Perturbation of the yeast mitochondrial lipidome and associated membrane proteins following heterologous expression of Artemia-ANT

Heterologous expression is a landmark technique for studying a protein itself or its effect on the expression host, in which membrane-embedded proteins are a common choice. Yet, the impact of inserting a foreign protein to the lipid environment of host membranes, has never been addressed. Here we demonstrated that heterologous expression of the Artemia franciscana adenine nucleotide translocase (ANT) in yeasts altered lipidomic composition of their inner mitochondrial membranes. Along with this, activities of complex II, IV and ATP synthase, all membrane-embedded components, were significantly decreased while their expression levels remained unaffected. Although the results represent an individual case of expressing a crustacean protein in yeast inner mitochondrial membranes, it cannot be excluded that host lipidome alterations is a more widespread epiphenomenon, potentially biasing heterologous expression experiments. Finally, our results raise the possibility that not only lipids modulate protein function, but also membrane-embedded proteins modulate lipid composition, thus revealing a reciprocal mode of regulation for these two biomolecular entities.


Heterologous expression of ArANT in yeasts affects the function of other membrane-embedded
proteins. In order to examine the effect of heterologous expression of ArANT in yeasts and how it affects the properties of other membrane-embedded proteins, we investigated in-gel and enzymatic activities of ETC complexes and probed for the presence of several key subunits. We further measured oxygen consumption rates in isolated mitochondria and quantitated ATP synthesis capacity and ATP hydrolytic activity of the F O -F 1 ATP synthase complex. Viability of yeast strains, mitochondrial membrane potential, insertion of ArANT to inner mitochondrial membrane, ArANT primary structure verification and functionality have all been addressed in 31 . In-gel activities of ETC complexes are shown in Fig. 3. No significant differences were found for the in-gel activities of Nde1, Nde2 and Ndi1 (belonging to pseudo-complex I) and of complex II among the ArANT-expressing vs Aac2-expressing yeast mitochondria, normalized to the amount of loaded protein deduced by Coomassie staining shown in upper-right panel A (see numbers indicated within upper-left and upper-mid panel A). Although the bands corresponding to pseudo-complex I seem diffuse, this is not unexpected as several of its subunits bind to other subunits of the respiratory chain forming supercomplexes in yeasts 32,33 ; thus, pseudo-complex I activity will be dispersed as a function of binding of its components to other subunits with a variable molecular weight. Lower in-gel complex III and IV activities were observed in the ArANT-expressing strain, compared to the Aac2-expressing strain (normalized to the amount of loaded protein deduced by Coomassie staining, bottom-right panel A). F O -F 1 ATP synthase complex did not show any alterations in either dimeric (V2) or monomeric (V1) form in BN-PAGE electrophoresis, normalized by Coomassie staining. However, there was a higher level in the free F 1 domain in the ArANT-expressing mitochondria, compared to those harboring Aac2 instead, depicted in upper-right panel B. The decrease of the in-gel complex III and IV activity were further investigated by probing for CytB (a complex III component) and Cox2 (a complex IV component by Western blotting after BN-PAGE (performed independently). No significant differences were found by comparing the density of bands obtained from ArANT-expressing strain vs the Aac2-expressing strain, normalized to the amount of loaded protein deduced by Coomassie staining, bottom-right panel C. The results shown in figure panel 3C imply that there were no appreciable differences in complex III and IV levels. It is to be noted that the lack of finding differences among complexes from the in-gel assays is limited by the methods used hereby; more sophisticated methods for addressing in-gel activities of mitochondrial complexes have been published elsewhere 34 . Finally, we must also stress that our experiments did not include measurements of transcript levels of ETC components as performed by Nůsková et al. 30 , thus it cannot be excluded that changes observed hereby were mediated on a genetic level.
To address the possibility that mitochondria and mitoplast fractions obtained from the ArANT and AAC2-expressing strains exhibited a different degree of contamination from non-mitochondrial components potentially affecting lipidomic measurements, we probed for the expression of plasma membrane (Gap1), ER (Wbp1) and ER-Golgi (Sec22) markers. As shown in Fig. 4A, yeast cells (spheroplasts: spher.), mitochondria (mito.), mitoplasts (mitopl.), and the mitochondria plus ER fraction obtained during the Percoll protocol (mito + ER) were probed for the aforementioned proteins by standard Western blotting. The plasma membrane marker Gap1 appeared only in the spheroplasts, while the ER-Golgi marker Sec22 was evident only in the mito + ER fractions. Although it should have also been present in the spheroplasts fraction, there it was absent probably due to the fact that it yields a weak antibody-antigen interaction ('weak band'). The extent of contamination by ER in the various fractions is better appreciated by inspecting the Wbp1 blot: there, it is apparent that the extent of ER contamination is largely diminished upon mitochondria and mitoplast purification, and there is no difference between those obtained from the ArANT-vs Aac2-expressing yeasts. The WB of Por1 confirms the extent of enrichment of mitochondrial outer membranes in the various fractions. Although porin is an exclusive component of the outer mitochondrial membrane, the association of this protein with the so-called 'contact sites' 35 , parts where the outer and inner mitochondrial membranes exhibit strong affinity, mitoplast fractions depleted from outer membrane still stain positive for porin, as it has been shown elsewhere 36 . From the results shown in Fig. 4A we concluded that if there is differential contamination of mitochondrial/mitoplast fractions by ER components, it is below the detection limit of the method used.
To investigate further the effect of heterologous expression of ArANT in yeasts, we probed for several key membrane-embedded ETC components, specifically CytB, Atp6, Atp7, Atp2, Atp9, Cox2, and normalized them to the expression level of porin (Por1: isoform 1) by Western blotting. As shown in Fig. 4B, except Cox2 (a subunit of complex IV) there were no other alterations in the contents of the probed membrane-embedded subunits. Densitometric analysis of the bands pooled from three replicate gels are shown in Fig. 4C. No statistical significance was observed by comparing band densities for any protein tested between ArANT (black bars) and Aac2 (grey bars) yeast samples.
To evaluate the impact of alterations conferred by the heterologous expression of ArANT observed in the BN-, CN-and SDS-PAGE experiments on mitochondrial functionality, we measured individual complex activities, oxygen consumption during various metabolic conditions, and ATP synthesis and hydrolysis rates. As shown in figure panel 5A, enzymatic activities of complexes II and IV were significantly decreased in mitochondria expressing ArANT, comparing to those harboring Aac2. The decrease in complex IV activity was also reflected in the significantly decreased rate of oxygen consumption when TMPD + ascorbate −obligate substrates of exclusively complex IV-were used, in ArANT expressing mitochondria (figure panel 5B). State 2 and 3 respiration did not reach statistical significance between the yeast strains, even though ArANT mitochondria exhibited a ~2-fold decrease in state 3 respiration compared to that obtained by Aac2-harboring mitochondria. Subsequent addition of the uncoupler CCCP though afforded a statistical significance in the comparison of respiration rates between ArANT-and Aac2-harboring mitochondria. Regarding ATP synthesis/hydrolysis rates mediated by F O -F 1 ATP synthase, results are shown in figure panels 5 C and 5D, respectively. The rate of ATP synthesis (measured at state 3 with NADH as a respiratory substrate) was ~60% lower in the ArANT-expressing than Aac2-expressing mitochondria. The lack of statistical significance between ArANT and Aac2-expressing mitochondria during state 3 respiration shown in Fig. 5B may seem at odds with the finding that ArANT-expressing mitochondria produce less amount of ATP per unit time, depicted in Fig. 5C; however, it is important to consider that the rate of state 3 respiration and rate of ATP production need not be in a linear relationship. It may well be possible that the flux-control coefficient of F O -F 1 ATP synthase working in the forward mode is not sufficiently high to warrant a strong influence in state 3 respiration. The rate of mitochondrial ATP hydrolysis was assessed at pH = 8.4 in non-osmotically protected conditions and in the presence of saturating amounts of ATP aiming for maximal ATPase activity rates; alkaline pH prevents the binding of IF1 to ATP synthase 37 , and ATP hydrolytic activity is not limited by a proton gradient. ATPase activity in ArANT-expressing mitochondria was reduced by 25% compared to the Aac2-harboring mitochondria, and free F1 likely mediated 67% of this activity since it was only partially sensitive to oligomycin. However, as a word of caution, the oligomycin-insensitive ATP hydrolysis rate could also be due to non-mitochondrial ATPases contaminating the mitochondrial fractions. The extent of this contamination cannot be reliably quantified, however, on the basis of the results shown in Fig. 4 it is anticipated that it is similar between ArANT-and Aac2-expressing mitochondria. In Aac2-harboring mitochondria, approximately 47% of ATPase activity was mediated by free F1 plus contaminating non-mitochondrial ATPases.

Discussion
In the present study we investigated the effect of heterologous expression of ArANT in yeast by examining their total, mitochondrial and mitoplastic lipidomes and interrogating the function of other proteins embedded in the same membrane as the heterologously expressed one. Our results demonstrated that there were both qualitative and quantitative lipidomic changes conferred by expressing ArANT in the inner mitochondrial membrane. Most notably, diacylglycerols, phosphatidylcholines, lysophosphatidylcholines, lysophosphatidylethanolamines, phosphatidylinositols, triacylglycerols and glycolipids were all significantly increased in the mitochondrial membranes of ArANT expressing yeasts. These changes suggest significant biophysical adaption of the subcellular mitochondrial membrane as well as inner mitochondrial membrane in response to metabolic activity. This is evidenced by lysolipid formation as well as neutral lipid and glycerol-lipid content changes. In addition to the lipidomic changes, heterologous expression of ArANT also affected the enzymatic activities of complexes II, IV and the F O − F 1 ATP synthase. In-gel assays of complex II did not show a difference between the ArANT and Aac2-harboring yeast strains, however, these assays are not as sensitive as the enzymatic ones.
It is to be noted that neither composition nor the amount of cardiolipins were different between the ArANTand Aac2 expressing yeast mitoplasts. This is important because cardiolipins are essential for the optimum activity of many inner mitochondrial membrane proteins, including the translocase itself [38][39][40][41][42][43] . It can be therefore stated that the alterations in mitochondrial lipidome caused by the heterologous expression of ArANT affecting the functions of other membrane-embedded proteins was not due to a catastrophic change in cardiolipins composition, in line with the biochemical stability and overall viability shown for this yeast strain 31 .
Our work outlines two ramifications: firstly, by heterologously expressing a membrane-embedded protein in a host, the membrane lipidome is altered. It cannot be overemphasized that our work concerns an individual case, that of expressing ArANT in the yeast strain MWY83/1; However, it sets the precedent of checking for changes in the membrane lipidome in experiments designed to include heterologous expression of other  44 . Thus, a 'reductionist approach' would probably be an oversimplification. A loss or decrease of one lipid class may have no effect, and the interaction of loss or decrease of two or more lipid classes may actually be responsible for the alteration. Also, research regarding lipids is lagging compared to research regarding proteins and nucleic acids. Systematic investigation is needed on the mechanisms by which lipids modulate protein function and structure 45 . Relevant to this, specialized protocols involving liposome-microarray-based assays (LIMA) are just being developed 46 .
Secondly, the alterations in membrane lipidome conferred by the heterologous expression of a foreign protein, in turn, may alter the function of other membrane-embedded proteins. The concept of lipid modulation of membrane-embedded proteins is now becoming recognized 15,19,[47][48][49] , at least to some extent 50 . Furthermore, it has been proposed that lipids can act as "lipochaperones" thus influencing protein folding in diseases such as Alzheimer's, prion/scrapie disease, cystic fibrosis, cataract formation, and type 2 diabetes 43 . It must be emphasized that protein-protein interactions may also be responsible for changes observed in view of the fact that The complexes were detected by using antibodies against CytB (for complex III) and Cox2 (for CIV)). For loading controls, gels shown in A, B and C were stained with Coomassie (three replicates); a representative gel is depicted on the right of each panel. Densitometric analysis of the gel bands shown in panels (A and C) are arbitrarily normalized to the signals obtained from Aac2 samples. Densitometric analysis of the gel bands shown in the rightmost panel (B) is represented as the ratio of free F1/(dimers + monomers). Data shown are Mean +/− S.E.M. (n = 5). For panel (C), the Densitometric analysis included all bands. CIII 2 -CIV 2 implies supramolecular forms of two complex III and two complex IV entities; CIII 2 -CIV implies supramolecular forms of two complex III and one complex IV entity, while CIII 2 implies supramolecular forms of two complex III entities. Images are representative of scanned gels from three to five independent experiments. *p < 0.05, and **p < 0.001.
Scientific REPORTS | (2018) 8:5915 | DOI:10.1038/s41598-018-24305-2 membrane lipid composition alterations can affect membrane-embedded proteins. Specifically, it is known that overexpressing a protein in a host may reduce the steady-state levels of other proteins, or disrupt stoichiometric complexes, compete for a shared subunit of two complexes or sequester a protein, reviewed in 1 . At least for human adenine nucleotide translocases It has been recently shown that they physically and functionally interact with the respirasome, i.e., membrane-embedded subunits of the electron transport chain 51 . Furthermore, in yeast it has been demonstrated that the synthesis of cytochrome c oxidase subunit 1 is downregulated in the absence of a functional F O -F 1 -ATP synthase, through a mechanism dependent on Cox1p synthesis 52 . Thus, it is impossible to decipher the extent of contribution of lipidome changes on affecting membrane-embedded functions vs that due to direct protein-protein interactions. Further studies will be needed to elucidate the mechanism(s) by which heterologously expressed proteins alter host lipidome and associated membrane proteins.
Isolation and bioenergetic analyses of yeast mitochondria. Assays were performed on mitochondria isolated from yeast cells grown in rich galactose medium (YPGalA, 1% Bacto yeast extract, 1% Bacto Peptone, 2% galactose, 40 mg/l adenine) at 28 °C. The mitochondria were isolated as previously described 53 . Oxygen consumption rates were measured with a Clark electrode in a buffer consisting of 0.65 M mannitol, 0.36 mM EGTA, 5 mM Tris-phosphate, 10 mM Tris/maleate, pH 6.8, as previously described 54 . For ATP synthesis rate measurements, the mitochondria (0.15 mg/ml) were placed in a 1 ml thermostatically controlled chamber at 28 °C in the same buffer as for the polarographic measurements. The reaction was started by adding 4 mM NADH and 0.15 mM ADP (for respiration assays) or 0.75 mM ATP (in the presence or absence of oligomycin, for determination of ATP synthesis or hydrolysis rates). 100 µl aliquots were taken either every 15 seconds and the reaction was stopped by adding 3.5% perchloric acid and 12.5 mM EDTA. Samples were neutralized to pH 6.5 by KOH and 0.3 M MOPS. ATP was quantified by Kinase-Glo Max Luminescence Kinase Assays (Promega) using a Beckman Coulter's Paradigm Plate Reader. Other additions were 12.5 mM ascorbate, 1.4 mM N,N,N,N,-tetramethyl-p-phenylenediamine (TMPD), 4 μM CCCP or 3 μg/ml oligomycin, where indicated. The specific ATPase activity was measured at pH 8.4 using a previously described procedure, quantitating the amount of liberated phosphate from ATP hydrolysis 55 .
In-gel catalytic activities assays. Regarding NADH dehydrogenases and complex II, CN-PAGE gels were washed with 5 mM Tris-HCl pH 7.4. For NADH dehydrogenases, gel was incubated in a buffer composed of 5 mM Tris-HCl pH 7.4, 1 mg/ml NBT and 0.2 mM NADH until blue strips appeared and photographed. For complex II assay, gels were washed in 5 mM Tris-HCl pH 7.4 several times and incubated in a buffer composed of 5 mM Tris-HCl pH 7.4, 1 mg/ml NBT, 20 mM sodium succinate and 0.2 mM PMS) until blue/violet strips appeared and photographed. Regarding complexes III and IV, CN-PAGE gels were washed with 5 mM Tris-HCl pH 7.4. For complex III activity, gels were incubated in a buffer composed of 5 mM Tris-HCl pH 7.4 and 0.5 mg/ml DAB until brown strips appeared and photographed. In gel activity assay for complex IV was performed in the same gels; gels were washed in 5 mM Tris-HCl pH 7.4 several times and incubated in a buffer consisting of 5 mM Tris-HCl pH 7.4, 0.5 mg/ml DAB and 0.05 mM cytochrome c until new brown strips appeared and photographed.
Liquid/Liquid Extraction of Structural Lipids. Mitoplasts from Percoll-purified yeast mitochondria, Percoll-purified yeast mitochondria, and intact yeast cells were thawed and diluted with a ten-times diluted PBS solution. All samples were homogenized in Omni bead tubes with 2.8 mm ceramic beads in the Omni Bead Ruptor 24 with Cryo Cooling Unit (Omni International, Kennesaw, GA) at 4 °C for 2 minutes. Protein concentration was determined by the bicinchoninic acid assay and 1 mg of protein from mitoplast, mitochondria, and intact yeast cell samples were aliquoted and a cocktail of deuterium-labeled and odd chain phospholipid standards from diverse lipid classes was added. Standards were chosen to represent each lipid class and were at designated concentrations to provide the most accurate quantitation and dynamic range for each lipid species (Supplementary  Table 5). To each sample, 4 mL chloroform:methanol (1:1, by vol) was added and lipidomic extractions were performed as previously described [Kiebish et al., 2010]. Lipid extraction was automated using a customized sequence on a Hamilton Robotics STARlet system (Hamilton, Reno, NV). Lipid extracts were dried under nitrogen and reconstituted in chloroform: methanol (1:1, by vol). Samples were flushed with nitrogen and stored at −20 °C.

Direct Infusion MS/MS ALL Structural Lipidomics Platform.
Samples were diluted 50 times in isopropanol:methanol:acetonitrile:water (3:3:3:1, by vol.) with 2 mM ammonium acetate in order to optimize ionization efficiency in positive and negative modes. Electrospray ionization-MS was performed on a TripleTOF ® 5600+ (SCIEX, Framingham, MA), coupled to a customized direct injection loop on an Ekspert microLC200 system (SCIEX). A sample volume of 50 µL was injected at a flow-rate of 6 µL/min. Lipids were analyzed using a customized data independent analysis strategy on the TripleTOF ® 5600+ allowing for MS/MS ALL high resolution and high mass accuracy analysis as previously described [Simons et al., 2012]. A customized in-house method was used for CL species with lower collision energy and shifted isolation windows. Quantification was performed using an in-house library on MultiQuant ™ software (SCIEX). Heat maps generated through unsupervised hierarchical clustering of the top 50 molecular lipid species was performed by MetaboAnalyst 3.0 62 .