Bifunctional Probes Reveal the Rules of Intracellular Ether Lipid Transport

Abstract

Ether glycerophospholipids bear a long chain alcohol attached via an alkyl or vinyl ether bond at the sn1 position of the glycerol backbone. Ether lipids play a significant role in physiology and human health. However, their cellular functions remain largely unknown due to a lack of tools for identifying their subcellular localization and interacting proteins. Here, we address this methodological gap by synthesizing minimally modified bifunctional ether lipid probes by introducing diazirine and alkyne groups. To interrogate the subcellular kinetics of intracellular ether lipid transport in mammalian cells, we used a combination of fluorescence imaging, machine learning‐assisted image analysis, and mathematical modelling. We find that alkyl‐linked ether lipids are transported up to twofold faster than vinyl‐linked species (plasmalogens), pointing to yet undiscovered cellular lipid transport machinery able to distinguish between linkage types differing by as little as two hydrogen atoms. We find that ether lipid transport predominantly occurs via non‐vesicular pathways, with varying contributions from vesicular mechanisms between cell types. Altogether, our results suggest that differential recognition of alkyl‐ and vinyl ether lipids by lipid transfer proteins contributes to their distinct biological functions. In the future, the probes reported here will enable studying ether lipid biology in much greater detail through identification of interacting proteins and in‐depth characterization of intracellular ether lipid dynamics.

The article introduces the first bifunctional ether lipids that were used for comparative studies of ester and ether lipid transport in mammalian cells. We find differential recognition of alkyl ether and vinyl ether lipids (plasmalogens) by lipid transfer proteins indicating distinct functional roles within cells. The new probes will advance the understanding of ether lipid biology in mechanistic detail.

Introduction

Eukaryotic ether lipids are primarily phosphatidylcholines (PCs) and phosphatidylethanolamines (PEs) featuring a long‐chain alcohol attached via alkyl (plamanyl) or vinyl (plasmenyl) ether bonds at the sn1 position of the glycerol scaffold.^{[ 1 ]} Between 10–20% of all glycerophospholipids in humans are ether lipids, with increased levels in brain, heart, kidney, and skeletal muscle tissues.^{[ 1 ]} Ether lipids are thought to have specific physiological roles despite the rather small structural deviation from the corresponding diacyl ester lipids. For example, disruption of peroxisomal ether lipid biosynthesis causes severe hereditary diseases, which are characterized by skeletal, renal, and cerebral abnormalities.^{[ 2 ]} Corresponding mice models show abnormal eye development, male infertility, and irregular behavior.^{[ 3} , ⁴ , ⁵ , ^{6 ]} Many studies also indicate a correlation between disturbed ether lipid metabolism and neurodegenerative diseases, psychiatric disorders, and cancer but the underlying molecular mechanisms are unclear.^{[ 1} , ⁷ , ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² , ^{13 ]}

Our knowledge of the cell biology of ether lipids is derived primarily from indirect evidence. Ether lipids have been proposed to play a role in membrane trafficking,^{[ 14 ]} the general architecture of the endomembrane system, lipid sorting in the cell, neurotransmitter release,^{[ 15 ]} and ferroptosis.^{[ 16} , ¹⁷ , ¹⁸ , ^{19 ]} Intriguingly, elevated levels of ether lipids were recently shown to be an adaptation of deep sea organisms to the high pressure environment.^{[ 20 ]} Perhaps the clearest evidence points to a role of ether lipids in modulating the export of GPI‐anchored proteins via a direct interaction with the transmembrane domain of mammalian p24 proteins, being metabolically co‐regulated with sphingolipids.^{[ 21 ]} Recent progress in ether lipid biology was mostly made by combining genetic or other perturbations such as oxidative stress with lipidomic approaches that now allow to distinguish between plasmenyl and plasmanyl lipids.^{[ 22 ]}

So far, very few ether lipid functions have been studied in mechanistic detail in biological assays. Thus, the possibility remains that the wide range of observed phenotypes are partially second‐order effects of genetic perturbations. The difficulties associated with mechanistic investigations into cellular functions of ether lipids can be explained by an almost complete lack of suitable chemical tools for studying ether lipids in situ. Apart from ether lipid analogues labeled with bulky fluorescent moieties,^{[ 23} , ^{24 ]} which are of limited use in cellular assays, the most promising approach has so far been the use of polyene lyso‐ether lipids, which are intrinsically fluorescent due to conjugated trans double bonds.^{[ 25} , ^{26 ]} Despite promising early results, their use has been limited, likely due to the relative complexity of the synthetic routes and issues with rapid photobleaching during microscopy. Further development of strategies for directly monitoring intracellular ether lipid localization, trafficking and interaction partners are needed.

Among the most versatile tools for studying lipid biology are minimally modified, bifunctional probes equipped with a photoactivatable (diazirine) and a clickable (alkyne) moiety. The diazirine group enables UV crosslinking to proximal proteins, whereas reporter groups can be attached to the alkyne group by copper‐mediated click chemistry. Bifunctional lipids have been used to study lipid‐protein interactions of fatty acyls, regular (diacyl) glycerophospholipids, sphingolipids, and sterols by utilizing affinity tags (e.g. biotin) and subsequent pulldown of lipid‐protein conjugates generated by crosslinking.^{[ 27} , ²⁸ , ²⁹ , ³⁰ , ³¹ , ³² , ³³ , ^{34 ]} In a limited number of examples, these probes have also been used to report on lipid localization and transport by attaching fluorophores after crosslinking and cell fixation and subsequent analysis by fluorescence microscopy.^{[ 29} , ³⁴ , ^{35 ]} Recently, we expanded upon this methodology by employing bifunctional phospholipids to quantify both lipid transport and metabolism using a combination of ultra‐high resolution mass spectrometry and fluorescence imaging.^{[ 36 ]} By comparing biophysical properties of bifunctional lipids to their native counterparts and comparing the metabolism of a bifunctional lipid probe with an isotope‐labeled probe in cells, we demonstrate that the combined effect of the diazirine and alkyne modifications on lipid behavior is small and equates to approximately one additional double bond.^{[ 36 ]}

We therefore reasoned that bifunctional ether lipid probes would be ideal tools to study ether lipid biology assuming that ether lipid specific lipid‐handling is captured by these near‐native lipid probes. As the primary probe of this study, we synthesized a PC ether lipid (1) featuring a bifunctional, alkyl ether‐linked sn1 chain in a 15‐step synthesis featuring a regioselective Sc(OTf)₃‐mediated epoxide opening. A complementary set of bifunctional phosphatidylcholines (2–5) that only differ in the sn1 linkage type via ester, alkyl ether, or a vinyl ether bond was generated to enable comparative studies of lipid transport in cells. We performed pulse‐chase experiments in two different cancer cell lines (HCT‐116 and U2OS) and iPSC‐derived neurons and found that the plasmalogen probe was consistently transported slower compared to both alkyl ether and ester lipid probes. Ether lipid transport was found to be primarily non‐vesicular in both investigated cell lines, but the fraction of material transported through the vesicular route was higher in U2OS versus HCT‐116 cells. Our data suggest that lipid transfer proteins distinguish plasmanyl and plasmenyl ether lipids, suggesting a plausible mechanism for the enrichment of distinct ether lipids in cellular membranes.

Results

Synthesis of Bifunctional Ether Lipid Probes

To study the influence of the linkage type on lipid transport, we synthesized bifunctional lipids with identical head groups and closely related side chain compositions. We focused on PC species as they are among the most abundant ether lipids in mammalian cells.^{[ 37 ]} As plasmalogens feature one unsaturation in the sn1 chain, we decided to use the monounsaturated oleic acid for the corresponding ester and alkyl ether version. Probes 1 and 2 were designed to assess the influence of alkyl ether versus ester linkage at the sn1 position (Figure 1a) as they are otherwise identical molecules. A complementary set of probes (3–5) was designed to perform a three‐way comparison of alkyl ether, vinyl ether, and ester linkages (Figure 1a). Together, the complete set of probes was designed to study the effects of the following structural elements that can plausibly affect transport mechanisms of ether lipids in cells: i) side chain linkage types (alkyl ether, vinyl ether, ester) and ii) positioning of a single double bond (sn1/sn2 chain, Δ9/Δ1).

Figure 1.

Overview of bifunctional ether lipid probes and synthesis of ePC(Y₁₈/18:1) (1) bearing a bifunctional ether linked side chain. a) Structures of synthesized bifunctional ether and ester PCs 1–5 differing in the linkage type at the sn1 position. b) Synthesis of the bifunctional ether lipid 1. The bifunctional alcohol 13 was synthesized in eight steps starting from 11‐Dodecynol (6) and coupled to the TIPS‐protected (S)‐glycidol 14. After oleic acid coupling and deprotection, phosphorylation according to Xu et. al.^{[ 38 ]} gave the final molecule 1. Reaction conditions: a) CrO₃, H₂SO₄, H₂O, acetone, 0 °C‐RT, 1 h, 85%; b) (COCl)₂, DMF, DCM, 1 h, 0 °C‐RT; c) 1‐morpholinocyclohexene, NEt₃, DCM, O.N., RT, 76% over 2 steps; d) aq. KOH, 15 min, 100 °C, 94%; e) LiAlH₄, THF, O.N., 0 °C ‐RT, 83%; f) DMTCl, pyridine, 3.5 h, RT, 95%; g) DMP, pyridine, DCM, O.N., RT, 82%; h) FeCl₃, MeOH, CHCl₃, O.N., RT, 90%; i) 1. ammonia (solution in MeOH), MeOH, 0 °C‐RT, 5.5 h. 2. H₂NOSO₃H, MeOH, O.N., 0 °C‐RT. 3. NEt₃, iodine, MeOH, 8 h, 16%; j) 14, 5 mol% Sc(OTf)₃, DCM, 2 d, RT, 59%; k) oleic acid, DMAP, EDC⋅HCl, DCM, O.N., 0 °C‐RT, 57%; l) TBAF, acetic acid, THF, O.N., 0 °‐RT, 32%; m) 1. NaBPh₄, water. 2. 2‐cyanoethyl‐N,N,N’,N’‐tetraisopropylphosphoroamidite, 1H‐tetrazole, MeCN, O.N., RT; n) 1. 18, 1H‐tetrazole, MeCN, 3.5 h, RT. 2. ^tBuOOH, MeCN, 1.5 h, 0 °C‐RT. 3. NEt₃, DCM, 20 h, RT, 47% over three steps.

The synthesis of the bifunctional lipids required two different strategies. For molecules 2–5, we first synthesized bifunctional fatty acids according to our recently published protocol (see details in Supporting Information).^{[ 36 ]} We then coupled the bifunctional fatty acids to the respective lyso‐lipids via EDC/DMAP‐mediated esterification (see Supporting Information for details). Probe 1, which bears the diazirine and alkyne functionalities in the ether chain at the sn1 position, required a 15‐step synthesis (Figure 1b, see Supporting Information for synthetic details and analytical data). We chose a synthetic route via the 1‐alkyl‐2‐acylglycerol 17 as this intermediate offers straightforward access to a range of other ether lipid classes via phosphoramidite chemistry.

The key step of the synthesis was the regioselective epoxide opening reaction used to attach the bifunctional long‐chain alcohol at the sn1 position of the glycerol scaffold. Similar transformations have been reported previously,^{[ 39} , ⁴⁰ , ^{41 ]} but the utilized reagents gave unsatisfactory yields in our hands, presumably due to the longer chain length and the presence of the diazirine and alkyne groups.

The installation of the two functional groups at the ether‐linked sn1 chain required the synthesis of a long‐chain alcohol 13 with an internal ketone and a terminal alkyne. Test reactions indicated that the yields of a Grignard reaction with ε‐caprolactone^{[ 42 ]} as well as a Rh‐catalyzed C–C‐coupling reaction^{[ 43 ]} were not sufficient (Scheme S3). Thus, we first synthesized a long‐chain fatty acid 8 bearing the internal ketone and a terminal alkyne. To generate the alcohol 13, we initially attempted to selectively reduce the carboxylic acid in the presence of the ketone. The use of mild reducing reagents, such as NBu₄BH₄, after activation of the carboxylic group via an acyl chloride only gave moderate yields (Table S4). The best overall yield was achieved with a complete reduction of carboxylate and ketone to the respective hydroxy groups, followed by a selective protection of the primary hydroxy group with DMTCl. After oxidation of the secondary hydroxy group with Dess–Martin–Periodinane and deprotection via FeCl₃, the long‐chain alcohol 12 was isolated in a yield of 58% over four steps. We note that the alternative route via ketone acetalization, reduction of the carboxylate and deprotection was also feasible but gave a lower overall yield. After introducing the diazirine group, the resulting bifunctional alcohol 13 was coupled to a glycerol backbone via an epoxide opening reaction to afford intermediate 15. Many different reagents, both bases and Lewis‐acid catalysts, have been described in the literature to afford high yields in similar transformations.^{[ 40} , ⁴¹ , ⁴⁴ , ⁴⁵ , ⁴⁶ , ⁴⁷ , ⁴⁸ , ⁴⁹ , ⁵⁰ , ^{51 ]} In our hands, the epoxide opening worked best with catalytic amounts of Sc(OTf)₃ (5 mol%) in dry DCM,^{[ 51 ]} whereas most published protocols gave either very low yields or no product at all (Table S5). The bifunctional intermediate 15 was isolated in 59% yield under optimized reaction conditions (3.5 equiv. of epoxide 14 added, longer reaction time and addition of a second batch of catalyst after 24 h). After subsequent esterification with oleic acid using EDC/DMAP, the silyl protection group at the sn3 hydroxy group was removed using TBAF. We note that the addition of acetic acid was necessary to minimize acyl chain migration from the sn2 to the sn3‐position. The synthesis was concluded by attaching the phosphocholine headgroup to intermediate 17 using phosphoramidite 18.^{[ 34 ]} After tetrazole‐mediated coupling, the phosphor atom was oxidized with ^tBuOOH and the cyanoethyl protecting group was removed with NEt₃ to obtain the bifunctional alkyl ether PC 1 in an overall yield of 0.3%.

Ether Lipid Transport Depends on the Linkage Type and Position of the Bifunctional Chain

With the set of bifunctional ether lipids in hand, we first asked how the linkage type at the sn1 position influences the intracellular transport behavior of PCs. We monitored ether lipid transport in HCT‐116 and U2OS cell lines. The HCT‐116 line was chosen as its origin tissue is involved in lipid resorption, and it furthermore features a relatively intact genome with few of the alterations present in most cancer lines. The U2OS line is the most frequently used line in the field of membrane contact sites and lipid transport due to its flat shape, which facilitates high‐resolution microscopy. Bifunctional ether lipid probes were incorporated into the outer leaflet of the plasma membrane of living cells by incubation with probe‐containing liposomes and α‐methylcyclodextrin.^{[ 36 ]} After the 4 min loading pulse, the loading mix was replaced with cell growth medium, and the cells were placed at 37 °C to enable retrograde lipid transport. UV‐crosslinking (10 s at 310 nm) after the respective chase times was followed by cell fixation and detergent treatment, achieving both cell permeabilization and the removal of noncrosslinked lipid probes. The generated covalent lipid‐protein conjugates were subsequently fluorescently labeled via Cu‐mediated click chemistry using an Alexa Fluor 594 picolyl‐azide dye and lipid localization was assessed by fluorescence microscopy (Figure 2a).

Figure 2.

Fluorescence signal derived from bifunctional probes is specific and allows to analyze ether lipid localization after intracellular transport. a) General workflow for pulse‐chase experiments to monitor ether lipid transport with fluorescence microscopy. b) Cellular localization and signal intensity of bifunctional lipids 1–5 in HCT‐116 cells after a 4 min pulse with and without UV irradiation. Lipid‐protein conjugates were labeled with an Azide‐AF594 dye via copper‐catalyzed click chemistry. The plasma membrane was labeled with NHS‐PEG4‐Biotin. The +UV and −UV images for the same probe are shown using identical settings, +‐UV image pairs are brightness‐contrast adjusted for better comparability as the signal intensity between lipid probes varied. Scale bar is 10 µm. c) Quantification of lipid signal intensity at 4 min for all probes (10–20 fields of view each with 20–50 cells) and control (3 field of views each with 20–50 cells) conditions. The mean fluorescence intensity of the cell‐covered area is shown; images were processed using a Gaussian blur (sigma = 4) filter and background corrected. The box plot (Python seaborn library) shows the mean and SD. A permutation test from the Python library Mlxtend (method = approximate, seed = 0, number of rounds = 1000) was used to test the significance. d) Colocalization of lipid signal with plasma membrane, ER, and Golgi markers allows to assign the lipid signal to the respective organelles. The plasma membrane was labeled with MemBrite®Fix 405/430. The ER and Golgi were labeled using immunofluorescence. Upper panels: Raw data of all four channels. Lower panels: Background corrected lipid signal and lipid signal assigned to the plasma membrane, the ER, and the Golgi, respectively. Scale bar is 10 µm. e) Cellular localization of pPC(18:1/C₁₆) (5) at 10, 60, and 120 min after lipid loading with co‐staining of the plasma membrane, ER, and Golgi. The plasma membrane was labeled with MemBrite®Fix 405/430. The ER and Golgi were labeled using immunofluorescence. Scale bar is 10 µm. Lipids were loaded in a α‐methyl‐cyclodextrin mediated lipid exchange reaction with bifunctional lipid‐containing liposomes (1.5 mM bifunctional lipid, 0.75 mM POPC, 0.75 mM cholesterol). P‐values are represented by asterisks indicating significance levels n.s.: p > 0.05, *: p ≤ 0.05, **: p ≤ 0.01, ***: p ≤ 0.001, ****: p ≤ 0.0001.

We first confirmed lipid incorporation into the plasma membrane by analyzing the colocalization of lipid signal with a plasma membrane marker in HCT‐116 cells after the 4 min loading pulse (Figure 2b). Control experiments without lipid loading and/or without UV irradiation demonstrate that the observed fluorescence signal is specific to the covalent lipid‐protein conjugates as very low background signal was observed. (Figure 2b lower panels, Figures 2c and S2–S7). We observed that both bifunctional lipids modified in the sn1 position (probes 1+2) exhibited a significant lower overall fluorescence intensity in contrast to their sn2‐functionalized counterparts (probes 3–5) (Figure 2c). This could be due to different crosslinking probabilities caused by the chain length difference or a higher likelihood of proteins to interact with the sn2 side chain.

In order to analyze lipid transport processes, we performed four‐color fluorescence microscopy experiments with iterative immunofluorescent co‐staining for the plasma membrane, the endoplasmic reticulum, the Golgi, and endosomes (Figure S1). We quantified the lipid signal in each organelle using our recently reported image analysis pipeline based on machine learning‐assisted image segmentation.^{[ 36 ]} Briefly, self‐trained Ilastik models were used to obtain probability masks to assign the fluorescence signal in the lipid channel to individual organelles and quantify lipid distribution in the cell (Figure 2d, see Supporting Information for details of image background correction and image analysis).

To quantify changes of lipid distribution over time, we performed pulse‐chase experiments with all lipid probes (1–5) covering a time period of 120 min in HCT‐116 cells (Figure 2e). Lipid signal intensity in the plasma membrane decreased over time for all lipids, indicating probe internalization via retrograde trafficking (Figure 3a,b, no UV‐controls in Figures S2–S7). A comparison between lipid probes revealed that the plasmalogen probe pPC(18:1/Y₁₆) (5) was retained at the plasma membrane to a larger extent compared to both alkyl ether and ester lipid probes (Figure 3a,b). Upon internalization, most of the signal for all probes was detected within the ER (Figure 3c). Apart from a brief spike at 20 min, very little fluorescence signal was assigned to endosomal structures (Figure 3d). Interestingly, all probes bearing the bifunctional fatty acid at the sn2 position (probes 3–5) exhibited more Golgi localization after 30 min (Figures 3a,e and S9). This suggested that intracellular lipid transport routes are specific to the presence of the double bond within the sn1 chain. At 120 min, the plasmalogen was more enriched in the Golgi (Figure 3e) implying a differential steady state distribution between the probes.

Figure 3.

Plasmalogens exhibit slower retrograde transport kinetics than the corresponding alkyl ether and ester lipid probes in HCT‐116 cells. a) Cellular localization of bifunctional lipids 1–5 in HCT‐116 cells at 4, 10, 20, 30, 60, and 120 min after lipid loading. The lipid was labeled with an Azide‐AF594 dye via copper‐catalyzed click chemistry. Scale bar is 10 µm. b)–e) Quantification of the relative lipid signal in the plasma membrane, the ER, endosomes, and the Golgi apparatus. The line plots were made with the Python seaborn library and show the mean and SE of each timepoint. Data is derived from six independent time points each containing 4–18 fields of view per lipid with 20–50 cells. f) Kinetic model for quantifying lipid transport from fluorescence microscopy data. Black arrows indicate vesicular transport and red arrows non‐vesicular transport. g) The corresponding model fits of the lipid transport kinetics for all lipids derived from 100 model runs. h) Comparison of rate constants describing retrograde vesicular transport from the plasma membrane to endosomes, retrograde non‐vesicular transport from the PM to the ER and total transport in the anterograde direction from the ER to the PM. Error bars: SD. i) Ratio of the rate constants for non‐vesicular lipid transport between the plasma membrane and the ER in the retrograde and anterograde direction. Error bars: SD. Lipids were loaded in a α‐methyl‐cyclodextrin mediated lipid exchange reaction with bifunctional lipid‐containing liposomes (1.5 mM bifunctional lipid, 0.75 mM POPC, 0.75 mM cholesterol).

To obtain a quantitative description of the lipid transport kinetics and to test the contribution of vesicular and non‐vesicular transport routes, we fitted a kinetic model to the obtained time‐course data (Figure 3f,g see Supporting Information for model details). The model distinguishes non‐vesicular transport from the plasma membrane to the ER from retrograde vesicular transport (plasma membrane via endosomes and the Golgi to the ER). Anterograde lipid transport is described by a summary rate constant from the ER to the plasma membrane capturing both vesicular and non‐vesicular transport modes. We found that transport via the non‐vesicular route in the retrograde direction (captured by the rate constant k_PM‐ER) was much faster than vesicular membrane trafficking (k_PM‐Endo) for all lipids (Figure 3h, Table S3). The alkyl ether and ester lipid probes bearing the bifunctional moieties at the sn1 chain (probes 1+2) were transported faster than the corresponding sn2‐modified probes (probes 3–5), which is likely due to both chain length and positioning. However, the linkage type via alkyl ether or ester had only negligible effects on transport kinetics. In contrast, the vinyl ether linkage of the plasmalogen 5 led to overall much slower transport. The steady‐state distribution of the plasmalogen probe pPC(18:1/Y₁₆) (5) was shifted toward the plasma membrane compared to the corresponding alkyl ether (ePC(18:1/Y₁₆) (3) and ester lipid (PC(18:1/Y₁₆) (4) probes (quasi‐equilibrium constants k_PM‐ER/k_ER‐PM, Figure 3i). In general, we observed that the effect of plasmalogen‐vinyl ether linkage supersedes the effect caused by the placement of the double bond in the sn1 or sn2 acyl chain, whereas the latter appears to be stronger than the difference between alkyl ether and ester linkage types.

Ether Lipid Transport Shows Similar Trends in Immortalized Cell Lines and iPSC‐derived Neurons

To assess whether the trends observed for ether lipid transport in HCT‐116 cells are more generally conserved, we conducted lipid transport experiments in U2OS cells. The two cell lines differ significantly in origin (bone marrow and colon) and morphology. U2OS cells are large, flat cells resembling mesenchymal cells, whereas HCT‐116 cells are smaller and retain most structural features of their epithelial origin. We used the sn2‐bifunctional lipid probes in pulse‐chase experiments to assess potential differences in lipid transport (negative control images in Figure S8). Endpoint quantification at 30 min showed that the plasmalogen was retained at the plasma membrane to a larger extent than both the corresponding ester and alkyl ether probes (Figure 4a,b). This effect was also reflected by lower plasmalogen lipid signal in the ER at all timepoints (Figure 4a,b). Overall, the relative differences between individual lipids were more pronounced for U2OS cells compared to HCT‐116 cells.

Figure 4.

Retrograde vesicular transport of ether lipids is cell‐ and linkage type specific. a) Cellular localization of sn2‐bifunctional lipid probes at 4, 15, and 30 min in U2OS cells with co‐staining of the plasma membrane and the ER. Scale bar is 10 µm. b) Quantification of the relative lipid intensity in the plasma membrane and the ER. The plots were made with the Python seaborn library. The line plots show the mean and SE and the box plots the mean and SD. Mean, SD, and SE of each timepoint were derived from 24–49 fields of view per lipid each with 8–12 cells. A permutation test from the Python library Mlxtend (method = approximate, seed = 0, number of rounds = 1000) was used to test the significance at the 30 min timepoint. c) Cellular localization of ePC(18:1/Y₁₆) (3) in U2OS and HCT‐116 cells at the bottom plane after 30 min with co‐staining of the Rab5 and Rab7 proteins. Scale bar is 10 µm. d) Quantification of the relative lipid signal in endosomes of the sn2‐bifunctional lipid probes in U2OS and HCT‐116 cells after 30 min. The box plot (Python seaborn library) shows the mean and SD. e) Comparison of ratios between the kinetic rates for non‐vesicular (k_PM‐ER) and vesicular (k_PM‐Endo) transport of PC(18:1/Y₁₆) (4) in HCT‐116 (this work) and U2OS cells.^{[ 36 ]} Error bars: SD. f) Cellular localization of the sn2‐bifunctional lipid probes in U2OS cells at 15 min with co‐staining of the Rab5 and Rab7 proteins. Scale bar is 10 µm. g) Quantification of the relative lipid signal in endosomes. The plots were made with the Python seaborn library. The line plot shows the mean and SE and the box plot the mean and SD. Mean, SD, and SE of each timepoint were derived from 12–33 fields of view per lipid with 8–12 cells. A permutation test from the Python library Mlxtend (method = approximate, seed = 0, number of rounds = 1000) was used to test the significance at the 15 min timepoint. h) Cellular localization of ePC(18:1/Y₁₆) (3) and pPC(18:1/Y₁₆) (5) at 4 min in iPSC‐derived neurons. Scale bar is 10 µm. Lipids were loaded in a α‐methyl‐cyclodextrin mediated lipid exchange reaction with bifunctional lipid‐containing liposomes (1.5 mM bifunctional lipid, 0.75 mM POPC, 0.75 mM cholesterol). P‐values are represented by asterisks indicating significance levels: n.s.: p > 0.05, *: p ≤ 0.05, **: p ≤ 0.01, ***: p ≤ 0.001, ****: p ≤ 0.0001.

One notable difference between the cell lines was bright vesicular structures in the lipid image observed in U2OS but not in HCT‐116 cells. These structures were identified as endosomes with immunofluorescence (Figure 4c). The quantification of the lipid signal fraction in endosomes confirmed that a larger amount of lipid material was routed through the endocytic route in U2OS cells (Figure 4d). This observation suggests that the non‐vesicular transport route is even more dominant in HCT‐116 cells compared to U2OS cells, which is also reflected in the difference between rate constants of PC(18:1/Y₁₆) (4) describing endocytosis and non‐vesicular transport from the plasma membrane to the ER in the retrograde direction (Figure 4e).^{[ 36 ]} A closer inspection of lipid incorporation in endosomes in U2OS cells revealed a slower uptake of the plasmalogen probe, which is also reflected in a slower appearance of lipid signal in the Golgi (Figures 4f,g and S10). This effect is not caused by a change in the bulk endocytic rate induced by probe loading as the number and size of endosomes were found to be constant for all probes (Figure S11). Taken together, while some cell line‐specific differences were apparent, the major trends for ether lipid transport in both cell lines were similar, in particular with regard to slower internalization kinetics of the plasmalogen probe and predominant non‐vesicular transport.

To study if the same trend holds in more physiologically relevant systems, we additionally visualized ether lipid localization in iPSC‐derived neurons. We observed that ePC(18:1/Y₁₆) (3) was internalized faster and pPC(18:1/Y₁₆) (5) was retained longer at the plasma membrane, which is in line with our observations in HCT‐116 and U2OS cells (Figures 4h and S12). This strongly supports the hypothesis that a specific, cell type independent recognition mechanism of the phospholipid sn1‐linkage type exists.

Discussion

We report a new set of bifunctional photoaffinity probes for studying ether lipid biology. We find notable differences in transport speed for plasmalogen and alkyl ether lipids, suggesting that the lipid transport machinery indeed exhibits specificity for distinct ether lipid types. While very little is known about the cellular roles of individual ether lipid species, our findings imply distinct functions of plasmalogen versus alkyl ether and ester lipids. The effect sizes found here are smaller compared to more structurally diverse lipids with different headgroups and saturation degrees, but nonetheless significant.^{[ 36 ]} These data imply that small structural differences are recognized by the cellular lipid handling machinery, providing further evidence for the biological importance of lipid diversity.^{[ 52 ]}

The utilized modular synthesis employing the long chain bifunctional alcohol 13, stereoselective Sc(OTf)₃ catalyzed epoxide opening, and phosphoramidite headgroup coupling enables straightforward synthetic access to further bifunctional ether lipid classes. This opens the way to study the biology of ether lipids both with regard to detailed analyses of individual lipid species as well as to expand the scope of investigations to the broader ether lipidome. Furthermore, while this study is focused on analyzing ether lipid transport by fluorescence imaging, bifunctional lipids can also be employed for tracing metabolism and identifying lipid‐protein interactions.^{[ 29} , ³¹ , ³⁵ , ^{36 ]} Taken together, the bifunctional ether lipid probes reported here represent a versatile toolkit for studying ether lipid biology in mechanistic detail. This capability will have significant benefits for understanding the functions of ether lipids in fundamental cell biology and their role in human diseases.

Conflict of Interests

The authors declare no conflict of interest.

Supporting information

Supporting Information

Böhlig K., Iglesias‐Artola J. M., Asaro A., Lennartz H. M., Link A. C., Drobot B., Nadler A., Angew. Chem. Int. Ed. 2025, 64, e202513360. 10.1002/anie.202513360

Data Availability Statement

The data that support the findings of this study are openly available in Bioimage Archive at https://www.ebi.ac.uk/biostudies/bioimages/studies/S‐BIAD1293, reference number 1293.