Distinct Functions of Evolutionary Conserved MSF1 and Late Embryogenesis Abundant (LEA)-like Domains in Mitochondria*

Background: PRELID1 homologs are highly conserved mitochondrial proteins that are poorly characterized. Results: PRELID1 homologs in yeast (UPS1–3) are essential for maintaining multiple mitochondrial processes. Conclusion: Ups2p exerts pleotropic effects on mitochondrial processes through distinct roles of the MSF1 and a newly identified LEA-like domain. Significance: Ups2p is the first cytoprotective LEA-like domain-containing protein identified in yeast. PRELID1, the only late embryogenesis abundant (LEA) domain-containing protein in humans, exerts cytoprotective effects through its LEA domain within the mitochondria. Although PRELID1 homologs in vertebrates contain the LEA domain, homologs in lower eukaryotes are thought to lack this domain. In this study, we identify a novel LEA-like domain in a yeast PRELID1 homolog, Ups2p, which contains sequence conservation with the LEA domain of human PRELID1. PRELID1 homologs, including Ups2p, are known to contain the PRELI/MSF1 domain. Our study reveals that the MSF1 domain of Ups2p maintains proper mitochondrial electron transport chain function, respiratory competency, and mitochondrial phosphatidylethanolamine metabolism. The Ups2p MSF1 domain mediates cardiolipin depletion in the absence of Ups1p. However, the Ups2p LEA-like domain is responsible for cardiolipin depletion resulting from UPS2 overexpression. The regulation of phosphatidylethanolamine levels by the MSF1 domain is antagonized by the Ups2p LEA-like domain. We demonstrate that the yeast LEA-like domain protects cells from oxidative stress and can be functionally replaced by the human LEA domain. Together our studies suggest distinct roles of MSF1 and LEA-like domains in mitochondrial function and resistance to oxidative stress.

triple UPS deletion mutants were created using PCR-mediated one-step gene replacement (22) utilizing the HIS3 and LEU2 auxotrophic markers from the pRS413 and pRS415 vectors (23), respectively. The ups2-⌬181 one-step allelic replacement cassette was constructed by PCR amplification of the ups2-⌬181-p426ADH plasmid. Primers were designed to create UPS2-homologous regions at the 5Ј and 3Ј ends of the replacement cassette comprised of the UPS2 truncation allele and the URA3 selectable marker (24). The cassette was used for allelic replacement in wild type and ups1⌬ cells, resulting in the ups2-⌬181 and ups1⌬ups2-⌬181 strains, respectively. A similar strategy was implemented for creation of the ups2-⌬210 and ups2-⌬181ϩhLEA strains. Primers used for the creation of these strains are listed in supplemental Table S1. Yeast transformations of plasmids and PCR cassettes were performed using the Frozen-EZ yeast transformation II kit (Zymo Research, Orange, CA). The strains were grown in 1% yeast extract and 2% peptone with either 2% dextrose (YPD); 2% glycerol, 2% lactic acid, pH 6.0 (YPGL); or 2% glycerol, 2% ethanol (YPGE). Plasmids in transformed yeast were maintained by growth in selective, synthetic medium lacking uracil (SDC-uracil or SGLC-uracil).
Plasmid Construction-PRELID1-p426ADH was constructed by PCR amplification of the open reading frame from PRELID1-pOTB7 (accession number BC007268; ATCC) followed by digestion and ligation into the EcoRI site of p426ADH (25). UPS2 was subcloned into the SpeI/XhoI sites of p426ADH from the UPS2-pEGH plasmid obtained from Open Biosystems (Huntsville, AL) (26,27). UPS1 and ups2-⌬181 were cloned into the HindIII and XhoI sites of p426ADH. The ups2-⌬181 truncation allele was constructed by introducing a nonsense mutation of Ser-182, which removed the LEA-like-containing C terminus, but left a functional MSF1 domain. The ups2-⌬210 truncation allele was generated from the UPS2-p426ADH plasmid via sitedirected mutagenesis, resulting in a Ups2 protein truncated after the LEA domain (G211X) (28). UPS2 alleles were verified by sequencing the entire UPS2 open reading frame. The ups2-⌬181ϩhLEA chimeric gene was created in the p426ADH vector by first subcloning UPS2 (from UPS2-p426ADH) into the SpeI site upstream of PRELID1 in the PRELID1-p426ADH plasmid and subsequently removing nucleotides between the Ups2p MSF1 domain and the human PRELID1 LEA domain, resulting in an in-frame yeast/human chimeric gene, as previously described (28). Primers used for plasmid creation are listed in supplemental Table S1.
Spontaneous Petite Formation-The frequency of spontaneous petite formation in the BY4741 background was determined as previously described (29). Briefly, overnight YPD cultures of single colonies were plated onto both YPD and YPGL medium and grown for 3 and 5 days, respectively. The percent of respiratory incompetent cells was calculated by quantifying the difference of colony-forming units between the two media. For transformants, the cells were grown under selection in SGLC medium lacking uracil to saturation. These cultures were then inoculated at an A 600 of 0.05 in SDC medium lacking uracil, grown for ϳ7 generations (2 days), diluted, and plated onto both YPD and YPG medium.
Mitochondrial Electron Transport Chain Complex Activities-Crude mitochondrial fractions were prepared from yeast grown to midlog phase in YPGL medium. The cells were disrupted, and mitochondrial fractions were obtained via differential centrifugation, as previously described (30,31). The mitochondrial pellets were then resuspended in 20 mM KP i buffer, pH 7.4, followed by three freeze-thaw cycles at Ϫ80°C (32). Protein concentration was determined using the Bradford protein assay and then diluted to 250 g/ml in KP i pH 7.4 buffer. Antimycin A-sensitive NADH-coupled cytochrome c reductase (NCCR) activity was measured as described by Powers et al. (33) with minor modifications. Briefly, 2.5 g of the mitochondrial fraction were solubilized in complex activity buffer (25 mM KP i buffer, pH 7.2, 5 mM MgCl 2 , 2 mM KCN, 2.5 mg/ml bovine serum albumin, 0.5 mM n-dodecyl-␤-maltoside) supplemented with 1.4 mM NADH. The rate of cytochrome c (50 M) reduction was measured (⌬Abs 550 ) via complex I oxidation of NADH before and after the addition of 2 g/ml antimycin A. Ubiquinol-coupled cytochrome c reductase (QCCR), succinate dehydrogenase (SDH), and cytochrome c oxidase (COX) activities were measured as described previously by Birch-Machin et al.  (32). Citrate synthase activity was measured as previously described by Barrientos et al. (34). Growth Assays-Yeast overexpressing UPS2 from UPS2-p426ADH were grown under selection in SGLC medium lacking uracil. Saturated cultures were used to inoculate fresh medium at an A 600 of 0.1. The A 600 absorbance readings were measured at 3-h intervals. Doubling times during exponential growth were then derived from an exponential best fit curve using Microsoft Excel (Redmond, WA). Sensitivity to oxidative stress was determined by the ability of the strains to grow in the presence of menadione sodium bisulfite, as previously described (35). Strains grown in nonfermentable YPGE medium overnight were diluted to an initial A 600 of 0.1 in YPGE medium with or without supplementation of menadione. At 48 h, the A 600 of the cultures were measured. Sensitivity to menadione was also determined by growth on solid medium. Overnight cultures were diluted to 2.4 ϫ 10 7 cells/ml (A 600 ϭ 0.8). 10-fold serial dilutions (10 l) were spotted onto solid YPGE medium containing menadione. The plates were visualized after 7 days of growth.
RT-PCR-Total RNA from transformed yeast strains under uracil selection was isolated with TRIzol reagent (Invitrogen) and treated with amplification grade DNase I (Sigma). RNA was reverse-transcribed using SuperScript III first-strand synthesis system for RT-PCR (Invitrogen). Expression of UPS2 and ACT1 was assessed by semi-quantitative PCR using the resulting cDNA template. Amplification of 0.4 ng of cDNA with Taq 2ϫ Master Mix (New England Biolabs) PCR reagent was conducted as follows: UPS2 amplicon (annealing temperature 50°C; 28 amplification cycles) and ACT1 amplicon (annealing temperature 55°C; 25 amplification cycles). The primer sequences are provided in supplemental Table S1.
Phospholipid Analysis-Non-transformed strains were grown in YPGE medium and transformed strains were grown while being maintained under selection in SDC medium lacking uracil. Phospholipids were extracted from the crude mitochondrial fractions of exponentially growing yeast, as described with minor modifications (36). Phospholipid quantification was determined by complexation with ammonium ferrothiocyanate in chloroform (37). Equal amounts of lipid were spotted onto 1.8% boric acid (in 100% ethanol)-impregnated high performance thin layer chromatography plates (Merck). The lipids were resolved twice using a developing solution of chloroform/ ethanol/water/triethylamine (30:35:7: (38). Phospholipids were visualized by charring the plates at 160°C for 20 min after treatment with 470 mM CuSO 4 in 8.5% o-phosphoric acid (5). CL and PE standards were obtained from Avanti Polar Lipids, Inc. (Alabaster, AL). Densitometric analysis of phospholipids was performed using ImageJ software (National Institutes of Health, Bethesda, MD).
Statistical Analyses-The results are expressed as averages with error bars representing the S.E. Statistical significance for mitochondrial complex activities was determined by analysis of variance using a generalized linear model procedure. In other figures, significance was determined by unpaired two-tailed Student's t test with equal variance using Microsoft Excel (Redmond, WA).

RESULTS
Yeast PRELID1 Homologs Maintain Mitochondrial Respiratory Function-Saccharomyces cerevisiae contains three PRE-LID1 homologs (UPS1-3). The role of Ups proteins in mitochondrial respiratory function has not been investigated. To determine the effects of UPS inactivation on mitochondrial respiration, spectrophotometric assays were performed to measure the activities of mitochondrial electron transport chain (mtETC) complexes. The steady-state activities of NCCR (complex I ϩ III), QCCR (complex III), SDH (complex II), and COX (complex IV) were measured in mitochondria isolated from wild type cells and UPS deletion mutants ups1⌬, ups2⌬[r], ups3⌬, ups1⌬ups2⌬, ups1⌬ups3⌬, ups2⌬ups3⌬, and ups1⌬ ups2⌬ups3⌬ (Fig. 1). NADH dehydrogenase activity of the mtETC was ascertained by comparing the relative activities of NCCR and QCCR (Fig. 1, A and B). In ups1⌬ cells, a decrease was observed in both NCCR and QCCR activity compared with the wild type strain. These results suggest a defect in steadystate levels of complex III activity in ups1⌬ cells. The additional inactivation of UPS3 (in ups1⌬ups3⌬ cells) reverses the defects in both NCCR and QCCR. Notably, the absence of Ups2p function (in ups2⌬, ups1⌬ups2⌬, ups2⌬ups3⌬, and ups1⌬ups2⌬ ups3⌬ cells) decreased NCCR activity by 35-50% in the absence of significant changes in QCCR activity. This indicates that Ups2p function influences the steady-state levels of yeast NADH dehydrogenase activity. NCCR and QCCR activities were unaltered in ups3⌬ cells compared with wild type cells.
The activity of SDH was decreased in ups1⌬ cells (Fig. 1C). However, deletion of UPS2 or UPS3 (ups1⌬ups2⌬ and ups1⌬ups3⌬ cells) reversed this defect. The ups1⌬ups2⌬ double mutant exhibited a slight increase in SDH activity compared with the wild type strain, whereas ups1⌬ups3⌬ cells exhibited wild type levels. The SDH activities of other UPS deletion mutants were unaffected.
Inactivation of UPS1 resulted in a 1.8-fold increase in the steady-state levels of COX activity (Fig. 1D). Similar COX activities were observed between ups1⌬ and ups1⌬ups3⌬ strains. The increased COX activity observed in ups1⌬ cells were reversed to wild type levels upon inactivation of UPS2 (ups1⌬ups2⌬), despite the slight increase in COX activity observed in the single UPS2 deletion mutant (ups2⌬). The triple UPS deletion mutant showed an increase in COX activity over wild type levels, comparable with that observed in ups2⌬ In summary, inactivation of UPS1 or UPS2 led to differential effects on steady-state levels of mtETC activities. In the absence of Ups1p function, mitochondria exhibited Ups2p-and Ups3pdependent decreases in SDH and QCCR activities and a Ups2pdependent increase in COX activity. In the absence of Ups2p function, a defect in NCCR activity emerged.
The MSF1 Domain of Ups2p Is Required for Optimal NCCR Activity-To determine whether the MSF1 domain is sufficient to maintain steady-state levels of NCCR activity, we created a UPS2 truncation allele, ups2-⌬181, which encodes only the MSF1 domain of Ups2p. Expression of either full-length Ups2p (UPS2) or the Ups2p MSF1 domain (ups2-⌬181) in ups2⌬ cells restored NCCR activity to levels observed in wild type cells, indicating that MSF1 domain function maintains optimal NCCR activity during respiratory growth ( Fig. 2A). Expression of human PRELID1 also restored NCCR activity in ups2⌬ cells.  QCCR activities in these strains were unaffected (Fig. 2B). Additionally, no significant differences were observed in mitochondrial citrate synthase activity, confirming that equal mitochondrial masses were analyzed in the mtETC activity assays (Fig. 2C).
The MSF1 Domain of Ups2p Maintains Respiratory Competence-To determine to role of Ups function in the maintenance of respiratory competency, the frequency of spontaneous petites (respiratory-incompetent cells) was measured in UPS deletion strains (Fig. 3A). Inactivation of the UPS2 gene led to an increase in petite frequency. The ups2⌬ single deletion mutant displayed frequencies ϳ3-fold higher than observed in wild type cells. Furthermore, in both double deletion mutants where UPS2 is inactivated (ups1⌬ups2⌬ and ups2⌬ups3⌬), the petite frequency was elevated to levels observed in the ups2⌬ strain. Inactivation of the UPS1 gene decreased the observed petite frequency. The additional inactivation of UPS2, but not UPS3, reverses the decreased petite frequency in ups1⌬ cells, indicating that the observed decrease is dependent upon Ups2p function but not Ups3p function. These results demonstrate that Ups2p maintains respiratory competence. However, inactivation of all UPS genes (in ups1⌬ups2⌬ups3⌬ cells) caused a synergistic increase in the frequency of petites (Fig. 3A), demonstrating that the Ups proteins are all required for maintaining respiratory competence.
To assess the contribution of the Ups2p MSF1 domain to maintaining respiratory competence, spontaneous petite frequency was measured in ups2⌬ cells exogenously expressing either full-length Ups2p (UPS2) or the Ups2p MSF1 domain (ups2-⌬181). The expression of full-length Ups2p was able to reverse the petite formation in the UPS2 deletion strain to wild type levels (Fig. 3B). Expression of the Ups2p MSF1 domain (ups2-⌬181) was also able to rescue this defect, indicating that the maintenance of respiratory competence is dependent upon MSF1 domain function. The rescue of the petite formation defect by expression of the Ups2p MSF1 domain and the inability of human PRELID1 expression to suppress this defect were reconfirmed in a second genetic background (data not shown).
Evolutionarily Conserved LEA-like Domain and Its Role in Oxidative Stress Resistance-Human PRELID1 is comprised of two domains, an N-terminal MSF1 domain and a C-terminal LEA domain (8,9). Although the MSF1 domain is common to all PRELID1 homologs, only vertebrate homologs contain a LEA domain (Fig. 4A). All three yeast PRELID1 homologs (Ups1p, Ups2p, and Ups3p) contain an MSF1 domain (Fig. 4B) (4,5,17). Our analysis revealed that, unlike Ups1p and Ups3p, Ups2p contains a unique, uncharacterized C terminus (Fig. 4B). Further analysis of protein sequences revealed that the C termini of yeast Ups2p and human PRELID1 align, exhibiting 40% sequence identity across the LEA domain (Fig. 4C). Conservation of these residues can also be seen in the Ups2p homologs of other yeast and insect species, comprising a LEA-like domain ( Fig. 4D and supplemental Fig. S1).
LEA-containing proteins were recently described to protect against ROS-producing insults in plant and animal species (9, 39 -41). To assess whether the conserved LEA-like domain of Ups2p performs a similar protective role in yeast, wild type, ups2⌬, and ups2-⌬181 strains were exposed to superoxide-producing menadione during growth in respiratory medium. No differences were observed between the growth of wild type and ups2⌬ cells under oxidative stress; however, ups2-⌬181 cells showed significant growth impairment (Fig. 5A). Treatment with menadione-inhibited growth of ups2-⌬181 cells, indicating that the LEA-like domain-containing C terminus of Ups2p provides oxidative stress resistance in these cells. To determine whether conserved LEA-like residues (Fig. 4C) play a role in  NOVEMBER 11, 2011 • VOLUME 286 • NUMBER 45

JOURNAL OF BIOLOGICAL CHEMISTRY 39145
protection against oxidative stress, the non-LEA-like sequence was truncated from the C terminus of Ups2p. This strain, ups2-⌬210, was not sensitive to menadione when compared with the ups2-⌬181 strain (Fig. 5A). These data demonstrate that the LEA-like sequence is required for protection against oxidative stress. The effects of menadione treatment observed during growth in liquid medium were recapitulated on solid medium (Fig. 5B).
To further characterize the effects of menadione on the ups2-⌬181 strain, we measured colony-forming units of treated cultures. We determined that cell death and petite formation did not significantly contribute to the optical density of ups2-⌬181 cultures (data not shown), revealing that menadione is cytostatic to ups2-⌬181 cells. On solid medium, the growth rate of ups2-⌬181 cells, not survival, was affected by menadione exposure; this became apparent when plates were incubated for additional time (data not shown). To determine whether this cytostatic effect was superoxide-specific, strains were also treated with tert-butylhydroperoxide (supplemental Fig. S2). In the presence of tert-butylhydroperoxide, the UPS2 mutant strains exhibited wild type levels of growth, indicating a superoxide-specific sensitivity in ups2-⌬181 cells.
We also tested whether the LEA domain of human PRELID1 could functionally replace the Ups2p LEA-like domain. A strain was generated (ups2-⌬181ϩhLEA) that expresses a chimeric protein containing the Ups2p MSF1 domain (encoded by the ups2-⌬181 allele) fused to the LEA-containing C terminus of human PRELID1 (ϩhLEA). This strain exhibited resistance to menadione similar to the ups2-⌬210 strain (Fig. 5, A and B). Together, these data suggest that the conserved Ups2p LEAlike domain is responsible for the protective effect conferred by the human LEA domain in the chimeric strain.
LEA-like Dependent Regulation of Cardiolipin Levels in UPS2-overexpressing Cells-Osman et al. (5) demonstrated that overexpression of UPS2 generates a defect in CL metabolism. To assess the contribution of the Ups2p MSF1 and LEAlike domains to CL metabolism, we overexpressed full-length Ups2p (UPS2) or the Ups2p MSF1 domain (ups2-⌬181) in ups2⌬ cells and measured CL levels. Exogenous expression increased transcripts of the UPS2 alleles greater than 20-fold over wild type levels (Fig. 6A).
The levels of mitochondrial phospholipids in these strains were determined. The ups1⌬ strain transformed with empty vector (ups1⌬ [p426ADH]) displayed a characteristic decrease in CL levels (Fig. 6B, panel I). As expected, the ups2⌬ [p426ADH] control strain exhibited wild type levels of CL. However, the decrease in PE levels associated with UPS2 deletion was not observed in these cells. Although the reason for this is unclear, it may be related to growth in minimal medium (SDC lacking uracil), which was required to maintain the p426ADH plasmids. The overexpression of full-length Ups2p (ups2⌬ [UPS2-p426ADH]) resulted in CL depletion (Fig. 6B,  panel I). CL levels in this strain were decreased ϳ40% (Fig. 6B,  panel II). Depletion of CL was not observed in the strain expressing the Ups2p MSF1 domain (ups2⌬ [ups2-⌬181-p426ADH]) (Fig. 6B, panel I and II). In these cells, the loss of the Ups2p LEA-like domain alleviates CL depletion caused by Ups2p overexpression. These results demonstrate that the LEA-like domain of Ups2p negatively regulates CL levels.
To determine the effect of UPS2 expression on growth, doubling time was calculated in strains that expressed full-length Ups2p at endogenous levels (WT [p426ADH]) or at elevated expression levels (ups2⌬ [UPS2-p426ADH]). The overexpression of Ups2p increased doubling time by ϳ50 min compared with the wild type strain (Fig. 6C). However, cells overexpressing the Ups2p MSF1 domain (ups2⌬ [ups2-⌬181-p426ADH]) exhibited wild type rates of growth. In these cells, the loss of the Ups2p LEA-like domain alleviates growth inhibition caused by Ups2p overexpression. Together, these data demonstrate that the concomitant decreases in CL levels and growth rate are dependent upon the LEA-like domain of Ups2p.
LEA-like Independent Regulation of Cardiolipin Levels in the Absence of Ups1p-The presence of Ups2p is responsible for CL depletion in the absence of Ups1p (4,5). We sought to determine whether the regulation of CL levels is dependent on the LEA-like domain, as we observed when Ups2p was overex- pressed. To address this question, mitochondrial CL levels were measured in the wild type strain, UPS deletion mutants lacking Ups1p and/or Ups2p (ups1⌬, ups2⌬, and ups1⌬ups2⌬ strains) and in genomically modified strains where the UPS2 wild type allele was replaced with the ups2-⌬181 truncation allele (ups2-⌬181 and ups1⌬ups2-⌬181 strains). Consistent with previous studies, ups1⌬ cells show a decrease in CL, whereas ups2⌬ and ups1⌬ups2⌬ strains contain wild type levels (Fig. 7). Our data demonstrate that the Ups2p LEA-like domain does not affect CL levels in the presence or absence of Ups1p, as observed in ups2-⌬181 and ups1⌬ups2-⌬181 strains, respectively. This indicates that the function of the Ups2p MSF1 domain, not the LEA-like domain, is responsible for the loss of CL in the absence of Ups1p.

DISCUSSION
PRELID1 homologs are highly conserved mitochondrial proteins. In this study, the functions of yeast PRELID1 homologs were investigated by determining the effects of UPS gene inactivation on mtETC complex activity and respiratory competence. The MSF1 domain of Ups2p was essential to its role in maintaining these processes. In addition to the MSF1 domain, we identified a LEA-like domain in Ups2p that exhibits a high degree of sequence conservation with the LEA domain of human PRELID1. Additionally, a conserved function of the yeast LEA-like domain in oxidative stress resistance was discovered. We also demonstrate that both the MSF1 and LEAlike domains contribute to mitochondrial phospholipid metabolism.
The three Ups proteins contain MSF1 domains (Fig. 4B), yet their sequences are divergent. The sequence of Ups1p differs more from Ups2p and Ups3p (30 and 25% identical, respectively), which are more similar in sequence (57% identical). Fur-thermore, Ups2p differs from its other two homologs in that it contains an additional C-terminal domain (Fig. 4B), which we have characterized as a LEA-like domain. Expression of the Ups2p MSF1 domain rescues the NCCR defect (Fig. 2) and petite formation defect (Fig. 3B) in ups2⌬ cells. Additionally, cells endogenously expressing the Ups2p MSF1 domain are able to maintain mitochondrial PE to levels slightly higher than in wild type cells (Fig. 7). These results demonstrate that the MSF1 domain is functional in the absence of the LEA-like domain-containing C terminus. Furthermore, it is the MSF1 domain, not the LEA-like domain, that is required for these functions of Ups2p. Thus, whereas all Ups proteins have an MSF1 domain, the sequence variation of this domain provides unique Ups protein functions.
The Ups proteins differentially influence the steady-state activity of mtETC complexes (Fig. 1). In ups1⌬ cells, NCCR, QCCR, and SDH activities are negatively affected (Fig. 1, A-C), and COX activity is increased (Fig. 1D). In the absence of Ups2p function, NCCR activity is negatively affected (Fig. 1A). Because NCCR (complex I ϩ III) activity is dependent upon QCCR (complex III) activity, the NCCR defect in ups1⌬ cells may result from the underlying defect in QCCR activity (Fig. 1B). In cells lacking Ups2p, the NCCR defect is not associated with a decrease in QCCR activity, revealing a defect in mtETC-associated NADH dehydrogenase activity. The SDH and QCCR defects in ups1⌬ cells are dependent upon both Ups2p and Ups3p function. These data suggest that, as previously reported for Ups2p (4), Ups3p may also antagonize Ups1p function. Inactivation of UPS1 also leads to an increase in the steady-state levels of COX activity (Fig. 1D). This alteration in COX activity is reversed in ups1⌬ cells upon inactivation of UPS2, but not upon inactivation of UPS3. Because CL synthesis and COX assembly are interdependent (42), the increase in COX activity in CL-deficient ups1⌬ and ups1⌬ups3⌬ cells was unexpected (Fig. 1D). However, a study utilizing a mutant of PGS1 (the enzyme catalyzing the rate-limiting step of CL synthesis) in CHO cells (43,44) recapitulates our CL (Fig. 7) and COX activity (Fig. 1D) data in ups1⌬ cells. Ups1p function may influence Pgs1p activity in yeast.
Inactivation of UPS1 results in a Ups2p-dependent decrease in the frequency of respiratory incompetence (Fig. 3A). The petite assay performed produces negative values for the calculation of petite frequency in ups1⌬ and ups1⌬ups3⌬ mutants, because more colony-forming units appeared on plates possessing a respiratory carbon source (glycerol/lactate) than a fermentable carbon source (glucose). This result demonstrates that these cells are better able to form colonies on medium requiring mitochondrial function for growth. The majority of UPS1 deletion phenotypes (including slowed growth, mitochondrial morphology, intramitochondrial protein import and processing, and mitochondrial membrane potential) occur in fermentable medium but are reversed upon growth in respiratory medium (4). It is known that growth of ups1⌬ and ups1⌬ups3⌬ cells exhibit slowed growth on glucose medium but not glycerol medium (4). The extent of this defect is not known, but our data suggest that the fermentable carbon source-dependent mitochondrial dysfunction limits the ability of some cells to form visible colonies.
Inactivation of UPS2 results in an increased frequency of respiratory incompetence (Fig. 3). Aside from nuclear-encoded mitochondrial proteins that exhibit an absolute requirement for mitochondrial respiration (characterized by complete respiratory incompetence in the presence of functional mtDNA), petite formation occurs because of mtDNA instability and/or complete mtDNA loss (producing Ϫ or 0 cells) (45,46). The UPS2 deletion mutant was previously described to exhibit an increase in mtDNA loss (5). Thus, Ups2p protects against petite formation through its ability to influence mitochondrial genome maintenance and stability. In both rich and synthetic medium, the loss of Ups2p function resulted in ϳ3-fold increase in petite frequency (Fig. 3, A and B, respectively). This role of Ups2p is provided by the Ups2p MSF1 domain (Fig. 3B). Expression of human PRELID1, the UPS1 ortholog, was unable to suppress the petite defect in ups2⌬ cells, exhibiting functional divergence of the Ups2p and PRELID1/Ups1p MSF1 domains.
It is known that disruption of mitochondrial PE synthesis, but not CL synthesis, increases the rate of petite formation (47)(48)(49). The ability of the Ups2p MSF1 domain to maintain both PE levels (Fig. 7) and respiratory competency (Fig. 3B) suggests that petite formation in cells lacking Ups2p is an indirect effect of PE depletion. The inability of human PRELID1 expression to suppress petite formation in the ups2⌬ strain (Fig. 3B) is in accordance with previous work demonstrating that PRELID1 expression failed to suppress PE depletion in the ups2⌬ strain (50). The synergistic increase in respiratory incompetence in the UPS triple deletion mutant reveals that, in the absence of Ups2p, Ups1p and Ups3p function protects against petite formation (Fig. 3A). This demonstrates that the mitochondrial functions of the conserved Ups MSF1 domains are required to maintain respiratory competence. The synergistic increase in petite formation may occur by influencing factors required for mtDNA maintenance, which interact synergistically with PE depletion.
We determined that the Ups2p MSF1 domain maintains respiratory competence (Fig. 3), mtETC-associated NADH dehydrogenase activity (Fig. 2), and mitochondrial PE levels (Fig. 7). Although a connection can be made between petite formation and PE levels, the mechanism by which NCCR activity (specifically, the NADH dehydrogenase component) is affected by UPS2 inactivation is less clear (Fig. 1A). Maintenance of NADH dehydrogenase activity appears to be a conserved role of the PRELI/MSF1 family of proteins, because expression of human PRELID1 suppresses the NADH dehydrogenase defect in ups2⌬ cells. Indeed, McKeller et al. (9) has shown that expression of PRELID1 in human cells leads to a substantial increase in respiratory function through increased NADH dehydrogenase activity, corresponding to our findings presented here. The ability of human PRELID1 expression to suppress the NADH dehydrogenase defect in ups2⌬ cells (Fig. 2) but not suppress the increased petite formation (Fig. 3B) or PE depletion (50) phenotypes suggests that the defect in NADH dehydrogenase activity does not result from PE depletion.
Cells lacking the LEA-like domain (expressing the Ups2p MSF1 domain only) were sensitive to oxidative stress (Fig. 5). However, the UPS2 deletion mutant did not exhibit sensitivity to oxidative stress compared with wild type cells. Thus, in the presence of Ups2p MSF1 domain function, the LEA-like domain of Ups2p is required for protection against oxidative stress (Fig. 5). Although there may be other possible explanations of this data, we propose that the LEA-like domain provides protection against oxidative stress through regulation of the MSF1 domain. Further evidence of MSF1 domain regulation by the LEA-like domain is provided through investigation of the role of Ups2p in mitochondrial phospholipid metabolism. Consistent with published data (4 -7), we demonstrate that mitochondrial PE levels are decreased in ups2⌬ cells (Fig.  7). However, in strains lacking the Ups2p LEA-like domain (ups2-⌬181 and ups1⌬ups2-⌬181), PE levels are increased compared with the wild type strain. This indicates that the LEA-like domain negatively regulates maintenance of mitochondrial PE levels by the Ups2p MSF1 domain.
Recently, LEA-containing proteins have been identified as essential cytoprotective proteins in nonplant species (9,13,39,51). In this study, we elucidate a conserved role for the LEA-like domain of Ups2p in oxidative stress resistance in yeast (Fig. 5). Consistent with our findings, recently identified mitochondriatargeted LEA proteins were shown to protect the TCA cycle and mitochondrial respiratory enzymes against reactive oxygen species-inducing water stress in plant and animal models (39,40). Furthermore, we demonstrate that the LEA domain of human PRELID1 can functionally replace the LEA-like domain in yeast and may provide cytoprotection through mechanisms similar to other mitochondria-targeted LEA-containing proteins. The LEA domain of human PRELID1 promotes mitochondrial respiration and protection against apoptosis (9). Because CL is required for mtETC supercomplex formation and provides apoptotic resistance (52, 53), PRELID1 function may be linked to CL. Recently, a mitochondrial LEA protein was shown to physically interact with CL (54). These data suggest that PRELID1 and possibly other LEA/LEA-like-containing proteins, may exert cytoprotective effects through interaction with mitochondrial phospholipids and, thus, through mitochondrial function.
A direct competition between Ups1p and Ups2p has been proposed, supported by the observations that (i) the CL depletion resulting from the loss of UPS1 is Ups2p-dependent and (ii) UPS2 overexpression results in a decrease in CL, a phenotype similar to UPS1 deletion (5,6). Our data reveal that the LEA-like domain is responsible for the loss of CL upon Ups2p overexpression (Fig. 6B), yet the MSF1 domain of Ups2p is responsible for the loss of CL in the absence of Ups1p (Fig. 7). Thus, by investigating the contributions of the MSF1 and LEAlike domains, we conclude that the decreases in CL levels in these two studies occur through different mechanisms. Our findings suggest that Ups2p does not directly antagonize Ups1p as proposed (5,6). This is supported by the observation that the Ups2p MSF1 domain (which is responsible for the CL depletion in the absence of Ups1p) does not lead to CL depletion upon overexpression (Figs. 6B and 7). Potting et al. (6) demonstrated that the UPS2 overexpression phenotypes (CL depletion and slowed growth) are Mdm35p-independent. It is unclear whether the UPS2 overexpression phenotypes result from biologically relevant Ups2p interactions.
In summary, we elucidated roles of the MSF1 and LEA-like domains of Ups2p in maintenance of mitochondrial function. By determining the contributions of the MSF1 and LEA-like domains to the regulation of mitochondrial PE and CL levels, we demonstrate that Ups2p may not directly antagonize Ups1p. Our studies suggest that the LEA/LEA-like domain is functionally conserved in yeast and humans, providing protection against oxidative stress. Further study of the Ups2p LEA-like domain may provide insight into mechanisms of protection conferred by mitochondrial LEA proteins across eukaryotes.