an essential event for the UPR activation in this condition. Although we did not observe differences in the expression of genes of Leloir pathway (GAL1 and GAL2) (Fig. 1C), we measured the galactose and galactose-1-phosphate levels in the ubx4⌬ mutant, and we observed an 82-85% decrease in galactose (Fig. 1D) and a 77-93% decrease in galactose-1-phosphate levels ( Fig. 1E) when compared with a control strain, even without lithium. These results suggest that the delayed UPR activation in the ubx4⌬ strain is due to the overall decreased levels of galactose-1-phosphate accumulated in this strain.
Ubx4p is known to participate in the ERAD pathway and to regulate proper proteasome localization (22,23). Thus, we reasoned that mutations in other components of these pathways would also influence UPR activation in lithium/galactose-treated cells. Nonetheless, the deletion of UBC7, a central effector of ERAD pathway (21,30), or the deletion of RPN4, a transcription factor and master regulator of proteasome genes (31-33), did not affect the capacity of the cell to induce HAC1 mRNA splicing when treated with lithium and galactose (Fig. 2).
A strategy to reduce lithium-galactose toxicity is to prevent galactose-1-phosphate accumulation (3,4,34). We assumed that because the deletion of UBX4 decreases intracellular galactose-1-phosphate content, the ubx4⌬ would be more tolerant to the treatment of lithium and galactose. Surprisingly, we observed that UBX4 deletion results in a decreased cellular tolerance to lithium in galactose-containing medium but not in glucose-containing medium (Fig. 3A). A decrease of ERAD or proteasome cellular function does not explain this increased sensitivity because neither the deletion of RPN4 nor the deletion of UBC7 modified the cellular sensitivity to lithium and galactose (Fig. 3B). We also tested mutations in other components of these pathways but could not observe significant differences in cell growth among those strains (Fig. S2).
Our data suggested that the deletion of UBX4 might be acting on the lithium-galactose tolerance via other mechanism rather than regulating galactose metabolism. Previous work suggested that the deletion of the UBX4 gene increases the frequency of petite colonies, which in general is related to deficiencies in mitochondrial metabolism (35). Hence, we monitored the respiration capacity of this strain, and we observed a greater than 10-fold decrease in oxygen consumption rate in the ubx4⌬ cells compared with the cells of a control strain when grown in galactose (Fig. 4A). The impaired galactose metabolism observed in ubx4⌬ does not seem to be solely responsible for the decreased mitochondrial respiration because the gal1⌬ strain, which cannot metabolize galactose because of the lack of galactokinase, presents only a slight decrease in mitochondrial oxygen consumption under the same conditions. This result is expected because these strains are growing in a rich medium containing other carbon sources, such as amino acids, which can sustain mitochondrial respiration when galactose metabolism is impaired. We also observed that the deletion of UBX4 reduced oxygen consumption in the absence of galactose metabolism, for example, when cells were grown on YPGly (medium containing glycerol, a nonfermentable carbon source) (Fig. S3A). The deletion of neither UBC7 nor RPN4 affected oxygen consumption rates ( Fig. 4A and Fig. S3A), suggesting that an impairment of ERAD or proteasome function does not contribute for this ubx4⌬ phenotype. The fact that mitochondrial preparations from ubx4⌬ cells present decreased oxygen consumption capacity when compared with mitochondrial preparations from a control yeast strain (Fig. S3, B and C) further supports the hypothesis that the effect of UBX4 deletion on oxygen consumption in vivo is not only due to its impact on galactose metabolism.
A hypothesis to explain the decreased tolerance to lithium and galactose observed in the ubx4⌬ strain is that, under this condition, cells are more dependent on respiration, and because the ubx4⌬ strain presents a decreased mitochondrial oxygen consumption rate, it cannot grow properly. This is in line with previous data from our group showing that lithiumgalactose-treated control cells have a metabolic shift from fermentation to respiration (11). Accordingly, if we simulate the ubx4⌬ defects in both galactose and mitochondrial metabolism by growing the gal1⌬ strain in the presence of antimycin A, an inhibitor of mitochondrial complex III, the cells become unable to grow on galactose (Fig. 4B).
We reasoned that mitochondrial function could be upstream of galactose metabolism and UPR activation in lithium-treated yeast cells. Accordingly, previous studies showed that GAL1 gene expression is down-regulated in strains with mutations in components of the mitochondrial electron transport chain or in yeast strains with reduced mitochondrial membrane potential (36,37). On the other hand, another study observed that yeast cells depleted of mitochondrial DNA (rho 0 ) present an increase in Gal4-dependent transcription (38). Nevertheless, it One-way ANOVA with Dunnett's multiple comparison test was performed, and we did not observe any statistically significant difference between the groups.

Ubx4p regulates yeast mitochondrial respiration
remains to be tested whether mitochondrial dysfunction promotes alterations in galactose metabolite levels. To test these hypotheses under our growth conditions, we first inhibited yeast mitochondrial function by deleting COQ3 or ATP7 genes, which impairs mitochondrial respiratory chain function (39 -41) (Fig. 5A). These mutants recapitulated all ubx4⌬ phenotypes, including decreased cell growth on lithium and galactose (Fig. 5A), the decrease in galactose-1phosphate levels (p Ͻ 0.0001 two-way ANOVA 30 mM of LiCl; control strain versus atp7⌬ or coq3⌬) ( Table 1), and the defect in HAC1 mRNA splicing induced by lithium and galactose treatment (Fig. 5B).
It was previously observed that the deletion of the yeast AMPK homolog SNF1 increases the oxygen consumption rate when yeast cells are grown in glucose medium (42). We observed a similar increase when snf1⌬ strains were grown in the presence of galactose (Fig. 6A). We then deleted the SNF1 gene in the ubx4⌬ strain to test whether this deletion could recover ubx4⌬ mitochondrial function. Surprisingly, SNF1 deletion totally recovered ubx4⌬ oxygen consumption capacity (Fig. 6A). Accordingly, the deletion of SNF1 was capable of recovering the adaption capacity to lithium and galactose treatment and the UPR activation of the ubx4⌬ strain (Fig. 6, B and  C). When we measured galactose-1-phosphate levels in these mutants, we noted that the deletion of SNF1 partially recovered galactose-1-phosphate levels in the ubx4⌬ strain growing in YPGal (Table 1). Despite the fact that snf1⌬ cells treated with lithium and galactose exhibited an ϳ40% reduction in galactose-1-phosphate levels when compared with the control strain (Table 1) (p Ͻ 0.0001 two-way ANOVA 30 mM of LiCl; control strain versus snf1⌬), the deletion of SNF1 partially restored the levels of galactose-1-phosphate in the ubx4⌬ strain (Table 1) (p ϭ 0.0677 two-way ANOVA 30 mM of LiCl; ubx4⌬ versus ubx4⌬snf1⌬). Galactose-1-phosphate is believed to be a toxic metabolite in the cell when accumulated, and although the ubx4⌬snf1⌬ strain has higher levels of galactose-1-phosphate than the ubx4⌬ mutant, the former grows better when treated with lithium and galactose, probably because of the increase in the oxygen consumption rate mediated by the deletion of SNF1.

Discussion
In this work, we showed that UBX4 is necessary for proper UPR activation in cells treated with lithium in the presence of galactose. We presented evidences that UBX4 does not regulate Ire1p activity directly. Instead, it diminishes ER stress by decreasing the intracellular content of galactose-1-phosphate (and possibly other Leloir pathway metabolites), accumulation of which is necessary for UPR activation in these conditions (3). Curiously, we did not observe any changes in UPR activation in

Ubx4p regulates yeast mitochondrial respiration
strains that are defective on ERAD or proteasome activity, suggesting that the effect of UBX4 deletion does not relate to these previously described Ubx4p functions. Importantly, we found that UBX4 deletion has an impact on mitochondrial oxygen consumption rates, which in turn affects galactose metabolism and consequently UPR activation in lithium-galactose-treated cells (Fig. 6D).
The human homolog of Ubx4p, TUG/ASPSCR1, was initially described as a GLUT4 tether, which interacts with this transporter via its conserved UBX domain, thus regulating GLUT4 membrane exposure under insulin stimulation (27). Curiously, our data show that ubx4⌬ has reduced levels of intracellular galactose, even when compared with that of atp7⌬ or coq3⌬ mutants. That could be explained if yeast Ubx4p could directly regulate hexose carrier activity. It will be interesting to test this possibility in the future. This is in agreement with data showing that the galactose transporter GAL2 mRNA level was not reduced on ubx4⌬ strain (Fig. 1C), indicating that GAL2 could be regulated at the post-transcriptional level.
To test whether UBX4 homologs have conserved this mitochondria-related function, we compared the mitochondrial oxygen consumption rates of the Caenorhabditis elegans strain asps-1(ok915), carrying a mutation in the UBX4 homolog gene, with that of a control strain N2. Although we observed that asps-1(ok915) mutant has slow larval development (data not shown), a phenotype observed in other C. elegans mutants with mild mitochondrial dysfunction (43-45), we could not observe alterations in mitochondrial oxygen consumption in this mutant (Fig. S4). Although Ubx4p may not have a conserved role in mitochondrial function in C. elegans, we also speculate that asps-1 can have a conditional or tissue-specific function that is not reflected by the assay performed with whole adult worms under the specific conditions used. Supporting this notion, it was recently observed that muscle-specific TUG/ ASPSCR1 deletion decreases oxygen consumption and influences mitochondrial morphology only when mice are subjected to a high-fat diet (46). Although other experiments need to be performed to fully address this issue, these results suggest that Ubx4p homologs can also impact mitochondrial function in metazoans.
So, how does the deletion of UBX4 causes dysfunction in mitochondrial metabolism? We measured the mitochondrial DNA content per cell as an indicator of the total mitochondrial content but could not observe any difference between ubx4⌬ and control strains (Fig. S5A). However, we did observe a decreased respiratory capacity of mitochondria preparations from ubx4⌬ cells when compared with preparations from control cells (Fig. S3, B and C), indicating that the mitochondria itself is functionally affected by the deletion of UBX4. We also reasoned that the ubx4⌬ strain could manifest redox alterations that might explain the decrease in cell growth. We tested whether the supplementation with the antioxidant N-acetylcysteine would recover ubx4⌬ growth on lithium and galactose, but no significant difference was observed (Fig. S5B). Recently, another protein that contains the UBX domain involved in ERAD pathway, Ubx2p, was shown to be necessary for solving defects in mitochondrial protein translocation via regulation of Cdc48p activity associated with the TOM complex (47). We tested whether the deletion of UBX2 would mimic the phenotype observed in ubx4⌬ of reduced growth on lithium and

Ubx4p regulates yeast mitochondrial respiration
galactose, but we did not observe any significant differences on ubx2⌬ strain growth under this condition (Fig. S2). This result suggests that Ubx2p and Ubx4p have distinct roles to sustain mitochondrial function, at least when cells are grown on galactose. Supporting this hypothesis, it was already shown that Cdc48p has distinct roles in regulating mitochondrial morphology, proteostasis, and yeast capacity to grow in medium containing respiratory carbon source (48 -51). Despite our results suggesting that proteasome function may not be involved with the mitochondrial phenotypes of the ubx4⌬ strain, it has been determined that the ubiquitin-proteasome system is involved in the control of mitochondrial function (52). We are currently trying to understand mechanistically the role of Ubx4p in the function of mitochondria, including a possible cooperation with Cdc48p. It has been previously shown that when yeast cells are grown in the presence of galactose, inhibition of phosphoglucomutase activity by either PGM2 gene deletion or lithium treatment results in an increase in cellular calcium uptake and UPR activation and negatively impacts cell growth (10,53,54). It was proposed that the increased cellular calcium accumulation is induced by an increase in the glucose-1-phosphate/glucose-6phosphate ratio because the deletion of the PFK2 gene (which encodes for phosphofructokinase) decreases the glucose-1phosphate/glucose-6-phosphate ratio and concomitantly sup-presses intracellular calcium accumulation (10,55). We found that PFK2 gene deletion did not affect UPR activation by lithium and galactose treatment (Fig. S6A), although it did confer some growth tolerance (Fig. S6B) as previously reported (10). These results suggest that the UPR activation observed under our experimental conditions is not influenced by the glucose-1-phosphate/glucose-6-phosphate ratio.
The deletion of the yeast AMPK/SNF1 gene increases oxygen consumption in control and ubx4⌬ cells, allowing a normal galactose metabolism and UPR activation with the treatment of lithium. Although in mammals the role of AMPK to sustain oxidative metabolism is well-characterized (56), in yeast the deletion of SNF1 seems to increase mitochondrial respiration by enhancing amino acid uptake and utilization (42). The deletion of SNF1 was already shown to regulate UPR activation in cells treated with tunicamycin (57). The relationship between ER stress and mitochondria was further explored by a recent study showing that mitochondrial function is necessary for cellular adaptation to DTT or tunicamycin (58). Thus, different ER stress conditions may share pathways to mediate the communication between mitochondria and the ER to allow the cell to adapt to stress.
In this work, we show interorganelle communication between mitochondria and ER through the regulation of galactose metabolism. Mitochondria during stressful conditions, such as

Ubx4p regulates yeast mitochondrial respiration
impaired proteostasis or energy deprivation, can send signals to the nucleus in a form of retrograde signaling to re-establish homeostasis (59). In yeast, one mechanism to mediate this communication is through the RTG pathway (60). Nevertheless, we did not observe a correlation between the decreased oxygen consumption rates and increased expression of the CIT2 gene, a target of the RTG pathway, in the ubx4⌬ strain (data not shown). Still, it will be interesting to test whether this signaling between mitochondria and galactose metabolism is conserved in other biological systems. In mammals, galactose metabolism is regulated by UPR after feeding (61). Thus, it is tempting to conceive that our results are a description of a primitive signaling that was modified during evolution to adapt to galactose toxicity in the context of postprandial metabolic changes. Our findings provide a link between mitochondrial homeostasis and UPR activation regulated by Ubx4p and contribute to our understanding of toxicity mediated by lithium and by galactose metabolism disturbances.

HAC1 transcript splicing analysis
Total RNA was extracted as previously described (3) from cells grown to the exponential phase (ϳ0.3 A 600 nm ) under the conditions described in each experiment. A total of 1 g of total RNA was used to prepare the first-strand cDNA using a highcapacity cDNA reverse-transcription kit following the manufacturer's protocol (Applied Biosystems, Foster City, CA). The cDNA was used as a template for the amplification of HAC1 cDNA by PCR. The PCR conditions were 94°C for 1 min followed by 35 cycles at 94°C for 30 s, 62°C for 30 s, 72°C for 90 s, and a final step of 72°C for 10 min. The primers used in these reactions were HAC1F (5Ј-GACCACGAAGACGCGTTG-3Ј) and HAC1R (5Ј-TCAAATGAATTCAAACCTGACTG-3Ј).
The percentage of HAC1 mRNA in the spliced form was quantified as previously described (3) using ImageJ software.

Plate growth assay
The cells were grown overnight, and by using a replica plater for 96-well plates (Sigma R2508), serial dilutions of the cultures (prepared in sterile-distilled water to A 600 nm values of 0.3, 0.03, and 0.003) were transferred to plates containing the indicated medium in the figure legends. The plates were incubated at 30°C for 3 days and then photographed. The images were processed using the Adobe Photoshop software.

Measurement of intracellular galactose and galactose-1phosphate levels
Yeast cultures were inoculated in YPGal and maintained at 30°C with agitation (200 rpm) until an A 600 nm of ϳ 0.2 was reached. At this point, lithium chloride (30 mM) was added to the media, and after 1 h, the cells were collected and processed as previously described (9).

Respiration rate in vivo
The cells were grown at YPGal or YPGly until an A 600 nm of 0.5 was reached and then collected by centrifugation. The cells were suspended at their respective medium to achieve a concentration of 0.1 A 600 nm , and their oxygen consumption rates were measured using the O2k oxygraph of high resolution (Oroboros) at 30°C. After the total oxygen consumption rate was measured, antimycin A was added (final concentration used was 1 M) to estimate the mitochondrial (antimycin A-sensitive) oxygen consumption rate.

Respiration rate in vitro
Yeast cells were grown in YPGly medium up to mid-log phase and ϳ100 A 600 nm of cells were harvested by centrifugation Ubx4p regulates yeast mitochondrial respiration (3,000 ϫ g, 5 min) for mitochondria isolation as described previously (67), using 100 mg/ml of the Lallzyme MMX enzyme for cell wall digestion (Lallemand/GER) instead of Zymolyase. After isolation, mitochondria protein content was determined by the Lowry method, and the amount relative to 200 g of proteins of the mitochondria preparation was added to the Oroboros O2k chamber containing respiration medium (0.6 M mannitol, 10 mM K 2 HPO 4 , 2 mM MgCl 2 , 20 mM Hepes, pH 6.8). Glycerol-3-phosphate (30 mM), pyruvate (10 mM), malate (5 mM), and succinate (20 mM) salts were added to load the electron transport chain of isolated mitochondria, and the oxygen consumption rate was determined (nonphosphorylation state of respiration). Then the oxygen consumption rate was measured in the presence of an excess (1 mM) of ADP (phosphorylation state of respiration), and the respiratory control ratio was calculated between the two states of respiration. After the total oxygen consumption rates were measured in the two states of respiration, antimycin A was added (1 microM) to estimate the mitochondrial (antimycin A-sensitive) oxygen consumption rate. Finally, cytochrome c (0.5 M) was added to inform the integrity of the membranes of the mitochondria preparation.

C. elegans experiments
Worms were grown at 20°C as previously described (68,69). We used the WT N2 and asps-1(ok915) V/RB994 strains. For the oxygraphy experiments, the worms were grown in Escherichia coli OP-50 as the food source until day 0 of adulthood at a density of ϳ12,000 worms/plate (90 ϫ 15 mm). Adult worms at day 0 of adulthood were washed out from plates with MilliQ water and transferred to microtubes at a final density of 2,000 -7,000 worms/biological replicate. The worms were washed two additional times with MilliQ water to remove bacteria. The water was removed, and M9 buffer was added. Oxygen consumption rate was measured at O2K oxygraph of high resolution (Oroboros) at 20°C in M9 buffer. After oxygen consumption measurement, sodium azide was added (final concentration used was 10 mM) to estimate the organism mitochondrial oxygen consumption rate. At the end of the assay, the worms were transferred to a safe-lock tube; supernatant was removed after decantation of the worms, and the tubes were quickly frozen in liquid nitrogen. For protein extraction, 500 l of radioimmune precipitation assay buffer was added to frozen samples and worms lysed using Bullet Blender Tissue Lyzer (two cycles of 3 min at speed 7). The samples were kept on ice for 30 min, with gentle mixing every 10 min, and centrifuged at 14,000 ϫ g for 30 min at 4°C. The supernatant containing proteins was diluted 10ϫ and used for protein quantification by BCA protein assay kit (Pierce). Oxygen consumption rates were normalized by protein levels.

Statistical analysis
All data are presented as means Ϯ S.D. or Ϯ S.E. We used GraphPad Prism version 6.00 for Windows (GraphPad Software) for all analysis in this work. The specific tests and post hoc tests are indicated in the figures legends.