Mechanism for Remodeling of the Acyl Chain Composition of Cardiolipin Catalyzed by Saccharomyces cerevisiae Tafazzin*

Remodeling of the acyl chains of cardiolipin (CL) is responsible for final molecular composition of mature CL after de novo CL synthesis in mitochondria. Yeast Saccharomyces cerevisiae undergoes tafazzin-mediated CL remodeling, in which tafazzin serves as a transacylase from phospholipids to monolyso-CL (MLCL). In light of the diversity of the acyl compositions of mature CL between different organisms, the mechanism underlying tafazzin-mediated transacylation remains to be elucidated. We investigated the mechanism responsible for transacylation using purified S. cerevisiae tafazzin with liposomes composed of various sets of acyl donors and acceptors. The results revealed that tafazzin efficiently catalyzes transacylation in liposomal membranes with highly ordered lipid bilayer structure. Tafazzin elicited unique acyl chain specificity against phosphatidylcholine (PC) as follows: linoleoyl (18:2) > oleoyl (18:1) = palmitoleoyl (16:1) ≫ palmitoyl (16:0). In these reactions, tafazzin selectively removed the sn-2 acyl chain of PC and transferred it into the sn-1 and sn-2 positions of MLCL isomers at equivalent rates. We demonstrated for the first time that MLCL and dilyso-CL have inherent abilities to function as an acyl donor to monolyso-PC and acyl acceptor from PC, respectively. Furthermore, a Barth syndrome-associated tafazzin mutant (H77Q) was shown to completely lack the catalytic activity in our assay. It is difficult to reconcile the present results with the so-called thermodynamic remodeling hypothesis, which premises that tafazzin reacylates MLCL by unsaturated acyl chains only in disordered non-bilayer lipid domain. The acyl specificity of tafazzin may be one of the factors that determine the acyl composition of mature CL in S. cerevisiae mitochondria.

detail. Phosphatidylglycerolphosphate synthase (Pgs1) catalyzes the first step of CL synthesis by converting CDP-diacylglycerol and glycerol 3-phosphate to produce phosphatidylglycerol phosphate (PGP) (6). PGP is dephosphorylated to phosphatidylglycerol by the PGP phosphatase Gep4 (7) (PTPMT1 in mammals (8,9)). CL synthase (Crd1) catalyzes the final step of CL synthesis by condensing phosphatidylglycerol and another molecule of CDP-diacylglycerol to produce CL, which is immature CL with primarily saturated acyl chains of various lengths (10,11).
Following the de novo synthesis of CL on the matrix side of the inner mitochondrial membrane (12), acyl chain remodeling is responsible for the final molecular composition of mature CL, which is typically defined by the symmetric incorporation of unsaturated fatty acyl chains (13). In this process, CL is deacylated to monolyso-CL (MLCL) by CL-specific lipase Cld1 (14). The reacylation of MLCL is accomplished by three distinct proteins in higher eukaryotes, tafazzin, monolysocardiolipin acyltransferase 1 (MLCLAT1), and acyl-CoA:lysocardiolipin acyltransferase-1 (ALCAT1), whereas yeast only undergoes tafazzin-mediated reacylation. Although MLCLAT1 and ALCAT1 utilize acyl-CoA as the acyl chain donor for the reacylation of MLCL (15,16), tafazzin is a transacylase that takes an acyl chain from another phospholipid, preferentially phosphatidylcholine (PC) or phosphatidylethanolamine (PE), and adds it to MLCL (17). Mutations in tafazzin cause Barth syndrome, resulting in cardio-and skeletal myopathy and respiratory chain defects (18,19). In Barth syndrome patients and models of Barth syndrome, the CL/MLCL ratio is decreased, and the remaining CL contains an altered acyl chain composition (e.g. low content of unsaturated acyl chains) (20 -23).
Because tafazzin exhibits little acyl chain specificity among potential phospholipid substrates (namely this enzyme randomly reacts with virtually all phospholipid and MLCL species) (13,17), the acyl chain compositions of all CLs may become identical if they are exposed to tafazzin for a sufficient period of time; however, this is not the case (24). Therefore, a critical question is how tafazzin causes specific patterns of acyl chains in mature CL; for example, tetralinoleoyl (18:2)-CL is a major CL in mammalian heart, and oleic (18:1) and palmitoleic (16:1) acids are major acyl chains in yeast mitochondrial CL (25).
Schlame and co-workers (17) investigated whether purified Drosophila melanogaster tafazzin can replicate the remodeling of CL in vitro. They demonstrated that efficient transacylation from phospholipids to MLCL only occurs in non-bilayer lipid aggregates (e.g. the inverted hexagonal lipid phase) and not in ordered lipid bilayer membrane like liposomes (26). Acyl chain specificity was also shown to be highly sensitive to the lipid structural order, namely a remarkable preference for linoleoyl (18:2) over oleoyl (18:1) groups was only observed in non-bilayer lipid membranes (26). Based on these findings, they proposed the so-called thermodynamic remodeling hypothesis to explain acyl specificity in the tafazzin-mediated remodeling of CL (26); tafazzin functions at non-bilayer-type relaxed lipid domains, which occur in curved membrane domains, because of the preferential mixing of phospholipids and MLCL species, and its acyl specificity is driven by the packing properties of these domains. In other words, multiple transacylations will change the acyl chain composition of CL until their free energy is minimal, at which point the chain composition is optimal for specifically curved membrane domains (27). CL, which has a relatively small volume of polar headgroup and unsaturated acyl chains (such as linoleic acid), has a shape asymmetry that preferentially leads to its accumulation in the negatively curved lipid monolayer (28).
Nevertheless, in light of the diversity of acyl chain compositions of CL in different organisms or even in tissues within the same organism, it currently remains unclear whether the thermodynamic remodeling hypothesis accounts for all details of tafazzin-mediated CL remodeling. For example, Schlame et al. (26) reported that only a portion (Ͻ1%) of endogenous mitochondrial phospholipids participates in transacylation (phospholipids ϩ [ 14 C]lyso-PC (LPC) 3 lyso-phospholipids ϩ [ 14 C]PC), suggesting that the action of tafazzin is highly limited to specific mitochondrial domains. However, this finding is difficult to reconcile with the fact that mammalian heart mitochondria contain an extremely high content of the uniform acyl chain composition, tetralinoleoyl-CL (ϳ80% of all CL (29)), which indicates that a large extent of the remodeling of CL takes place in this organ. Although mitochondria contain curveshaped membrane domains (the cristae), it has not yet been established whether there are substantial domains that are sufficiently curved to be categorized as non-bilayered membrane domain (30). Furthermore, hexagonal and micellar phases have both been shown to facilitate tafazzin-mediated transacylation (26), which is unexpected because the two lipid states are characterized by opposite curvatures of the lipid-water interface. The shape of CL makes it favorable for being positioned at negatively curved membrane regions (28), as mentioned above. Thus, it is still debatable whether the thermodynamic remodeling hypothesis is a general feature of tafazzin-mediated transacylation. A critical point to be verified is whether transacylation accompanying the remarkable acyl specificity (linoleoyl (18:2) Ͼ Ͼ oleoyl (18:1)) from PC to MLCL only occurs in non-bilayer lipid aggregates and not in ordered lipid bilayer membranes.
We herein investigated the mechanism underlying tafazzinmediated transacylation using purified S. cerevisiae tafazzin with a liposomal system composed of various pairs of PC (acyl donor) and MLCL (acyl acceptor) species (Fig. 1). Because all CL species used in this study were synthesized according to the procedures developed in our laboratory (31)(32)(33), we were able to arbitrarily fix their acyl chain compositions. This study clearly demonstrated that S. cerevisiae tafazzin catalyzes efficient transacylation among phospholipid substrates even with the liposomal system, the highly ordered lipid bilayer environ-ment. Tafazzin elicited unique acyl chain specificity against PC as follows: linoleoyl (18:2) Ͼ oleoyl (18:1) ϭ palmitoleoyl (16:1) Ͼ Ͼ palmitoyl (16:0). On the basis of these results, we conclude that the thermodynamic remodeling hypothesis is not necessarily a general scenario to describe the mechanism underlying tafazzin-mediated CL remodeling. The substrate specificity of tafazzin may be one of the factors that determine the acyl composition of mature CL.

Results
Effects of the GST Tag and Ca 2ϩ on the Catalytic Activity of Tafazzin-We investigated transacylation from PC to MLCL catalyzed by isolated S. cerevisiae GST-tagged tafazzin with FIGURE 1. Structures of CL, MLCL, DLCL, and PC species synthesized in this study. The synthetic procedures for these phospholipids are described in the supplemental material. Regarding the abbreviated designation of CL species, four acyl chains are arranged to make the chiral carbon of the central glycerol to have the R configuration. For MLCL species, the synthetic products are racemic mixtures of (S) sn-1-and (R) sn-1Ј-MLCL or (S) sn-2-and (R) sn-2Ј-MLCL. Similarly, one of DLCL species is a racemic mixture of (R) sn-1/sn-2Ј-and (S) sn-1Ј/sn-2-DLCL. liposomes in the highly ordered lipid bilayer environment. The PC/MLCL molar ratio was fixed at 9:1 throughout this study to maintain the ordered lipid bilayer structure (34,35). We preliminarily examined the effects of two factors (the GST tag and Ca 2ϩ ) on the catalytic activity of tafazzin to optimize the reaction conditions.
To determine the effect of the GST tag on the catalytic activity of tafazzin, we compared activities between GST-tagged and tag-free enzymes in the reaction PC(18:1-18:1) ϩ sn-2Ј-MLCL(18:1-18:1/18:1-OH) 3 LPC ϩ CL(18:1-18:1/18:1-18:1). The results obtained showed that the GST tag did not affect enzyme activity (Fig. 2, column b). Therefore, we used GSTtagged tafazzin throughout this study because of the less laborious purification process. In this and following experiments, transacylation did not occur in the absence of tafazzin (data not shown); therefore, a mere chemical reaction was not responsible for the transacylation observed.
Schlame et al. (26) previously used a high concentration of Ca 2ϩ (20 mM) to prepare inverted hexagonal lipid phase from lipid mixture dispersion. Because the possibility that a high concentration of Ca 2ϩ affects the catalytic activity of tafazzin cannot be excluded, we investigated the effect of Ca 2ϩ (1 and 20 mM) on enzyme activity in the reaction PC(18:1-18:1) ϩ sn-2Ј-MLCL(18:1-18:1/18:1-OH) 3 LPC ϩ CL(18:1-18:1/18:1-18:1). Enzyme activity determined in the presence of 1 mM Ca 2ϩ was almost identical to that determined in the absence of Ca 2ϩ , but it was reduced by ϳ30% in the presence of 20 mM Ca 2ϩ (Fig. 2,  columns c and d). Because the presence of high concentrations of Ca 2ϩ was unfavorable for tafazzin-mediated transacylation under our experimental conditions, we omitted Ca 2ϩ from the reaction mixture in this study.
sn-2 Acyl Chain of PC Is Selectively Transferred to Monolyso-CL-In tafazzin-mediated acyl transfer from PC to MLCL, the acyl chain of PC (sn-1, sn-2, or both positions) that is deacylated has not yet been identified. This previously unresolved issue is important to obtain a more comprehensive understanding of the mechanism underlying the tafazzin-mediated transacylation. To investigate this issue, sn-1-LPC and sn-2-LPC are needed as the standard samples. It is the general nature of lyso-phospholipids that the acyl group in the sn-2 position tends to migrate to the sn-1 position if the sn-1 position is deacylated (i.e. free glycerol OH) (31,36,37). Therefore, we carefully synthesized sn-1-LPC(OH-18:1) and sn-2-LPC(18: 1-OH) separately (Fig. 1). Judging from their 1 H NMR spectra, a content of sn-2-LPC(18:1-OH), which was formed by automigration during the synthetic and/or purification processes, in sn-1-LPC(OH-18:1) sample was ϳ5%.
We  (17,23,24,26,38) evaluated the catalytic activity of tafazzin by determining the amounts of lipids produced after a certain reaction period by means of counting the radioactivity of 14 C-incorporated products or a quantitative MALDI-TOF MS analysis. However, because any kinetic parameter of the catalytic reaction cannot be evaluated by such single-point measurements, we herein followed the time course of the formation of the product of interest.
Efficient acyl transfer from PC to sn-2-MLCL was observed in all donor-acceptor pairs even in liposomes (as shown in   Table  1. Our results demonstrated that PC(18:2-18:2) is slightly superior to PC(18:1-18:1) as an acyl donor to sn-2-MLCL; the rate constants of the former were ϳ1.5-fold greater than those of the latter. It is important to note that the remarkable preference   (Table 1).
Furthermore, to investigate effects of membrane components on the catalytic activity of tafazzin, we also examined the reaction using liposomes composed of PC(18:2-18:2) and sn-2Ј-MLCL(18:1-18:1/18:1-OH) at a 8:2 molar ratio. The mean diameter of liposomes made up of the 8:2 mixture was identical to that of liposomes of the 9:1 mixture (data not shown). Although the apparent reaction rate observed with the 8:2 mixture was considerably faster (by ϳ1.3-fold) than that with the 9:1 mixture, the rate constants were almost identical between the two reactions (0.10 Ϯ 0.02 min Ϫ1 for 8:2 versus 0.098 Ϯ 0.008 min Ϫ1 for 9:1).
There may be two possibilities that are the cause of the poor transacylation activity for PC(16:0-16:0) as follows: one is that tafazzin cannot remove the palmitoyl group from PC, and the other is that tafazzin has this ability but cannot transfer it to MLCL. To verify these two possibilities, we determined the amount of PC(16:0-16:0) remaining in the reaction mixture after the 40-min reaction by HPLC analysis, and we found that the residual amount is almost identical to the initial amount (a total of ϳ45 nmol). This result strongly suggests that tafazzin does not have the ability to remove the palmitoyl group from PC.
Regiospecificity of Tafazzin-mediated Transacylation from PC to Monolyso-CL Isomers-All four acyl chains of CL participate in the tafazzin-mediated remodeling process, which requires a mechanism that does not discriminate between the sn-1-and sn-2-glycerol positions. Using two MLCL isomers (sn-1-MLCL and sn-2-MLCL), which lack one of the four acyl groups at the sn-1 or sn-2 position, Malhotra et al. (24) demonstrated that tafazzin transfers an acyl group into the sn-1 and sn-2 positions at almost equally efficient rates. Nevertheless, because the features of tafazzin-mediated transacylation were significantly different between previous studies (24,26) and our work, as described so far, we reinvestigated the regiospecificity of acyl transfer from PC to MLCL isomers by comparing the reaction rates of sn-1Ј-MLCL (  We carefully synthesized sn-1-MLCL species and handled them in subsequent experiments to avoid automigration of the acyl group in the sn-2 position, as mentioned above. The sn-1 and sn-2-MLCL isomers were determined accurately based on their 1 H NMR spectra (Fig. 5, A and B); sn-1-MLCL, but not sn-2-MLCL, yielded a specific resonance signal at ϳ5.0 ppm, which is characteristic of the POCH 2 -CH(OR)-CH 2 OH proton (H(8) in Fig. 5A). The 1 H NMR spectra of our sn-1-MLCL samples indicated that the contents of sn-2 isomers are less than 5%, as shown in Fig. 5A taking sn-1Ј-MLCL(18:1-18:1/ OH-18:1) as an example.
We established an HPLC analytical method that enables discrimination between sn-1-MLCL and sn-2-MLCL. Fig. 5C shows an example of the HPLC charts observed for sn-1Ј-MLCL(18:1-18:1/OH-18:1) alone (panel i), sn-2Ј-MLCL(18:1-18:1/18:1-OH) alone (panel ii), and a mixture of them (panel iii), which indicated that the extent of automigration is negligibly small under the HPLC analytical conditions. By the HPLC analysis of MLCL isomers recovered from the assay mixtures with or without tafazzin, we also confirmed that no significant migration occurs during the assay process or the preparation of liposomes (data not shown). In contrast, judging from 1 H NMR spectra reported in Ref. 24, a small fraction of sn-2-MLCL (ϳ20%) was included in the previous sn-1-MLCL sample that was prepared by treating commercially available CL with Rhizopus arrhizus lipase, which removes an acyl chain preferentially from the sn-1 position. It is unclear whether the authors checked the automigration in subsequent experiments.
As summarized in Table 1, we corroborated that tafazzin is able to efficiently transfer an acyl group into the sn-1 and sn-2 positions of MLCL isomers; there was no preference for the glycerol positions. The acyl chain compositions of acyl donors and acceptors did not significantly affect this feature. It is noteworthy that the slightly greater preference for PC(18:2-18:2) than PC(18:1-18:1) as an acyl donor was again observed with sn-1-MLCL, as noted with sn-2-MLCL. The lack of regiospecificity in tafazzin-mediated transacylation is a prerequisite for uniformity and/or symmetry in the acyl distribution of CL (24).
Monolyso-CL Functions as an Acyl Donor to Lyso-PC-Using commercially available MLCL and DLCL, Malhotra et al. (24) demonstrated that the role of MLCL in tafazzin-mediated transacylation is limited to that of an acyl acceptor from phospholipids to produce CL, and DLCL does not function as an acyl acceptor. Based on this information, the authors designed acyl donor-acceptor pairs for transacylation experiments in their subsequent study (26). However, if MLCL serves as an acyl donor, the complexity of transacylation patterns may markedly increase. Therefore, it is important to clarify whether MLCL serves as an acyl donor for other lyso-phospholipids.
We examined this issue using an MLCL-LPC mixture at a 1:9 molar ratio (sn-2Ј-MLCL(18:1-18:1/18:1-OH) ϩ sn-2-LPC(18: 1-OH) 3 DLCL ϩ PC(18:1-18:1)). Dynamic light scattering Detailed HPLC analytical conditions were described under "Experimental Procedures." measurement revealed that this lipid mixture failed to form liposomes but may form micelle-like morphology with an averaged diameter of ϳ20 nm, which is drastically smaller than that of liposomes prepared from other acyl donor-acceptor pairs (an averaged diameter of ϳ110 nm). This may be due to the lack of PC as a major component when preparing liposomes. We note that PC was not used in this reaction because substantial transacylation occurs from PC to LPC and MLCL.
As shown in Fig. 6, the amount of sn-2-MLCL in the reaction medium gradually decreased, whereas the reaction was almost complete after ϳ5 h with ϳ20% consumption of initial sn-2-MLCL. We detected the production of PC(18:1-18:1), the amount of which corresponds to that of consumed sn-2-MLCL. The unique lipid phase of this lipid mixture may be responsible for the incomplete reaction. We were able to identify the acyl composition of the DLCL produced as sn-2/sn-2Ј-DLCL(18:1-OH/18:1-OH) by an HPLC analysis using synthetic DLCL isomers as the standard samples, which have two oleoyl groups at the sn-1-sn-1Ј, sn-1-sn-2Ј, or sn-2-sn-2Ј position (Fig. 1), and their averaged retention times were 13.8, 14.7, and 15.5 min, respectively. This result indicates that an oleoyl group in the sn-2 position of sn-2-MLCL was selectively transferred to sn-2-LPC to produce PC. Thus, our results clearly reveal that MLCL has an inherent ability as an acyl donor.
As shown in Fig. 7A, sn-2/sn-2Ј-DLCL was efficiently converted to sn-2-MLCL (open circles). The rate constant for this reaction (0.072 Ϯ 0.005 min Ϫ1 ) was similar to that determined for transacylation from PC(18:1-18:1) to sn-2Ј-MLCL(18:1-18: 1/18:1-OH) ( Table 1). Because we detected the formation of a small fraction of CL(18:1-18:1/18:1-18:1) (Fig. 7A, closed circles) during the first 20 min, we continued to monitor the formation of CL. The amount of CL gradually increased with a   JULY 22, 2016 • VOLUME 291 • NUMBER 30 slight decrease in the total amount of sn-2-MLCL (Fig. 7B). Taking the inherent transacylation ability of tafazzin into consideration, it is feasible that, once formed, sn-2-MLCL was further transformed to CL by accepting an acyl chain from PC. However, this (second) transacylation reaction was significantly slower than that observed for the direct production of CL from a mixture of sn-2-MLCL and PC (e.g. Fig. 4A). This may be because the initial concentration of sn-2-MLCL in the former was significantly lower than that in the latter. The total amount of sn-2-MLCL and CL produced after ϳ400 min was almost equivalent to the initial amount of sn-2/sn-2Ј-DLCL (5.0 nmol), indicating that almost all sn-2/sn-2Ј-DLCL was transformed to these CL species during the reaction. In contrast to previous findings (24), our results clearly demonstrated that MLCL and DLCL function as an acyl donor to LPC and acyl acceptor from PC, respectively, in tafazzin-mediated transacylation.

Remodeling of the Acyl Chain Composition of Cardiolipin
Transacylation Activity of the H77Q Mutant-Claypool and co-workers (23,38) established Barth syndrome-associated tafazzin mutants in S. cerevisiae mitochondria and classified them into seven functional classes based on their biochemical and cell biological characterization. Of the seven classes, the class 4 mutations resulted in catalytic inactive tafazzin, although expression levels, localization in mitochondria, and assembly in macromolecular complexes were almost normal. His-77 is located in the HX 4 D motif, a conserved structural motif in acyltransferase, and the H77Q mutant is a typical example of the class 4 mutants (23). We prepared the H77Q mutant and isolated and determined its transacylation activity in the reaction PC Our results showed that this mutant almost completely lacks the catalytic activity in our model membrane system (Fig. 2, column e), as observed with the assay using S. cerevisiae mitochondria (phospholipids (in mitochondria) ϩ [ 14 C]MLCL 3 [ 14 C]CL) (23). Lu et al. (42) recently demonstrated that the human equivalent of the H77Q mutant (H69Q in human) remarkably lacks the transacylation activity, indicating conservation of the functional importance of this histidine.

Discussion
Because tafazzin is considered to exhibit little acyl chain selectivity among possible phospholipid substrates (13,17), the mechanism by which it determines the acyl chain composition of mature CL has yet to be explained. The thermodynamic remodeling hypothesis (26) appears to ingeniously explain the marked enrichment of the CL fraction of D. melanogaster and mammalian mitochondria with CL(18:2-18:2/18:2-18:2). This hypothesis, in principle, is based on the idea that tafazzin reacts only at non-bilayer-type lipid domains, which may occur in curved membrane zones, and the specificities of acyl chain compositions are derived from the physical properties of these domains (i.e. lipid packing). In other words, highly curved membranes promote the catalytic reaction of tafazzin with multiple unsaturated acyl chains, and tafazzin, in turn, stabilizes the membrane curvature (26,27). To verify this hypothesis, it is critical to prove that tafazzin does not function in the reacylation of MLCL by an unsaturated acyl chain in structurally ordered lipid bilayer membrane. Schlame et al. (26) did not necessarily prove this issue in the strict sense because they used the dispersion of acyl donor-acceptor lipid mixtures, which were prepared by sonication for 2 min in a bath-type sonicator. On the basis of the chemical shift anisotropy of magic angle spinning 31 P NMR, they regarded the lipid phase primarily as hexagonal and lamellar phases in the presence and absence of 20 mM Ca 2ϩ , respectively; the former is less structurally ordered than the latter. However, it may be difficult to unequivocally define the phases (or polymorphism) of lipid dispersion solely by the 31 P NMR technique under their experimental conditions. Therefore, we herein investigated this issue with unilamellar liposomes, ordered lipid bilayer membranes, using purified S. cerevisiae tafazzin.
The results of this study revealed that isolated tafazzin efficiently catalyzes transacylation from PC possessing unsaturated acyl chains to MLCL, even in liposomes. The acyl donors (PC) exhibited a slight, but not strong (26), preference for linoleoyl (18:2) over oleoyl (18:1) groups irrespective of the acyl chain compositions of the acyl acceptors (MLCL). In these reactions, tafazzin selectively removed the sn-2 acyl chain of PC and transferred it into the sn-1 and sn-2 positions of MLCL isomers at almost equivalent rates. It is important to note that tafazzin exhibited the remarkable acyl specificity for PC; it substantially transferred the palmitoleoyl (16:1) group but not the palmitoyl (16:0) group in PC to MLCL. Moreover, we revealed for the first time that sn-2-MLCL and sn-2/sn-2Ј-DLCL have inherent abilities to function as an acyl donor to sn-2-LPC and an acyl acceptor from PC, respectively, in the tafazzin-mediated in vitro reaction. However, it currently remains unclear whether sn-2-MLCL and sn-2/sn-2Ј-DLCL actually function as an acyl donor and acceptor, respectively, in CL remodeling in mitochondria. Additionally, we corroborated that His-77, located in the conserved structural motif (HX 4 D) of acyltransferase, is critical for the catalytic activity. Collectively, we conclude that the thermodynamic remodeling hypothesis is not necessarily a general scenario to describe the mechanism underlying tafazzin-mediated CL remodeling, at least for S. cerevisiae tafazzin. On the basis of our results, we will discuss the mechanism responsible for CL remodeling later.
Some discrepancies concerning the features of tafazzin-mediated transacylation exist between earlier studies (24, 26) and our work. To identify these discrepancies, we compared the specific activity of isolated tafazzin between both studies, taking the reaction (PC 2) mixtures with various molar ratios to investigate this reaction (see Fig. 5 in 26). The averaged specific activity of our enzyme was markedly higher than the highest activity reported in their study (ϳ20 versus ϳ1.5 pmol of CLs/min/g of tafazzin). Although we are unable to specify a critical causal factor for this difference, it may be related to differences in the origin as well as the expressed form of tafazzin; D. melanogaster maltose-binding protein (ϳ43 kDa)-tagged tafazzin and S. cerevisiae GST (ϳ27 kDa)-tagged tafazzin were used in the earlier study (24,26) and our work, respectively. We confirmed that the existence of the GST tag does not affect the catalytic activity of S. cerevisiae tafazzin. Based on previous findings (17,24,26), it is unclear whether the maltose-binding protein tag affects the catalytic activity of D. melanogaster tafazzin. Moreover, besides differences in the origin and expressed form of tafazzin, we may not exclude possible differences in the morphological states of lipid mixtures (liposome versus lipid dispersion), which are substrates of tafazzin as well as the microenvironment in which tafazzin works. It is also important to stress that our HPLC analytical method enabled the accurate quantification of the lipids produced because of diminishing sample loss by omitting the extraction step of the lipids with an organic solvent(s) (see "Experimental Procedures").
The results of this study demonstrated unique acyl chain specificity in tafazzin-mediated transacylation from PC to MLCL as follows: linoleoyl (18:2) Ͼ oleoyl (18:1) ϭ palmitoleoyl (16:1) Ͼ Ͼ palmitoyl (16:0). Nevertheless, the mechanism underlying CL remodeling in mitochondria may not be described solely by the functions of tafazzin. Contribution of other enzymes to CL remodeling has been discussed such as Cld1 (43) and enzymes with lysophospholipid:acyl-CoA acyltransferase activity, MLCLAT1 (15) and ALCAT1 (16). In particular, Cld1, which is a CL-specific phospholipase and predominantly removes saturated acyl groups such as palmitoyl and stearoyl (18:0) groups (14), may play an important role in the initiation of CL remodeling in yeast mitochondria (40,44). The present observations that tafazzin could not efficiently remove the palmitoyl group from PC and CL support this notion. Although the content of the palmitoyl group in PC and PE species (i.e. acyl donors) is as high as that of palmitoleoyl group in yeast mitochondria (14), palmitoleoyl and oleoyl groups are two major acyl components of mature CL (25,40). The enrichment of these acyl groups in yeast CL may be explained by the preferential removal of saturated palmitoyl and stearoyl groups from de novo synthesized CL by Cld1 to provide MLCL (14,40) and the subsequent tafazzin-mediated reacylation of MLCL by oleoyl and/or palmitoleoyl groups, but not palmitoyl, in PC. The enrichment of the linoleoyl group in CL may not be feasible because its content in PC and PE species is originally low in yeast mitochondria (14). Thus, CL remodeling may require the coordinated action of phospholipase(s) and transacylase(s), such as tafazzin. Nevertheless, because tafazzin and Cld1 (both contain no transmembrane segment) are associated with the intermembrane space-facing and matrix-facing leaflets of the mitochondrial inner membrane, respectively (4,45), MLCL generated by Cld1 must be transported to the intermembrane space-facing leaflet to gain access to tafazzin, suggesting the existence of an as yet unidentified protein(s) capable of redistributing MLCL across the inner membrane. Further studies are needed to establish all of the involved players and elucidate their roles in CL remodeling in mitochondria.
Information on the physiological importance of the regulation of CL remodeling remains very limited, thereby preventing a more comprehensive molecular understanding of how this clinically relevant process promotes mitochondrial functions. Two groups recently and independently showed that defined changes in the acyl chain composition of CL do not significantly alter either the mitochondrial morphology or mitochondrial bioenergetic functions in yeast (40,44). Both groups suggested alternative physiological roles for the CL remodeling process, for example a repair mechanism that removes and replaces acyl chains damaged by oxidative stress, thereby restoring the oxidative phosphorylation capacity of mitochondria (40,44).
Synthesis of PC and CL Analogues and Their Abbreviated Designation-The PC and CL analogues synthesized in this study are shown in Fig. 1. The synthetic procedures for these phospholipids are described in the supplemental Schemes 1-6. We used abbreviated designations for PC and CL analogues to characterize their acyl compositions with discrimination between the sn-1 and sn-2 glycerol positions. The former and latter figures in parentheses represent the acyl group in the sn-1 and sn-2 positions, respectively; for example, PC(18:2-18:1) means PC having linoleoyl and oleoyl groups in the sn-1 and sn-2 positions, respectively, and sn-2Ј-MLCL(18:1-18:1/18:1-OH) means monolyso-CL that has three oleoyl groups in the sn-1, sn-2, and sn-1Ј positions but lacks an acyl chain in the sn-2Ј position of one glycerol moiety.
Preparation of S. cerevisiae Tafazzin and its H77Q Mutant-The ORF of the S. cerevisiae gene encoding tafazzin (TAZ1, YPR140w) was amplified by PCR using KOD Fx Neo DNA polymerase (Toyobo, Japan), with the genome gained from the BY4741 strain used as a template DNA and the oligonucleotides (ScTaz1InfusionFw, TCGAAACGAGGAATTCATGTCTTT-TAGGGATGTCCTAGAAAGA; ScTaz1InfusionRv, GAGTT-TTTGTTCTAGATCATCCTTACCCTTTGGTTTACC) as primers. The amplified fragment was introduced into the pGEX6P-1 vector (GE Healthcare, Japan) linearized with EcoRI and NotI restriction enzymes (New England Biolabs, Japan), using an In-fusion HD Cloning kit (Takara Bio, Japan) to yield the pGST-Taz1 plasmid. To prepare a point-mutated derivative of the yeast tafazzin (H77Q), mutagenesis PCR was performed using PrimeStar Max DNA polymerase (Takara Bio, Japan) with pGST-Taz1 as a template and the oligonucleotides (ScTaz1H77Qfw, CCTTATGACGGTCATGAACCAAATG-AGTATGGTCGATGATCCG; ScTaz1H77QRv, CGGATCA-TCGACCATACTCATTTGGTTCATGACCGTCATAAGG) as primers, which produced the pGST-Taz1H77Q plasmid.
Escherichia coli Rosetta (DE3)pLysS (Merck Millipore, Japan) strains harboring the constructed plasmids were used for the expression of yeast tafazzin and its mutant derivative tagged N-terminally with glutathione S-transferase (GST). Cells were grown in 1 liter of LB Broth medium (Sigma-Aldrich, Japan) at 37°C to an OD 610 of 1.0 and subjected to the induction of expression by the addition of 0.4 mM (final concentration) isopropyl 1-thio-␤-D-galactopyranoside and cultured at 20°C for 16 h. After the induction, cells were harvested, washed with 50 ml of PBSE (137 mM NaCl, 27 mM KCl, 10 mM Na 2 HPO 4 , 1.8 mM KH 2 PO 4 , and 1 mM EDTA, pH 7.4), and pressed in 20 ml of PBSE containing cOmplete TM , EDTA-free (Sigma-Aldrich, Japan), using a One Shot model cell disruption system (Constant Systems, UK). The sample was mixed with 0.1% (w/v, final concentration) Triton X-100, incubated at an ambient temperature for 10 min, and centrifuged at 2,000 ϫ g at 4°C for 5 min. The supernatant fraction was retrieved and mixed with 1.5 ml of glutathione-Sepharose 4B resin (GE Healthcare, Japan) at an ambient temperature for 30 min. The resin was then washed with PBSE containing 0.1% (w/v) Triton X-100 and incubated with 7 ml of elution buffer (50 mM Tris-HCl, pH 7.6, 1 mM EDTA, 0.1% (w/v) Triton X-100, and 10 mM reduced glutathione) at an ambient temperature for 15 min. The eluted sample was dialyzed against 3 liters of dialysis buffer (50 mM Tris-HCl, pH 7.6, 1 mM EDTA, and 0.1% (w/v) Triton X-100).
For the preparation of tag-free tafazzin, the purified GSTtagged tafazzin was treated with 320 units/ml (final concentration) of PreScission protease (GE Healthcare, Japan) in cleavage buffer (150 mM NaCl, 1 mM EDTA, 1 mM DTT, and 50 mM Tris-HCl, pH 7.6) at 4°C for 3 h. This reaction mixture was then applied to glutathione-Sepharose 4B resin (GE Healthcare) to obtain pure tag-free tafazzin. SDS-PAGE of purified GSTtagged and tag-free tafazzin and GST-tagged H77Q mutant were shown in Fig. 8.
Preparation of Liposomes-Liposomes (large unilamellar vesicles) made of an acyl donor and acceptor mixture (PC/MLCL, PC/dilyso-CL (DLCL), or LPC/DLCL at a 9:1 molar ratio) were prepared by the extrusion method (46,47). Stock solutions of an acyl donor and acceptor in a methanol/chloroform (1:3, v/v) solution were mixed in the required proportion. A thin lipid film, which was obtained by evaporating organic solvents and was left under a vacuum overnight to remove residual organic solvents, was hydrated with Tris buffer (50 mM Tris-HCl, pH 7.4) and vortexed (final concentration of 9.0 mM choline, equivalent to 1.0 mM CL derivatives). After seven rounds of freezethawing under a N 2 atmosphere, the lipid suspension was extruded through a 100-nm pore polycarbonate filter using a LiposoFast device (Avestin, Ottawa, Canada). The diameter of liposomes was determined from dynamic light scattering measurements (ELSZ-2Plus; Otsuka Electronic, Osaka, Japan). The particles scarcely varied in diameter with lipid compositions; the mean diameter estimated based on light scattering intensity was ϳ110 nm, as shown in Fig. 9A, taking PC(18:1-18:1)/sn-2Ј-MLCL(18:1-18:1/18:1-OH) liposomes (9:1 molar ratio) as an example. The concentrations of PCs in liposomal preparations were determined using an enzyme assay kit for choline (Wako, Osaka, Japan) (47).
The reaction was quenched after definite incubation periods by the addition of 5 l of solvent A (CH 3 CN/H 2 O/(C 2 H 5 ) 3 N/ CH 3 COOH, 89:10:0.5:0.5 by volume) and 5 l of solvent B (2-propanol/H 2 O/(C 2 H 5 ) 3 N/CH 3 COOH, 89:10:0.5:0.5 by volume). Note that the enzyme reaction was quenched within ϳ20 s by this treatment. To accurately quantify lipid products, we directly subjected the reaction samples, without an extraction step using an organic solvent, to the HPLC analysis described below. The recovery of products in our HPLC analytical method was greater than 95%. We adopted the reverse-phase ion pair HPLC with acidified triethylamine to efficiently separate and quantify lipid products (48). HPLC analysis (SCL-10, Shimadzu, Japan) equipped with an evaporative light scattering detector (ELSD) (model 300S, Softa Co.) was performed in the gradient mobile phase using solvents A and B, as described above. The RP-18 GP Aqua 5-m ODS column (150 ϫ 4.6 mm, KANTO Chemical, Japan) was connected with the guard cartridge (5C18-AR-II, 10 ϫ 4.6 mm, Nacalai Tesque, Japan), both of which were incubated at 50°C in the column oven. ELSD (58 and 72°C for the spray chamber and drift tube, respectively) was kept at a pressure of 1.4 bar (2.5 liters/min) for nebulization gas (N 2 ). To separate lipids, except for DLCL, the gradient profile used was as follows: 0 -10 min, an isocratic run with 53% of solvent B; 10 -30 min, 53-100% of solvent B with a linear gradient; 30 -40 min, an isocratic run with 100% of solvent B. Flow rates were 0.8 and 0.4 ml/min for 0 -18 and 18 -40 min, respectively. For the separation of DLCL species, the HPLC system was operated under isocratic conditions with 20% of solvent B for 20 min (flow rate, 0.8 ml/min). The amounts of the lipid products were estimated by interpolating the light scattering intensity to calibration plots (in a range of 3-40 g) obtained with each synthetic standard sam- ple. Because chromatographic behavior of Triton X-100 and tafazzin in the HPLC analysis was remarkably different from that of lipid products of interest, they did not interfere with the detection of lipid products.
We note that ELSD is a powerful tool for detecting lipids that have no strong chromophore (49,50). Based on our experience, the sensitivity of ELSD is markedly superior, by at least ϳ100fold, to that of a UV-visible absorption detector (e.g. monitoring absorption at 205 nm) for the detection of lipids; accordingly, our HPLC analytical method enabled the reproducible detection of lipids at one-digit microgram levels.