Ethanolamine ameliorates mitochondrial dysfunction in cardiolipin-deficient yeast cells

Cardiolipin (CL) is a signature phospholipid of the mitochondria required for the formation of mitochondrial respiratory chain (MRC) supercomplexes. The destabilization of MRC supercomplexes is the proximal cause of the pathology associated with the depletion of CL in patients with Barth syndrome. Thus, promoting supercomplex formation could ameliorate mitochondrial dysfunction associated with CL depletion. However, to date, physiologically relevant small-molecule regulators of supercomplex formation have not been identified. Here, we report that ethanolamine (Etn) supplementation rescues the MRC defects by promoting supercomplex assembly in a yeast model of Barth syndrome. We discovered this novel role of Etn while testing the hypothesis that elevating mitochondrial phosphatidylethanolamine (PE), a phospholipid suggested to overlap in function with CL, could compensate for CL deficiency. We found that the Etn supplementation rescues the respiratory growth of CL-deficient Saccharomyces cerevisiae cells in a dose-dependent manner but independently of its incorporation into PE. The rescue was specifically dependent on Etn but not choline or serine, the other phospholipid precursors. Etn improved mitochondrial function by restoring the expression of MRC proteins and promoting supercomplex assembly in CL-deficient cells. Consistent with this mechanism, overexpression of Cox4, the MRC complex IV subunit, was sufficient to promote supercomplex formation in CL-deficient cells. Taken together, our work identifies a novel role of a ubiquitous metabolite, Etn, in attenuating mitochondrial dysfunction caused by CL deficiency.

Mitochondrial membrane phospholipid composition determines the function and formation of the mitochondrial respiratory chain (MRC). 4 Previous work on cardiolipin (CL), a signature phospholipid of mitochondrial membranes, has identified its critical roles in electron transport and energy transformation reactions (1,2). Both in vivo and in vitro experiments have demonstrated that CL exerts its effect on the MRC by facilitating supercomplex formation (3)(4)(5). MRC supercomplexes consist of supramolecular assemblies of the MRC complex III and complex IV in the yeast Saccharomyces cerevisiae (6) and complexes I, III, and IV in higher eukaryotes (7). These respiratory supercomplexes are proposed to stabilize individual MRC complexes, minimize the generation of reactive oxygen species, and enhance catalytic efficiency by substrate channeling (8,9). The recent cryo-EM-based determination of highresolution structures of the respiratory supercomplexes (10 -12) have firmly established their existence and suggested the dynamic nature of their assembly. However, apart from CL, no other metabolite has been known to regulate their formation.
The biomedical relevance of CL in MRC function and supercomplex assembly can be gleaned from the finding that its depletion results in a life-threatening disorder in humans called Barth syndrome (BTHS). BTHS is a severely debilitating X-linked genetic disorder characterized by cardiomyopathy, skeletal muscle myopathy, neutropenia, growth-delay, and exercise intolerance (13). BTHS is caused by loss-of-function mutations in the evolutionarily conserved tafazzin (TAZ) gene (14), which encodes a mitochondrial transacylase that remodels CL (15,16). Studies from BTHS patient cells as well as from a number of model systems, including the yeast S. cerevisiae taz1⌬ cells, have shown that mitochondrial dysfunction caused by perturbation in CL remodeling is the primary cause of BTHS pathology (17)(18)(19)(20)(21)(22). The downstream consequences of this altered mitochondrial phospholipid composition include the destabilization of MRC supercomplexes and increased oxidative stress (23)(24)(25).
In addition to its well-characterized role in MRC supercomplex formation, CL exerts its effect on mitochondrial bioenergetics at multiple levels. For example, CL is required for the expression of both mitochondrial and nuclear DNA-encoded MRC subunits (26,27) and optimal MRC complex IV activity (4). The loss of CL also results in decreased mitochondrial membrane potential and protein import (28). Finally, the loss of CL perturbs efficient coupling of electron transport with ATP synthesis (29,30). Like CL, phosphatidylethanolamine (PE) is also required for optimal mitochondrial bioenergetics by preserving the catalytic activities of the MRC complexes (31)(32)(33), whereas PC has been shown to be redundant as far as MRC function and assembly is concerned (31,34,35). Both PE and CL are synthesized in situ, unlike other mitochondrial phospholipids (Fig. 1). In addition to the mitochondrial pathway, PE can be biosynthesized either by the nonmitochondrial ethanolamine-Kennedy pathway or by a lyso-PE-requiring pathway (Fig. 1).
Because of their common site of biosynthesis, reciprocal regulation of their levels, and the propensity to form nonbilayer structures, PE and CL have been proposed to have overlapping functions (36 -43). These findings formed the basis of our hypothesis that MRC defects due to disruptions in CL biosynthesis can be rescued by elevating mitochondrial PE levels. We have recently shown that exogenous Etn supplementation can increase mitochondrial PE levels (31), allowing us to test the proposed hypothesis. In this study, we used an Etn supplementation strategy in yeast models of CL deficiency to demonstrate that Etn can rescue respiratory chain defects of CL-depleted strains. Surprisingly, Etn-mediated rescue was independent of its incorporation into PE. Thus, our work identifies a novel role of Etn as a regulator of MRC biogenesis.

Ethanolamine supplementation rescues the respiratory growth of CL-deficient cells
Previous reports have shown that CL deficiency in yeast cells is accompanied by an increase in PE levels (19,42,43), suggesting a compensatory response to the lack of CL. Therefore, we asked whether further elevation of mitochondrial PE levels in CL-deficient taz1⌬ and crd1⌬ cells could compensate for the loss of CL and rescue the respiratory growth defect (20). To elevate mitochondrial PE levels, we utilized our recently described Etn supplementation strategy (31) and showed that Etn rescues the respiratory growth of taz1⌬ and crd1⌬ cells in a dose-dependent manner, with an EC 50 of 0.5 and 1.04 mM, respectively (Fig. 2, A-C). Whereas Etn fully rescued the respiratory growth of CL-depleted taz1⌬ cells, it only partially rescued the growth of crd1⌬ cells, which are completely deficient in CL (Fig. 2, A and C). Notably, the rescue was also observed in BY4742 genetic background (Fig. S1).
To confirm that Etn supplementation results in increased mitochondrial PE in taz1⌬ and crd1⌬ cells, we analyzed the mitochondrial phospholipid composition of cells grown with and without Etn supplementation. Consistent with the previous reports (19,42,43), mitochondrial phospholipid analysis of CL-deficient cells revealed that, when compared with WT cells, PE was already elevated by ϳ25 and 50% in taz1⌬ and crd1⌬ cells, respectively (Fig. 2D). Addition of Etn to the growth media further increased mitochondrial PE levels in taz1⌬ cells by 25% (Fig. 2D). The increase in PE was accompanied by a concomitant decrease in PC levels (Fig. 2D). However, in crd1⌬ cells, neither PE nor PC was significantly altered upon Etn supplementation (Fig. 2D). These results indicate that exogenous supplementation of phospholipid precursor Etn leads to increased levels of PE in taz1⌬ but not in crd1⌬ mitochondria. Crd1 synthesizes nascent CL p from PG. The resulting CL p is deacylated by the phospholipase Cld1 to produce monolysocardiolipin (MLCL), which is then reacylated by the transacylase Taz1 to form mature cardiolipin (CL m ). Mutations in the human homologue of TAZ1 result in Barth syndrome. CL biosynthesis depends on the import of PA from endoplasmic reticulum, which is converted to CDP-DAG by Tam41. Pgs1 catalyzes conversion of CDP-DAG to PGP, which is then dephosphorylated to PG by Gep4. PE biosynthesis in yeast can occur by the following: 1) Psd1-catalyzed decarboxylation of PS in the mitochondria; 2) incorporation of Etn via the cytosolic/endoplasmic reticulum Kennedy pathway enzymes Eki1, Ect1, and Ept1, respectively; and 3) the acylation of lyso-PE by Ale1. The Kennedy pathway enzymes Cki1, Pct1, and Cpt1 can utilize choline to biosynthesize PC, a bilayer-forming phospholipid that is imported into mitochondria. CL p , precursor cardiolipin; CL m , mature cardiolipin; PG, phosphatidylglycerol; PGP, phosphatidylglycerol phosphate; CDP, cytidine diphosphate; DAG, diacylglycerol.

Ethanolamine-mediated rescue of CL deficiency is independent of its incorporation into PE
Although significant elevation of PE in taz1⌬ cells upon Etn supplementation can explain the rescue of respiratory growth, the lack of a significant increase in PE upon Etn supplementation in crd1⌬ cells suggests that PE may not be responsible for the observed rescue. To test this idea, we deleted the gene encoding the rate-limiting enzyme of the Kennedy pathway, ethanolaminephosphate cytidylyltransferase (Ect1), in taz1⌬ and crd1⌬ backgrounds. We analyzed the growth of ect1⌬taz1⌬ and ect1⌬crd1⌬ cells in respira-tory media with and without Etn and observed that Etn supplementation rescued growth of the double mutants despite the absence of Ect1 ( Fig. 3A and Fig. S2A). To confirm that the deletion of ECT1 abolished the increase in PE levels upon Etn supplementation, we measured the cellular PE levels in WT and individual single and double mutants with and without Etn supplementation. As expected in the ect1⌬ genetic background, the cellular PE did not increase with Etn supplementation ( Fig. 3B and Fig. S2B).
To independently test the role of PE in rescuing CL deficiency, we used lyso-PE supplementation, which activates an

Ethanolamine rescue of cardiolipin deficiency
alternative pathway for mitochondrial PE elevation (44). Lyso-PE supplementation did not rescue the respiratory growth of either the taz1⌬ or the crd1⌬ cells (Fig. S3, A and B) but, as expected, did rescue respiratory growth of psd1⌬ cells that are deficient in mitochondrial PE (Fig. S3C). Together, these results suggest that Etn-mediated rescue of CL-depleted cells is independent of PE biosynthesis and is mediated by either Etn itself or its downstream metabolites.

Kennedy pathway intermediates are not required for the ethanolamine-mediated rescue of CL-deficient cells
To test the requirement of the Kennedy pathway intermediates in Etn-mediated rescue of taz1⌬ cells, we constructed a triple mutant eki1⌬cki1⌬taz1⌬ strain that cannot convert Etn to phosphoethanolamine (PEtn) (45). We found that Etn sup-plementation was able to rescue the respiratory growth of the triple mutant in a manner similar to taz1⌬ cells (Fig. 3C). This result suggests that Etn itself, not PEtn or the downstream metabolites of the Kennedy pathway, mediates the respiratory growth rescue observed in CL-deficient cells. Furthermore, we found that Etn-mediated rescue was independent of the Etn/ Cho transporter, Hnm1, because deletion of HNM1 failed to abrogate Etn-mediated rescue (Fig. 3D). To test whether Etn is able to enter cells lacking Hnm1, we measured Etn levels in hnm1⌬ and hnm1⌬taz1⌬ cells by LC-MS (LC-MS) as well as by using radiolabeled [ 14 C]Etn. Both of these experiments demonstrated that Etn could indeed enter hnm1⌬ cells, albeit at ϳ5-fold lower levels (Fig. 3, E and F). Consistent with the lower efficiency of Etn uptake in hnm1⌬ cells, the EC 50 of Etn was found to be ϳ4-fold higher in taz1⌬hnm1⌬ cells as compared

Ethanolamine rescue of cardiolipin deficiency
with taz1⌬ cells (Fig. 3G). In addition to Etn, Hnm1 is known to transport choline (Cho), which also regulates the abundance of Hnm1 (46). Therefore, co-supplementation of Etn with Cho is expected to decrease Etn uptake. Indeed, co-incubation of 10 mM choline with radiolabeled [ 14 C]Etn decreased its uptake by ϳ8-fold (Fig. 3H). Under similar experimental conditions, Etnmediated rescue of taz1⌬ and crd1⌬ cells is abrogated indicating that cellular uptake of Etn is required to rescue the respiratory growth defects of CL-deficient cells (Fig. 3I). These results suggested that either Etn or Etn-containing metabolites are responsible for the rescue. To identify other intracellular Etncontaining metabolites that may confer the rescue phenotype, we utilized LC-MS to investigate downstream products of Etn that may rescue CL-deficient cells. We found Etn supplementation resulted in an increase in N-palmitoylethanolamine levels ( Fig. S4), raising the possibility that N-palmitoylethanolamine may serve as the bioactive mediator. However, we did not observe a positive genetic interaction between ECT1 and TAZ1 (Fig. 3A), i.e. the growth rate of ect1⌬taz1⌬ cells, which have higher levels of N-palmitoylethanolamine even without Etn supplementation (Fig. S4), was not higher than that of taz1⌬ cells, thus ruling out this possibility. Taken together, these results suggest a specific role of Etn in ameliorating the respiratory growth defect in CL-depleted cells.

Respiratory growth rescue of CL-deficient cells by Etn analogues
To test the specificity of Etn-mediated rescue, we supplemented respiratory media with other water-soluble lipid precursors, including monomethylethanolamine (MME), dimethylethanolamine (DME), choline (Cho), serine (Ser), and propanolamine (Prn) (Fig. 4A). MME, DME, and Cho act as substrates for the biosynthesis of phosphatidylmonomethylethanolamine (PMME), phosphatidyldimethylethanolamine (PDME), and PC, respectively. Both PMME and PDME serve as interme-diates of PC, the most abundant mitochondrial phospholipid. Serine is a precursor for phosphatidylserine (PS), a less abundant mitochondrial phospholipid, whereas Prn is an Etn analog that is a substrate for phosphatidylpropanolamine (PP), a nonnatural phospholipid (47). We found that Prn, and to a lesser extent MME, supplementation rescued respiratory growth of taz1⌬ and crd1⌬ cells, whereas DME, Cho, and Ser supplementation did not restore the respiratory growth of taz1⌬ or crd1⌬ cells (Figs. 4, B and C, and Fig. S5). These results show that the rescue is specific to Etn and its structurally similar analogues MME and Prn.

Ethanolamine supplementation rescues MRC supercomplex levels in taz1⌬ cells
To determine the biochemical mechanism of Etn-mediated respiratory growth rescue of CL-depleted cells, we determined the supra-molecular assembly of MRC complexes in taz1⌬ cells, which are known to have reduced levels of MRC supercomplexes (3,4,23). Etn supplementation in taz1⌬ cells partially restored the formation of the large supercomplex (III 2 IV 2 ) and reduced the amounts of the free complex III dimer (III 2 ) (Fig. 5, A and B) and the complex IV monomer (Fig. 5, C and D). Consistent with the respiratory growth rescue, Prn supplementation resulted in PP biosynthesis (Fig. S6A) and restoration of MRC supercomplex levels in taz1⌬ cells (Fig. S6B). To determine whether the Etn-mediated restoration of supercomplex levels in taz1⌬ cells was independent of an increase in mitochondrial PE levels, we examined supercomplex formation in taz1⌬ect1⌬ cells, which cannot synthesize PE via the Kennedy pathway. Similar to taz1⌬ cells, Etn was able to promote the formation of the large supercomplex in taz1⌬ect1⌬ cells (Fig. 5, E and F). These results identify Etn and Prn as novel regulators of respiratory supercomplex formation in CL-depleted taz1⌬ cells.

Ethanolamine rescue of cardiolipin deficiency Ethanolamine supplementation restores the expression of MRC subunits in crd1⌬ cells
Previous reports have shown that a complete lack of CL in crd1⌬ cells results in a more pronounced decrease in supercomplex formation and that CL is specifically and absolutely required for supercomplex reconstitution in vitro (5). Consistent with these reports, we found that the supercomplex formation in crd1⌬ cells was not restored with Etn supplementation (Fig. 6, A-D). However, Etn supplementation did increase the levels of the complex IV monomer (Fig. 6, C and D), which were due to increased expression of MRC subunits of complex IV (Fig. 6E). Additionally, Etn supplementation also restored the levels of MRC complex III subunits (Fig. 6E). Consistent with an increase in MRC complex IV levels, the MRC complex IV activity was restored in crd1⌬ mitochondria (Fig. 6F). These results suggest that critical levels of CL are required for supercomplex formation and that the observed partial rescue of respiratory growth of crd1⌬ cells is due to the restoration of complex IV activity.

Cox4 overexpression rescues MRC supercomplex levels in CL-deficient cells
To determine whether increased complex IV biogenesis upon Etn supplementation is sufficient to restore supercomplex formation in CL-deficient cells, we overexpressed the Cox4 subunit of MRC complex IV in taz1⌬ and crd1⌬ cells (Fig.  7A). Indeed, overexpression of Cox4 was sufficient to restore MRC supercomplexes in CL-deficient cells (Fig. 7, B and C). These results are consistent with the idea that Etn stimulates supercomplex formation by increasing the expression of MRC complex IV subunits.

Ethanolamine supplementation reduces protein carbonylation in CL-deficient cells
One of the hallmarks of CL depletion and BTHS pathology across multiple model systems, including the yeast BTHS model taz1⌬, is increased oxidative stress (25,48,49). To determine whether Etn supplementation could alleviate oxidative stress in CL-deficient cells, we measured cellular protein carbonylation, a sensitive indicator of oxidative stress (25). Consistent with a previous report (25), we observed that taz1⌬ and crd1⌬ cells had increased cellular protein carbonylation indicative of enhanced oxidative stress. Etn supplementation reduced protein carbonylation in both taz1⌬ and crd1⌬ cells (Fig. 8A). To determine whether Etn-mediated reduction of protein carbonylation was independent of PE biosynthesis, we measured protein carbonylation in the double mutants taz1⌬ect1⌬ and crd1⌬ect1⌬ and found that Etn was able to reduce carbonylation (Fig. 8B). Together, these results demonstrate that Etn supplementation reduces oxidative stress in taz1⌬ and crd1⌬ cells.

Ethanolamine supplementation does not rescue respiratory growth defects of cells lacking MRC assembly factors
Next, we wanted to determine whether Etn-mediated rescue is specific to CL deficiency or whether any MRC assembly defect could be rescued. To test this possibility, we performed rescue experiments on a number of MRC assembly factors, including the recently discovered mitochondrial proteins, Rcf1 and Rcf2, which are proposed to promote supercomplex formation (50 -52). Consistent with the previous report (51), rcf1⌬ cells exhibited more pronounced growth defect compared with rcf2⌬ cells in respiratory media (Fig. 9A). However, Etn supplementation did not rescue the respiratory growth of either rcf1⌬ or rcf2⌬ cells (Fig. 9A). In addition, Etn supplementation could not rescue the respiratory growth of sdh2⌬, bcs1⌬, shy1⌬, and atp12⌬ cells, which are mutants defective in MRC complexes II-V, respectively (Fig. 9, B and C). Together, these results suggest that Etn-mediated rescue of mitochondrial defects is specific to CL deficiency.

Discussion
We report a novel role for Etn in ameliorating mitochondrial dysfunction caused by perturbations in mitochondrial membrane phospholipid composition due to CL deficiency. We serendipitously discovered this new role of Etn while testing the hypothesis that elevating mitochondrial PE by exogenous Etn  (n ϭ 3). E, mitochondria from WT, taz1⌬, and ect1⌬taz1⌬ cells grown in SC ethanol Ϯ 2 mM Etn were solubilized by 1% digitonin and subjected to BN-PAGE/Western blot analysis and MRC complex III; F, complex IV was detected by probing with anti-Rip1 and anti-Cox2 antibodies, respectively. III 2 IV 2 , large supercomplex; III 2 IV, small supercomplex; III 2 , complex III dimer; IV, complex IV monomer.

Ethanolamine rescue of cardiolipin deficiency
supplementation could rescue mitochondrial defects caused by CL deficiency. We found that Etn ameliorates mitochondrial dysfunction in CL-deficient cells but, surprisingly, without increasing PE levels. Specifically, Etn exerted its protective effect by increasing the expression of MRC complex III and IV subunits and thereby restoring MRC supercomplex formation in CL-depleted cells.
Whereas the phospholipid precursors choline and inositol are known to influence cellular physiology (53), very little is known about the role of Etn outside of its incorporation into PE.
Previous work has shown that Etn and PEtn can directly modulate mitochondrial function (54,55). In fact, pharmacological elevation of PEtn by meclizine, an anti-nausea drug, has been shown to protect the heart, brain, and the kidney from the ischemia-reperfusion injury by altering mitochondrial bioenergetics (56,57). Therefore, we asked whether Etn mediated its effect by incorporation into PEtn. We addressed this question by constructing a triple knockout strain, taz1⌬eki1⌬cki1⌬, which cannot phosphorylate ethanolamine. Our results show that Etn supplementation was still able to rescue the respiratory-  (n ϭ 3). E, MRC subunits of complexes II (Sdh1), III (Cor1 and Qcr2), and IV (Cox2 and Cox4) were analyzed by SDS-PAGE/Western blotting. Mitochondrial proteins aconitase and porin were used as a loading control. Data are representative of at least three independent experiments. F, digitonin-solubilized MRC complexes from WT and crd1⌬ cells were separated by CN-PAGE, followed by in-gel activity staining for complex IV. In-gel activity of complex IV was quantified by densitometric analysis, and relative activity was plotted. Data were normalized to WT cells and expressed as mean Ϯ S.D. (n ϭ 3); *, p Ͻ 0.05.

Ethanolamine rescue of cardiolipin deficiency
deficient growth of the triple knockout cells, indicating that PEtn is unlikely to contribute to the observed rescue (Fig. 3C). The protective effect of Etn on CL-deficient cells was specific because other phospholipid precursors, including Ser and Cho, were unable to rescue the respiratory growth defect of CL-deficient cells (Fig. 4). We determined that the amine group of ethanolamine is critical for its biological activity because addition of the methyl groups on the nitrogen of the amine group progressively diminished its efficacy, as exemplified by the lack of rescue with DME and Cho and only a partial rescue with MME (Fig. 4). Consistent with this observation, Prn, a structural analogue of Etn containing the primary amine group, was also able to rescue the respiratory growth of CL-deficient cells with the same efficacy as that of Etn.
Because the Etn-mediated rescue was independent of the Etn-Kennedy pathway, we considered the possibility that Etn can be incorporated into some other metabolite, which in turn can exert its protective effect. Therefore, we utilized a LC-MS system to discover Etn-containing molecules and found Npalmitoylethanolamine levels to be significantly increased upon Etn supplementation (Fig. S4). This observation raised the possibility that the protective effect of Etn may, in fact, be mediated by N-palmitoylethanolamine, which has been shown to increase upon oxidative stress in yeast (58). However, wenoticedthattherewerehigherbasallevelsofN-palmitoylethanolamine in ect1⌬taz1⌬ cells even without Etn supplementation (Fig. S4), and these cells still exhibit respiratory growth defect (Fig. 3A), thus ruling out any possible role of N-palmitoylethanolamine in mediating the protective effect. Interestingly, higher basal levels of N-palmitoylethanolamine in ect1⌬taz1⌬ cells suggest that blocking the Etn-Kennedy pathway reroutes endogenous Etn to N-palmitoylethanolamine.
To understand the biochemical basis by which Etn restored the respiratory growth of CL-depleted cells, we examined the formation and function of MRC supercomplexes. Previous reports have shown that CL deficiency results in diminished MRC supercomplexes, leading to respiratory defects in both the yeast and mammalian models of BTHS (23,24). Remarkably, exogenous supplementation of Etn was able to rescue MRC supercomplex levels in the CL-depleted mitochondrial membranes of taz1⌬ cells (Fig. 5). The observed Etn-mediated rescue occurred despite MLCL accumulation (Fig. 2D), which has been reported to be the primary cause of mitochondrial dysfunction (20,21). Thus, our finding suggests that detrimental effects of MLCL accumulation can be overcome.
Previous studies have shown that the loss of CL and its precursor phosphatidylglycerol causes an inhibition of translation of MRC subunits, specifically decreasing the levels of complex III and IV subunits (26,27). Thus Etn-mediated rescue of complexes III and IV subunit levels provides a biochemical mechanism for the rescue of respiratory growth of CL-deficient cells (Fig. 6E). We further demonstrated that overexpression of the complex IV subunit is sufficient to restore supercomplex formation in CL-deficient cells (Fig. 7), a finding consistent with

Ethanolamine rescue of cardiolipin deficiency
the recent report showing that supercomplex formation is the function of MRC complex IV abundance (59). Based on these observations, we propose a model where the primary effect of Etn is to restore the expression of MRC complex III and IV subunits (Figs. 6, E and F, and 7), which in turn rescue complex IV activity and promote supercomplex formation in CL-deficient cells. Although the molecular mechanism by which Etn increases the expression of MRC subunits remains elusive at this time, a number of important conclusions can be drawn from our study. First, Etn mediates its effect on mitochondria either directly or by incorporation into a yet unidentified metabolite that is neither PE nor N-palmitoylethanolamine. Second, the protective effect of Etn is specific to CL deficiency because Etn supplementation fails to rescue mitochondrial dysfunction in cells lacking MRC assembly factors. Third, Etn is able to overcome the accumulation of MLCL, which has been shown to be the primary cause of mitochondrial dysfunction in yeast models of BTHS (20,21). Fourth, increased cytosolic Etn levels in Etn-supplemented hnm1⌬ cells suggest existence of alternative Etn import machinery. Finally, exogenously supplemented Etn or endogenous Etn, which builds up in cells with a disruption in the Etn-Kennedy pathway, is rerouted to Nacylethanolamine biosynthesis.
In a recent review, Henry et al. (53) predicted novel roles of lipid precursors in cellular physiology and suggested the suitability of the yeast S. cerevisiae model system to uncover such processes. Our work, demonstrating a novel role of the soluble lipid precursor Etn in ameliorating mitochondrial dysfunction caused by CL deficiency in yeast model of BTHS, provides strong support for their prediction.

Yeast strains, growth medium composition, and culture conditions
S. cerevisiae strains used in this study are listed in Table 1. These strains were confirmed by PCR as well as by replica plating on dropout plates. Single, double, and triple knockout yeast strains were constructed by one-step gene disruption using hygromycin or clonNat (nourseothricin) resistance cassettes amplified from pFA6a-hphNT1 or pFA6a-natNT2 plasmids, respectively (60). The primers used for one-step gene disruption are listed in Table 2. For growth in liquid media, strains were pre-cultured in YPD medium (1% yeast extract, 2% peptone, and 2% dextrose) and inoculated into SC media (0.2% dropout mix containing amino acid and other supplements as described previously (61), 0.17% yeast nitrogen base without amino acids and ammonium sulfate, and 0.5% ammonium sulfate) containing either 2% glucose, 2% lactate, pH 5.5, or 2% ethanol and grown to late logarithmic phase. Solid media were prepared by the addition of 2% agar to the media described above. Yeast strains were inoculated into liquid SC glucose, SC lactate, and SC ethanol at a starting A 600 of 0.1, and growth was monitored for up to 1 day (SC glucose), 2 days (SC lactate), or 3 Figure 9. Ethanolamine-mediated rescue is specific to CL deficiency. A, growth of WT, rcf1⌬, and rcf2⌬ cells in SC glucose and SC ethanol Ϯ 2 mM Etn media was monitored at 30°C (upper panel) and 37°C (lower panel) by measuring absorbance at 600 nm. Data are representative of at least two independent measurements. B, 10-fold serial dilutions of WT, taz1⌬, sdh2⌬, bcs1⌬, shy1⌬, and atp12⌬ cells were seeded onto the indicated media, and images were captured after 2 days of growth on SC glucose and 5 days on SC ethanol Ϯ 2 mM Etn at 30°C. Data are representative of at least three independent trials. C, mitochondria from YP galactose grown WT, sdh2⌬, bcs1⌬, and shy1⌬ cells were subjected to Western blot analysis. Mitochondrial respiratory chain complexes II, III, and IV subunits were probed using anti-Sdh2, anti-Rip1, and anti-Cox2 antibodies, respectively. Porin was used as a loading control. days (SC ethanol), respectively. For growth on solid media, 10-fold serial dilutions of overnight pre-cultures were seeded on SC glucose or SC ethanol plates and incubated at 30 or 37°C for 2 days (SC glucose) and 5 or 7 days (SC ethanol), respectively. For Etn supplementation experiments, 2 mM Etn was added to SC growth medium. For media containing lyso-PE, 1% (v/v) Tergitol Nonidet P-40 was included, and lyso-PE was added to the final concentration of 0.5 mM from a sterile 25 mM stock solution in 10% (v/v) Tergitol Nonidet P-40 (44).

Plasmids
Yeast COX4 gene was cloned into a multicopy plasmid (pRS426) under control of the native promoter using primers listed in Table 2.

Mitochondrial isolation
Isolation of mitochondria was performed as described previously (62). Mitochondria were isolated from yeast cells grown to late logarithmic phase and were subsequently used for SDS-PAGE Western blot analysis as well as in-gel activity assays. For obtaining gradient-purified mitochondrial fractions, crude mitochondria were loaded onto a sucrose step gradient (60, 32, 23 and 15%) and centrifuged at 134,000 ϫ g for 1 h. The intact mitochondria re-covered from the gradient interface (60 and 32%) were washed in isotonic buffer, pelleted at 10,000 ϫ g, and subsequently used for BN-PAGE/Western blot analysis and mitochondrial phospholipid quantification. Protein con-centrations were determined by the BCA assay (Thermo-Fisher Scientific).

Mitochondrial and cellular phospholipid measurements
For the quantification of mitochondrial phospholipids, lipids were extracted from gradient-purified mitochondria (1.5 mg of protein) using the Folch method (63), and individual phospholipids were separated by two-dimensional TLC using the following solvent systems: chloroform/methanol/ammonium hydroxide (65:35:5) in the first dimension followed by chloroform/acetic acid/methanol/water (75:25:5:2.2) in the second dimension (47). Phospholipids were visualized with iodine vapor, scraped into acid-washed glass tubes, and P i was quantified (64). For the quantification of cellular PE, phospholipids were extracted from yeast cells (0.5 g wet weight) using the Folch method (63); P i was quantified, and 30 nmol of phospholipids were separated into individual phospholipid class on HPTLC Silica Gel 60 plates (EMD Millipore 1.11764.0001) by one-dimensional TLC using chloroform/methanol/ammonium hydroxide (50:50:3) (65). Phospholipids were visualized with copper sulfate charring, and bands were quantified using ImageJ (65).

SDS and blue native-PAGE
SDS-PAGE was performed on mitochondrial samples solubilized in lysis buffer (150 mM NaCl, 1 mM EDTA, 50 mM Tris-HCl, pH 7.4, 1% Nonidet P-40, 0.5% sodium deoxycholate,  Table 2 Primers used in this study The boldface letters represent sequences that are complementary to the flanking sequences of the hygromycin and clonNat (nourseothricin) resistance cassettes of pFA6a-hphNT1 or pFA6a-natNT2 plasmids, respectively.

In-gel activity measurements
In-gel activity measurements for mitochondrial respiratory chain complexes were performed as described previously (67). Clear native-PAGE (CN-PAGE) was used to avoid interference of Coomassie Blue with activity measurements. Briefly, mitochondria solubilized in 1% digitonin were resolved on a 4 -16% gradient native PAGE BisTris gel (Life Technologies, Inc.) with the addition of 0.05% n-dodecyl ␤-D-maltoside and 0.05% sodium deoxycholate in the cathode buffer. Gels were loaded with 90 g of protein and incubated in MRC complex IV activity staining solutions as reported previously (67). Equal loading was determined by Coomassie Blue stain, and total protein and band intensity quantification was determined using the gel analysis method in ImageJ.

Steady-state labeling of yeast cells with [ 14 C]ethanolamine
The yeast strains were grown in the presence of [ 14 C]ethanolamine (1 Ci) for 24 h at 30°C with shaking. 1 ml of ϳ2.5 A 600 cells were centrifuged at 3000 ϫ g for 5 min, followed by washing four times with ice-cold water. The [ 14 C]ethanolamine incorporation was measured using a scintillation counter by resuspending the cell pellets into the scintillation mixture.

Quantification of protein carbonyl content
The protein carbonyl content was measured by determining the amount of 2,4-dinitrophenylhydrazone (DNP) formed upon reaction with 2,4-dinitrophenyl hydrazine (DNPH), as described previously (25,68). Yeast cells were grown at 30°C to the early stationary phase in SC ethanol medium, and cell extracts were used for subsequent protein carbonylation measurements. Nucleic acids were removed from the cell extracts with 1.0% streptomycin sulfate, and the resulting protein samples were incubated with 10 mM DNPH in 2 M HCl at room temperature for 60 min in the dark. Proteins were precipitated by addition of TCA to a final concentration of 10%, and the pellets were washed with ethanol/ethyl acetate (1:1) to remove the free DNPH. The final protein pellets were dissolved in 6 M guanidine hydrochloride solution containing 20 mM potassium phosphate, pH 2.4. The carbonyl content was calculated from the absorbance maximum of DNP measured at 370 nm.

Mass spectrometry
Ethanolamine and N-palmitoylethanolamine levels were determined in yeast cells using LC-MS (LC-MS) method, as described previously (55). Yeast cells cultured in 5 ml of SC ethanol medium with and without 2 mM Etn supplementation were pelleted, washed, and frozen. Cell pellets were resuspended in an adjusted volume of ice-cold 80:20 methanol/water with 150 l of solvent per 10 mg of cells that were transferred into a 2.0-ml impact-resistant tube containing 300 mg of 1-mm zirconium beads. Samples underwent three 15-s homogenization cycles at 6400 Hz in a Precellys 24 tissue homogenizer. To ensure complete cell lysis, samples were thereafter sonicated for 2 min and vortexed for 30 s. Samples were then placed in a Ϫ20°C freezer for 30 min to allow for complete protein precipitation. Samples were thereafter vortexed again for 30 s and centrifuged at 14,000 ϫ g for 10 min at 4°C, and supernatants were transferred to LC-MS vials containing 200-l glass inserts. An injection volume of 2.0 l was used so that ϳ134 g of yeast cells were injected for all samples. LC-MS/MS-based metabolomics analysis was performed using a ThermoFisher Scientific QExactive Orbitrap mass spectrometer coupled to a Thermo-Fisher Scientific Vanquish UPLC system. Chromatographic separation of polar metabolites was achieved using a Millipore (Sequant) Zic-pHILIC 2.1 ϫ 150-mm, 5-m column maintained at 25°C. Compounds were eluted via a 19-min linear gradient starting from 90:10 acetonitrile, 20 mM ammonium bicarbonate to 45:55 acetonitrile, 20 mM ammonium bicarbonate. Chromatographic separation of nonpolar metabolites was achieved using an Agilent Eclipse Plus C18 RRHT 1.8 uM 2.1 ϫ 50-mm column maintained at 50°C with a 13-min linear gradient beginning with 5:95 water with 0.2% acetic acid, acetonitrile:isopropanol (1:1) with 0.2% acetic acid and ending with 100% acetonitrile:isopropanol (1:1) with 0.2% acetic acid. A ThermoFisher Scientific Q-Exactive Orbitrap mass spectrometer was operated in positive and negative ion modes using a heated electrospray ionization source at 35,000 resolution for polar metabolites and 70,000 resolution for non-polar, 100-ms ion trap time for MS1 and 17,500 resolution, and 50-ms ion trap time for MS2 collection. For polar compounds, data were collected over a mass range of m/z 59 -885, using a sheath gas flow rate of 40 units, auxiliary gas flow rate of 20 units, sweep gas flow rate of 2 units, spray voltage of 3.5 and 2.5 kV for positiveand negative-ion modes, respectively, capillary inlet temperature of 275°C, auxillary gas heater temperature of 350°C, and Ethanolamine rescue of cardiolipin deficiency an S-lens RF level of 45. For MS2 collection, MS1 ions were isolated using a 1.0 m/z window and fragmented using a stepped normalized collision energy of 15, 30, and 45. For nonpolar compounds, the mass range was m/z 120 -1800, with an auxiliary gas flow rate of 10 units, spray voltage of 3.55 kV for negative ion mode, and capillary inlet temperature of 265°C with all other parameters kept the same. Fragmented ions were placed on dynamic exclusion for 30 s before another round of fragmentation. Collected data were imported into the mzMine 2.26 software suite for analysis.