Exogenous Ether Lipids Predominantly Target Mitochondria

Ether lipids are ubiquitous constituents of cellular membranes with no discrete cell biological function assigned yet. Using fluorescent polyene-ether lipids we analyzed their intracellular distribution in living cells by microscopy. Mitochondria and the endoplasmic reticulum accumulated high amounts of ether-phosphatidylcholine and ether-phosphatidylethanolamine. Both lipids were specifically labeled using the corresponding lyso-ether lipids, which we established as supreme precursors for lipid tagging. Polyfosine, a fluorescent analogue of the anti-neoplastic ether lipid edelfosine, accumulated to mitochondria and induced morphological changes and cellular apoptosis. These data indicate that edelfosine could exert its pro-apoptotic power by targeting and damaging mitochondria and thereby inducing cellular apoptosis. In general, this study implies an important role of mitochondria in ether lipid metabolism and intracellular ether lipid trafficking.


Introduction
One fifth of all glycerophospholipids in humans are ether lipids featuring an alkyl chain at the sn-1 position and commonly an ethanolamine or choline head-group [1]. Highly abundant in certain tissues ether lipids are linked to pathologies and genetic disorders although their biological function remains unclear [2]. An involvement in membrane organization, fusion and trafficking or intracellular signaling and protection against oxidative stress is discussed [1,3,4]. Changed levels link to neurological dysfunction and degeneration [1,5]. Inactivation of ether lipid biosynthesis in mice causes male infertility, defects in eye development, cataract and optic nerve hypoplasia [2]. Little detail of the subcellular localization of ether lipids is known; although synthesized in peroxisomes and the endoplasmic reticulum (ER), they also occur in the plasma membrane (PM), post-Golgi compartments and lipid droplets, LDs [1,2,6,7]. Neutral ether lipids contribute 20% to the LD core and etherphosphatidylcholine, ePC, and ether-phosphatidylethanolamine, ePE were detected in the surrounding monolayer [7].
Polyene lipids are fluorescent lipid analogues of high structural similarity to natural lipids [27]. Microscopy of polyene lipids holds the invaluable advantage of studying lipids in living cells avoiding compromises inherent to subcellular fractionation and purification approaches. Here, we employ polyene-ether lipids to visualize their subcellular distribution. Polyfosine, a fluorescent analogue of edelfosine, is used to elucidate the cellular localization of the bioactive compound and gain insights into its action.
Edelfosine was from Sigma, other ether lipids ( Fig. 1) were synthesized as described in Text S1.

Methods
COS7 (from ATCCH, Number: CRL-1651) cell culture and delivery, extraction, TLC and detection of lipids were performed as described [27].
Quantification of signals on TLC plates and Western blotting films was performed using ImageGaugeV3.3 (Fuji, Duesseldorf, GER).

Cell fractionation
Cells were washed and collected as above, followed by homogenization in a cooled EMBL cell cracker (HGM, Heidelberg, GER) with 8 strokes using a maximum clearance of 18 mm. To purify the mitochondria by differential centrifugation the supernatant (PNS, post nuclear supernatant) after a first centrifugation step (100-1,0006g; 2-5 min, respectively) was separated from the pellet (PNP, post nuclear pellet) and centrifuged again (10,0006g; 10 min) to obtain a crude mitochondria pellet (PMP, post mitochondrial pellet) and a supernatant (PMS, post mitochondrial supernatant). The PMP was redissolved in HES and layered on top of a step gradient (40%, 26% and 12% Percoll in HES). The gradient was centrifuged (25 min, 150,0006g) before harvesting 8 fractions from the top. Samples were subjected to SDS-PAGE and Western blotting.

Microscopy
Two-photon-excited fluorescence microscopy of living cells was performed as described [27]. Epifluorescence microscopy of living and fixed cells was performed using a Zeiss Observer.Z1 microscope (Carl Zeiss, Oberkochen, GER) equipped with a C-Apochromat 636 (1.20 NA) and a Photometrics Evolve camera, or a Plan-Apochromat 636 (1.40 NA) DIC and a Photometrics Coolsnap K4 camera, respectively. Live cells imaging was performed at 37 degree C in a 5% carbondioxid atmosphere. Light source was a Polychrome V 150 W xenon lamp (TillPhotonics, Grä felfing, GER). Confocal immuno-fluorescence laser scanning microscopy of fixed cells was performed using a Leica TCS SP2 microscope (Leica, Wetzlar, GER) equipped with a HCX PL APO CS 406 (1.25 NA). Relief contrast bright field microscopy of living cells was performed using an Olympus CKX31 microscope (Olympus, Hamburg, GER) equipped with a LCACHN 20xRC (0.4 NA) and a halogen light source. Images were acquired with Canon Powershot digital camera (Canon, Amsterdam, NED). Phase-contrast video microscopy of living cells was performed using an Olympus IX70 microscope (Olympus) equipped with a U-Plan S Apo 1006 (1.40 NA) and a primary halogen lamp light source. A secondary light source, a Polychrome II 75 W xenon lamp (TillPhotonics) with its soft shutter wavelength set to 280 nm, and an FT395-LP415 (Carl Zeiss) was used to illuminate the cells continuously with broadband UV light. Digital images were acquired with a NTE/CCD-512-EBFT camera (Roper Scientific, Ottobrunn, GER). All images were processed employing Adobe Photoshop 6.0 (Adobe).

Polyene-ether lipids derived from alkyl glycerol precursors
We fed polyene-alkyl glycerols to cells and analyzed the fluorescent metabolic products by TLC. From 50 mM concentrations of c16:5-alkyl-glycerol or c20:5-alkyl-glycerol, the cells produced primarily neutral ether lipids (ether-diglycerides, eDG; ether-triglycerides, eTG) and ether glycerophospholipids (ePE, ePC and ether-phosphatidic acid, ePA) during the experiment ( Fig. 2A and B). Differences in alkyl chain lengths did not influence the labeling pattern, also when 3 mM concentrations were used (data not shown). Considerable amounts of precursor, whose concentration decreased only slowly during a chase, were detectable. Studying the distribution of polyene-ether lipids in living cells by two-photon-microscopy ( Fig. 2C) revealed a staining of the nuclear envelope, ER, LDs and mitochondria. To unequivocally identify the latter we performed epifluorescence microscopy colocalization studies (Fig. 2D). LDs and mitochondria were confirmed to contain polyene-ether lipids. Peroxisomes did not accumulate ether lipids as shown before [27].

Polyene-ether lipids derived from lyso-ether lipid precursors
A shortcoming of using alkyl glycerols for ether lipid tagging is the limited specificity. Labeling selectivity was greatly increased when cells were incubated with polyene lyso-ether lipids. Application of polyene-lyso-ePE yielded fluorescent ePE upon cellular acylation (Fig. 3A). At later timepoints (60-120 min) minimal labeling of ePC could be detected as also natural ePC is synthesized from ePE [1]. During a 2 h chase the precursor concentration was reduced by 70%. When cells were incubated with polyene-lyso-ePC, cellular acylation yielded ePC with high specificity and rate (Fig. 3B). During the 2 h chase the precursor was converted to only ePC. Uptake and specificity of various polyene precursors were quantified (Fig. 4). The alkyl glycerol was taken up 5-fold more efficiently (Fig. 4A), but after 60 min incubation the lyso-ether lipids showed a 8-fold or 5-fold higher labeling specificity for the desired lipid class ( Fig. 4B and C), which increased further during a chase.
To identify the fatty acids used for cellular acylation of polyene lyso-ether lipids MS analyses were performed (Table 1). Arachidonic acid (20:4) was the predominant acylation partner for both polyene-lyso-ePE and -ePC. Other endogenous unsaturated fatty acids, oleate (18:1), linoleate (18:2), and palmitoleate (16:1) were also detected. These findings are expected, since ePE and ePC species are rich in polyunsaturated fatty acids in untreated COS7 cells (Fig. S1) and other cells [1]. The 6-fold higher abundance of palmitate (16:0) and the presence of myristate (14:0) in ePC, but not in ePE (Table 1) might point to a different specificity of the acyltransferases with a higher acceptance of saturated acyl chains in ePC compared to ePE. The concentration of vinylated polyene plasmenyl-PE was 24-fold lower than that of polyene plasmanyl-PE. Due to peak overlay plasmenyl-PC could not be analyzed. MS analysis did not indicate any significant conversion of polyene ePE or ePC into ether-phosphatidylserine, ePS, or ePA. Taken together, polyene-lyso-ether lipids are acylated by cellular enzymes without apparent disturbances by the tag.

Ether lipids in mitochondria and ER
The specific labeling of ePE or ePC obtained using the respective polyene-lyso-ether lipid allowed for the localization of fluorescent ePE and ePC on the subcellular level in living cells for the first time. Two-photon-excitation microscopy revealed that ePE and ePC prominently stained the mitochondria, nuclear envelope and ER (Fig. 5). LD staining was not detected, as only minor amounts of fluorescent neutral ether lipids were biosynthesized during the experiment (Fig. 3).
When cells were incubated with polyfosine, a fluorescent analogue of the metabolically stable, synthetic ether lipid edelfosine, no significant metabolites could be detected within    20 h (Fig. 6). Fluorescence microscopy analysis revealed a pronounced polyfosine staining of mitochondria with a noticeable marking of nuclear envelope, ER and PM (Fig. 7). Long incubation with polyfosine resulted in shortened or swollen mitochondria ( Fig. 7B and C). Colocalization studies confirmed the accumulation to mitochondria (Fig. 7D), while lysosomes contained hardly any detectable polyfosine (Fig. 7E).

The impact of polyfosine and edelfosine on mitochondria
To compare the potentials of edelfosine and polyfosine to induce cellular apoptosis the activation of caspase-3 in immortal COS7 cells was monitored by microscopy and Western Blotting (Fig. 8). The number of caspase-3 positive cells and its concentration increased over time, faster for edelfosine compared to polyfosine treatment. Long incubation times resulted in decreased cell viability and number, due to apoptosis-induced detachment ( Fig. 8A and Fig. S2).
To compare the inhibition of PC biosynthesis by edelfosine and polyfosine the incorporation of a radio-labeled fatty acid into PC was assayed (Fig. 9A). After an initial decrease in the incorporation rate, edelfosine treated cells stopped the synthesis and remodeling of PC. Longer polyfosine treatment also reduced the amount of labeled PC, although less efficiently. To corroborate these findings we analyzed the major cellular lipid classes by MS (Fig. 9B). The PC and PE contents were quantified relative to the sum of major lipids. After 1 h treatment, edelfosine decreased the total PC content by about 38%, polyfosine by 18% compared to control cells (Fig. 9B).  The morphological changes of mitochondria upon polyfosine accumulation were studied by fluorescence microscopy employing a dye sensitive to mitochondrial activity and vitality (Fig. S3A). In control cells mitochondria were often elongated and stained with similar intensities. Many polyfosine and all edelfosine treated cells had short and fragmented mitochondria (Fig. S3B). While many apoptotic cells upon polyfosine treatment did not accumulate the dye, edelfosine treated cells with condensing nuclei or arrested in mitosis often showed intensively stained mitochondria with an increased cellular background stain (Fig. S3A). Mitochondrial fragmentation leads to changes in organelle size, shape and density, which prevented their isolation by density gradient centrifugation to a satisfying purity from contaminating ER (Fig.  S4).
Mitochondrial fragmentation was analyzed by phase-contrast video microscopy ( Fig. 10 and Movie S1). After 10 min some polyfosine and edelfosine treated cells started mitochondrial fragmentation, while others followed after 15-25 min. Shortly after fragmentation the mitochondria of polyfosine, but not edelfosine treated cells condensed and disintegrated, followed by nuclear condensation. Control cells incubated with carrier or polyene-lyso-ePC, the latter sharing the fluorophore and the cellular localization with polyfosine ( Fig. 5C and D) showed no morphological changes. Blebbed mitochondria upon polyfosine treatment were also Apparently, the phototoxicity potentiated the membrane-disrupting effects of fluorescent polyfosine, but not edelfosine, known for its radiosensitizing properties [34].

Discussion
Information on the subcellular localization of most ether lipids is limited. Using polyene lyso-ether lipids, which upon cellular acylation yielded fluorescent ePC or ePE, allowed for their unprecedented individual microscopic analysis. Both ether phospholipids strongly accumulated to mitochondria, besides the ER. The fact that fluorescent ePC or ePE was not detectable on LD monolayers [7] or the PM [4] likely reflects limitations in optical resolution and detection sensitivity. Polyene-sphingolipids were not accumulating on mitochondria arguing against an erratic targeting of the fluorophore to mitochondria [27]. Comparing the labeling patterns of polyene lyso-ether lipids and alkylglycerols we established lyso-ether lipids as supreme precursors for phosphoether lipid tagging as they represent a good compromise between labeling specificity, similarity to natural lipids and applicability to cells. Cell derived polyene-ePC and -ePE, like natural ether glycerophospholipids [1], often contained unsaturated fatty acids most prominently arachidonic acid. Placed to mitochondria ether lipids might serve as a sink for harmful oxidants resulting from oxidative phosphorylation. However, only a small amount of polyene-ePE was detected as vinylated plasmenyl-ePE with antioxidative properties [2]. Alternatively, mitochondria undergo extensive fusion and fission cycles [35] as do synaptic vesicles, which show high plasmalogen concentrations [36]; ether phospholipids might facilitate membrane reorganization in both cases.
After lyso-ether lipid feeding mitochondria became the first labeled organelles justifying speculations of the underlying targeting mechanism. The structure of lyso-ether lipids supports a non-facilitated transport between membranes. For instance, the related lyso-PC is spontaneously released from membranes within milliseconds, which allows for rapid kinetic equilibration [37]. Its release and cytosolic transport are supported by fatty acid binding proteins [38]. Mitochondria may provide the enzyme for lysoether lipid acylation, which would arrest the resulting ether lipid in the mitochondrial membrane. Tafazzin, a phospholipid acyltransferase catalyzing transacylation between lyso-PC and cardiolipin [39] represents a candidate.
Alternatively, ePC and ePE generated by non-mitochondrial acyltransferase(s) could be targeted to mitochondria by a mechanism also accessible to the synthetic ether lipids polyfosine and edelfosine. Especially for the latter, a raft-dependent endocytosis step has been discussed and is supported by the enrichment of both edelfosine [16,40] and ether lipids [4] in rafts. Alternatively, membrane continuities between organelles could be involved as reported for lipid trafficking before [41]. Noteworthy, many other cellular organelles including LDs, which act as a cellular sink for hydrophobic xenobiotics [42] did not accumulate polyfosine, arguing for a specific rather than a stochastic enrichment at mitochondria.
One may not expect polyfosine enrichment in mitochondria as edelfosines established molecular targets are located to the ER and PM [16]. Labeling by a fluorescent phenylpolyene-analogue of edelfosine was restricted to the PM in leukemic cells or predominantly found in the ER of solid carcinoma yielding a model where the cell specific distribution pattern correlates with different cellular targets [43,44]. A recent study performed in HeLa cells uses fluorescent borondifluorodipyrromethene (BODIPY) coupled edelfosine analogues and confocal laser microscopy to describe a mitochondrial localization of these probes [45]. Polyfosine, the analogue matching best the chemical structure of edelfosine confirms mitochondria as a target for the drug in COS7 cells, while verifying an ER and PM localization. Cell type or different cellular handling of the various analogues may explain the discrepancy in localization of the fluorescent analogues. Mitochondrial localization may be hard to detect if only non-confocal microscopic images are available [43,44] as ER staining can interfere.
While polyfosine and edelfosine activated caspase-3 and interfered with PC biosynthesis our study supports the notion that the accumulation of these lipids to mitochondria per se results in a structural damage of the organelle that ultimately leads to apoptosis. These inverted cone-shaped lipids possibly generate membrane curvature if high local concentrations are met [46]. Indeed, edelfosine induced swelling of isolated mitochondria [45] and membrane destabilization in liposome model systems by formation of interdigitated structures, micelles and small vesicles [47]. Edelfosine's detergent properties [10] and very low critical micellar concentration were found to cause lipid solubilization and, importantly, release of content from lipid vesicles [48]. Physical membrane destabilization and loss of mitochondrial integrity results in release of mitochondrial proteins like cytochrome c [49], which ultimately induces apoptosis [35].
In summary, this study establishes polyene lyso-ether lipids as supreme precursors for phosphoether lipid tagging. Fluorescent phosphoether lipids accumulated to mitochondria, suggesting an important role for the organelle in lipid trafficking.   Figure S3 Morphological changes of mitochondria and nuclei upon polyfosine or edelfosine treatment of COS7 cells. Cells were incubated with 50 mM polyfosine, edelfosine or carrier for the indicated times. Mitotracker, whose accumulation depends on mitochondrial activity and vitality, was added prior fixation and fluorescence microscopy (A). Merged color images show green nuclei and red mitochondria, stained by DAPI or Mitotracker, respectively. Bars, 50 mm. (B) Cells with fragmented mitochondria were counted from microscopy images (70-200 cells total for each time point). (TIF) Figure S4 Isolation of mitochondria from COS7 cells incubated with polyfosine or edelfosine. Cells incubated with carrier (control) or 50 mM polyfosine or edelfosine for 1 h were harvested and homogenized. A crude mitochondria pellet was prepared as described under Material and Methods and separated from a cytosol-enriched supernatant (PMS, post mitochondrial supernatant). The crude mitochondria fraction (crude) was loaded onto a Percoll gradient and centrifuged again (150,0006g; 25 min) before 9 fractions were collected from the top. Aliquots were analyzed by SDS-PAGE and Western blotting for the mitochondrial proteins cytochrome c and cytochrome c oxidase subunit IV (COX IV) and calnexin, an ER marker protein. Note, that in contrast to control cells the mitochondria of polyfosine or edelfosine treated cells cannot be separated from the ER by density gradient centrifugation. (TIF) Figure S5 Disintegrated, blebby mitochondria and condensed nuclei of COS7 cells upon polyfosine treatment. Cells were incubated with 50 mM polyfosine for 5 h. For the last 15 min before fixation the incubation medium was supplemented with 20 nM Mitotracker dye whose accumulation to mitochondria depends on their activity and vitality. The vitality and morphology of mitochondria was analyzed by fluorescence microscopy. Merged color images show nuclei in green, mitochondria in red, stained by DAPI or Mitotracker, respectively. Bar, 50 mm.

(TIF)
Movie S1 UV light induced changes of cell morphology upon polyfosine treatment. COS7 cells were incubated with carrier (upper left) or 50 mM of lyso-ether PC (upper right), edelfosine (lower left) or polyfosine (lower right). Living cells were followed by phase contrast video microscopy for 33 min. During the course of the experiment the cells were constantly illuminated by broadband UV light from a second light source, a Xenon lamp. Bar, 20 mm. Note the fragmentation of mitochondria (lower left and right) and the disintegration of mitochondria and nuclei (lower right). (MOV) Text S1. (DOC)