A Functionalized Sphingolipid Analogue for Studying Redistribution during Activation in Living T Cells

Ceramide accumulates in cellular membranes as a result of de novo synthesis, breakdown of complex sphingolipids, including sphingomyelin (SM), or the salvage pathway. Ceramide molecules act as hubs in the sphingolipid pathway as they are readily converted by sphingomyelinases (SMases), ceramidases, ceramide kinase, or glucosyltransferases, thereby maintaining ceramide levels under homeostatic conditions (1, 2). In response to cellular stress or ligation of specific receptors, it is predominantly because of activation of SMases that ceramides condense into microdomains, thereby locally altering the biophysical properties of cellular membranes (3–6). Ceramide-enriched membrane microdomains segregate membrane receptors and their associated signalosomes, but also affect membrane curvature, thereby promoting vesicle exocytosis or endocytosis (4, 5, 7–9).

Availability of suitable tools is a prerequisite to visualize and analyze sphingolipid localization, trafficking, and lipid–lipid or lipid–protein interactions. Ceramide-specific Abs have been successfully used mainly on fixed cells to avoid artificial clustering of target molecules (10, 11). Synthetic analogues to visualize ceramide-enriched membrane domains in fixed and living cells are available, such as for 6-([N-(7-nitrobenz-2-oxa-1,3-dimensionaliazol-4-yl)amino]hexanoyl)sphingosine (NBD-C₆-cer), an established marker for visualizing the Golgi compartment (12, 13). These, however, bear bulky fluorophores in their fatty acid or sphingosine chains that perturb membrane architecture, limiting their potential use (14, 15).

For the study of biomolecules, the advent of click chemistry has brought about powerful new tools to probe trafficking, metabolism, or formation of higher molecular complexes (15–20). Click chemistry uses compounds containing terminal azido groups or alkynes allowing for bio-orthogonal reactions with suitable fluorophores. This means that the necessary modifications introduced into the biomolecules, including functionalized lipids, are minimally invasive, thus not interfering with their biology. So far, the use of alkyne compounds has mainly been restricted to fixed specimens, because the requirement of copper ions to catalyze the click reaction precluded experiments with living cells. More recently, a tetrazine-functionalized C₆-ceramide analogue (modified by a terminal trans-cyclo-octene group at the fatty acid) was successfully labeled under copper-free conditions and used to image the Golgi compartment without perturbing its function and architecture in living HeLa cells (21). Incorporation and labeling kinetics of this compound were, however, slow. This, of course, limits its potential use in live-cell experiments addressing ceramide accumulation, redistribution, and trafficking early after stimulation, which is especially important in primary cells. Thus, ceramide analogues rapidly integrating into cellular membranes and allowing for short, efficient bio-orthogonal click labeling in primary living cells have so far not been available. We therefore synthesized azido-functionalized ceramides (ω-azido-C₆-ceramide [N₃-C₆-cer] or ω-azido-C₁₆-ceramide [N₃-C₁₆-cer]) and explored their incorporation into resting and stimulated T cells and their redistribution after T cell stimulation.

As for other cell types, exposure to high levels of synthetic exogenous C₂/C₆ ceramides also promotes apoptosis in T cells (22–24). Similarly, physiological acid SMase (ASM) activation after death receptor ligation followed by ceramide release interferes with relay of TCR signaling by abrogating store-operated Ca²⁺ entry, NF-AT activation, and IL-2 synthesis (25, 26). However, in other situations, activation of SMases appears to modulate T cell activation (27–29), indicating that both the amount and/or compartmentalization of ceramides at stimulatory interfaces needs to be tightly controlled. In support of this hypothesis, Ab-based detection studies have revealed that ceramides are largely excluded from the center of an established immune synapse (IS) in favor of the lamellum (29).

With this study we establish that our new azido-functionalized ceramides are nontoxic and are efficiently incorporated into membranes of living T cells, even after activation, where they can be detected as early as 5 min after labeling. We found N₃-C₆-cer in plasma membrane microdomains and proximal vesicles of resting T cells. After TCR, and especially CD28 ligation, it segregated into clusters indicating efficient sorting into known plasma membrane domains associated with T cell activation. Mirroring the Ab detection experiments, N₃-C₆-cer was efficiently excluded from the IS center and rather accumulated in the periphery of the established IS. This indicates that N₃-C₆-cer can be used for ceramide membrane domain compartmentalization studies in living cells.

Azido-sphingosines N₃-C₆-cer, N₃-C₁₆-cer, and α-azido-C₁₆-ceramide were prepared by adding a solution of azide modified fatty acid (1 eq.) in CH₂Cl₂ (60 ml per mmol sphingosine) dropwise to a mixture of sphingosine (1.00 eq.), hydroxybenzotriazole monohydrate (1.20 eq.), 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (1.20 eq.), and N,N-diisopropylethylamine (1.8 eq.) in dry CH₂Cl₂ (60 ml per mmol sphingosine) within 2 h at 0°C, followed by an 18-h stirring under nitrogen atmosphere. The solvent was removed and the residue was purified via column chromatography (CH₂Cl₂/MeOH 100:1).

Peripheral blood monocytic cells from healthy donors were subjected to Ficoll gradient centrifugation and were used as source for primary human T cells (enriched by nylon wool columns) or monocytes (separated by plastic adherence). Monocytes were differentiated into mature dendritic cells (DCs) by culture in RPMI 1640 containing 10% FCS and human GM-CSF (500 U/ml; Berlex)/IL-4 (250 U/ml; Miltenyi Biotec) for 3–6 d followed by a 24-h addition of LPS (100 ng/ml; Sigma) and subsequent loading with superantigen (1 μg/ml for 30 min; SEB; Sigma) before T cell conjugation. Primary human and Jurkat T cells were maintained in RPMI 1640/10% FCS. For stimulation, primary T cells were kept at high density (8 × 10⁶ cells/well of a 12-well plate) for 3–5 d, or, alternatively, were incubated with CD3-specific (clone UCHT-1) and/or CD28-specific Abs (clone CD28.2) (each 1 μg/ml) (both Becton Dickinson Biosciences Pharmingen) for 48 h. When indicated, cells were exposed to bacterial SMase for 30 min at 37°C (12.5 μM; Sigma).

A total of 2 × 10⁵ Jurkat or primary T cells were extensively washed and resuspended in HBSS containing NBD-C₆-cer (stock 5 mM in DMSO, if not stated otherwise, final 5 μM; Life Technologies), N₃-C₆-cer, or N₃-C₁₆-cer (stocks 10 mM in EtOH, if not stated otherwise, 25 μM final) and incubated for 30 min at room temperature (RT) and washed three times with HBSS. For click reactions, Click-IT Alexa Fluor 488 DIBO Alkyne (20 μM; Life Technologies) or DBCO-Sulfo-Cy5 (20 μM; Jena Bioscience) was added for 5 min, if not stated otherwise, and cells were kept at RT until microscopical analysis. Primary T cells followed the same feeding and labeling protocol: when indicated, the feeding buffer was supplemented with Pluronic F-127 (20% in DMSO [VWR International], final concentration 0.25%) and when indicated, were preincubated with anti-CD3/CD28 Abs on ice and transferred onto slides precoated with 10 μg/ml anti-mouse IgG (Dianova) (1 h at 37°C), added to anti-CD3/CD28–coated dynabeads (Invitrogen) or conjugated to DCs at a ratio of 4:1 by capture on poly-l-lysine–coated slides (μ-Slide 8 well, ibidi) or 8-well coverslips (LabTekII; Nunc).

Confocal laser scanning microscopy imaging was performed using an LSM 780 (Zeiss, Germany), equipped with an incubation system and a 40× Plan-Apochromat oil objective (NA 1.4) and laser lines 488 and 633. Alternatively, LSM 700 (Zeiss, Germany) equipped with a 63× 1.4 oil objective and a 639 laser line was used. Images were processed using confocal laser scanning microscopy software ZEN2012. When indicated, z-stacks were acquired and three-dimensional reconstruction was done with a ZEN software tool. For cluster quantification, images were taken after 15 min of Ab stimulation on the costimulatory slides at the z-height of maximal cell spreading, and clusters per cell were counted. Pictures were analyzed using ImageJ, and clusters were defined as signals revealing a higher intensity as the membrane ring. For statistical analysis of cluster frequencies, at least 150 cells per condition from at least three donors were analyzed. GraphPad 6.0 software was used for graphs and statistical analysis.

Structured illumination microscopy (SIM) imaging of stained Jurkat T cells was performed using a Zeiss ELYRA S.1 system (Zeiss). SIM images were acquired with an inverted Axio Observer Z1 microscope equipped with a Plan-Apochromat oil-immersion objective (63×, NA 1.4). DBCO-Sulfo-Cy5 was excited with a 642-nm solid-state laser and recorded with a PCO.edge 5.5 sCMOS camera (PCO). Z-stacks were generated with three grid rotations and five phases for each z section, separated by 100 nm. For SIM calculations, the software Zen 2012 SP1 was used. Image processing was realized with Fiji software.

For detection of cell death, cells were incubated with PI Staining Solution (5 μl/test; eBioscience) or Viability dye 780 (1:1000; eBioscience) after treatment with functionalized lipids and immediately analyzed by FACSCalibur. For detection of ceramide incorporation, T cells were fed with unmodified (C₆-cer, C₁₆-cer; both Avanti Polar Lipids), functionalized (N₃-C₆-cer, N₃-C₁₆-cer), or NBD-C₆-cer at the concentrations indicated for 30 min at RT, washed and stained 30 min on ice using anti-ceramide IgM (1:1000; Enzo Life Science) followed by Alexa 488–conjugated anti-IgM (1:1000; Thermo Fisher). When indicated, cells were DIBO488-clicked 5 min at RT. For detection of CD69, primary T cells were left untreated or fed with each 25 μM ceramide (unmodified or N₃-functionalized) for 30 min at RT, washed stimulated for 16 h with 1 μg/ml anti-CD3/CD28 or PMA/ionomycin (or not), and stained using anti-CD69–allophycocyanin (1:50; Immunotools Friesoythe) for 45 min on ice. Incorporation of functionalized ceramides was controlled after the 16-h stimulation period by DIBO488-clicking. Ca²⁺ mobilization was detected in primary T cells fed with 25 μM ceramide (unmodified or N₃-functionalized) for 30 min, washed, loaded with Fluo-4AM (1 μM; Molecular Probes, Thermo Fisher) for 30 min, 30 min de-esterified (both at 37°C), and stimulated with 20 μg/ml anti-CD3/CD28 (preligated with goat anti-mouse IgG) by flow cytometry by gating on living cells (>94% in all populations analyzed).

Ceramide-treated cells were collected and dissolved in 1 ml methanol by sonification on ice for 30 min. After centrifugation (15,000 × g, 5 min), the organic phase was transferred into glass tube and evaporated using a vacuum system (Thermo Fisher Scientific, Waltham, MA). The dried lipids were then resolved in methanol, and sample analysis was carried out by rapid-resolution liquid chromatography–tandem mass spectrometry (MS) using a Q-TOF 6530 mass spectrometer (Agilent Technologies, Waldbronn, Germany) operating in the positive electrospray ionization mode. The compounds were separated by reverse-phase liquid chromatography using a 2.1 × 150 mm Zobrax Eclipse Plus C8 column (Agilent Technologies) and a binary solvent system at a flow rate of 0.5 ml/min. Before injection of the samples, the column was equilibrated for 1 min with a solvent mixture of 40% mobile phase A (H₂O/HCOOH, 99.9/0.1 v/v) and 60% mobile phase B (CH₃OH/C₂H₃N/HCOOH, 49.95/49.95/0.1, v/v/v), and after sample injection, the A/B ratio was maintained at 40/60 for 3 min, followed by a linear gradient to 95% B over 9 min, which was held at 95% B for 3 min, followed by a 1-min gradient return to 40/60 A/B. N₃-C₆-cer, C₆-cer, N₃-C₁₆-cer, and C₁₆-cer were detected via accurate mass of m/z 439.3643, m/z 398.3629, m/z 579.5208, and m/z 538.5194, respectively, and quantified by the use of an external calibration curve. Quantification was performed with Mass Hunter Software (Agilent Technologies).

Statistical analysis was performed using unpaired Student t test with ****p < 0.0001, ***p < 0.001, **p < 0.01, *p < 0.05.

To obtain ceramides for bio-orthogonal labeling in living cells, we synthesized N₃-C₆-cer (Fig. 1B) in which the azide group can be linked to a chemical reporter via bio-orthogonal click chemistry. N₃-C₆-cer was synthesized from ω-azido hexanoic acid and sphingosine via carbodiimide-mediated amide coupling. We also synthesized N₃-C₁₆-cer (Fig. 1C) and α-azido-C₁₆-ceramide (Fig. 1D) for use in some experiments as outlined later (30).

A first set of experiments comparatively analyzed the toxicity induced by the functionalized ceramide (N₃-C₆-cer) and by NBD-C₆-cer (Fig. 1A) in Jurkat and primary T cells over time using a constant concentration of the compounds (25 μM). Although NBD-C₆-cer caused a substantial amount of cell death already after 10 min of incubation, this was not observed with N₃-C₆-cer even after longer exposure (Fig. 2A). Moreover, toxicity of NBD-C₆-cer, but not the functionalized compounds, appeared to be concentration dependent in both Jurkat and primary T cells (Fig. 2B, t = 30 min). Notably, toxicity of the functionalized compounds did not exceed that of their unmodified counterparts under any of the conditions tested (Fig. 2B, upper panel). To select functionalized dyes suitable for click-reactions in living T cells, we separately incubated T cells with 11 different dyes for 30 min at 37°C and analyzed by confocal microscopy. Although some dyes readily penetrated into the T cells and accumulated in large aggregates, others were detected in dying but were excluded from living T cells (Fig. 3A, Table I). From the 11 dyes tested, only DIBO488 and DBCO-Sulfo-Cy5 were selected for bio-orthogonal click-reactions, which were performed under copper-free conditions in living cells. Clicking of DIBO488 or DBCO-Sulfo-Cy5 to the compounds revealed no dye-related differences with regard to staining patterns (data not shown). Detection of the click-labeled compounds appeared to faithfully reflect the amount of compound incorporated because labeling intensities were dose-dependent and unconjugated (not clickable) alkyne dye, but not conjugated DBCO-Sulfo-Cy5 dye diffused into the cytosol (exemplified for N₃-C₆-cer in Jurkat cells; Fig. 3B, 3C). Finally, the clicked products did not reveal enhanced cytotoxicity as determined in Jurkat T cells after 120 min of incubation (Fig. 2A).

We next established the incorporation kinetics and efficiencies of the functionalized ceramides in living target T cells with the feeding period followed by 5-min fluorophore coupling. Using flow cytometry, we compared NBD-C₆-cer– and N₃-C₆-cer–fed Jurkat T cells showing that N₃-C₆-cer highly efficiently accumulated in these cells with 25 min of feeding giving optimal results (Fig. 4A). We validated our flow cytometry data by mass spectrometry confirming that N₃-C₆-cer highly efficiently incorporated into both Jurkat and primary T cells within 25 min (roughly 70% as determined on a pmol/sample basis; Fig. 4B).

To comparatively analyze incorporation efficiencies of NBD-C₆-, unmodified (C₆-cer and C₁₆-cer), and functionalized ceramides, these were fed to primary T cells for 30 min at the concentrations indicated and then detected by Ab staining (unmodified ceramides), a 5-min DIBO488-click (functionalized ceramides), or directly (NBD-C₆-cer) by flow cytometry. Only C₆-cer, but not C₁₆-cer, was found to be incorporated by Ab-mediated detection (as compared with the EtOH solvent control) (Fig. 4C). This indicates that C₁₆-cer might be inefficiently incorporated within this time interval. In contrast, both functionalized and NBD-C₆-cer were well incorporated with signals decreasing to control levels (DIBO488 with or without DMSO) upon dilution. Again, incorporation levels of N₃-C₁₆-cer at 25 μM were substantially lower than those of the short-chain ceramides (Fig. 4C). Incorporation of C₁₆-ceramides was also lower than that of C₆-ceramides when determined by MS in both Jurkat and primary T cells (Fig. 4D). Interestingly, functionalization enhanced accumulation of C₆-ceramides but strongly reduced that of C₁₆-ceramides as compared with that of the unmodified controls (Fig. 4D).

After 30-min feeding with N₃-C₆-cer and subsequent DBCO-Sulfo-Cy5 clicking, specific signals could be detected in the plasma membranes, as well as in intracellular compartments of Jurkat T cells, whereas preclicked N₃-C₆-cer (dye clicked before feeding) primarily accumulated in intracellular compartments (Fig. 5A). To exclude that specific signals were derived from incorporation of N₃-C₆ fatty acid alone (potentially generated as a result of N₃-C₆-cer breakdown), cells were exposed to the functionalized fatty acid, which did, however, only result in background signals (Fig. 5B). Feeding of Jurkat T cells with N₃-C₁₆-cer (azide group at the ω-position C₁₆) resulted in faint plasma membrane and intracytoplasmic signals (Fig. 5C), most likely reflecting its low incorporation (Fig. 4D). Interestingly, N₃-C₁₆-cer functionalized at the α-position C2 (α-azido-C₁₆-ceramide; Fig. 1D) could not be detected at all after dye-clicking (data not shown), most likely indicating that, in addition to incorporation efficiencies, accessibility of the reactive group is an important determinant for the suitability of functionalized compounds. Because of the low incorporation efficiencies of N₃-C₁₆-cer, studies addressing subcellular redistribution of our functionalized compounds were performed with short-chain ceramides only.

To analyze whether the incorporation pattern of the functionalized compounds is similar to that of ceramides, we monitored subcellular distribution of N₃-C₆-cer and NBD-C₆-cer over time in Jurkat T cells. NBD-C₆-cer (used at 5 μM because of its toxicity; Fig. 2) efficiently accumulated at the plasma membrane and in the Golgi compartment (Fig. 6A, upper row). However, even when used at this low concentration, we observed substantial morphological changes associated with loss of cell integrity after 75 min of incubation. This was not seen for N₃-C₆-cer (used at 25 μM), which otherwise gave similar results as the NBD-C₆-cer with regard to the kinetics of the observed distribution patterns (Fig. 6A, bottom row). In addition to staining the plasma membrane, N₃-C₆-cer accumulated in intracellular vesicular structures (Fig. 6B, Supplemental Fig. 1) and a compartment also targeted by the Golgi marker NBD-C₆-cer (Fig. 6C).

Compared with Jurkat T cells, unstimulated primary T cells incorporated less N₃-C₆-cer into their membranes (Figs. 4B, right panel, 7A). The faint N₃-C₆-cer–specific signals could, however, be increased by supplementation of Pluronic F-127 during the labeling step (Fig. 7A). Preactivation of primary T cells by culture at high cell density (31) also enhanced labeling efficiencies, indicating that activation-induced alterations of T cell membranes might facilitate uptake of our functionalized ceramides (Fig. 7C). In line with this hypothesis, N₃-C₆-cer labeling efficiencies were even better in T cells preactivated with anti-CD3/CD28 Abs than after high-density culture, and this also applied to that of N₃-C₁₆-cer labeling. In costimulated T cells, N₃-C₆-cer accumulated both at the plasma membrane and in cytosolic vesicles (Fig. 7C). Mirroring our results using Ab detection of ceramides (29), the clicked N₃-C₆-cer appeared in vesicles just beneath the contact plane in bead-stimulated T cells (Fig. 7C, arrow, Supplemental Fig. 2).

We aimed at developing a tool for monitoring ceramide redistribution in T cells upon stimulation. Therefore, we first investigated whether incorporation of functionalized ceramides would interfere with T cell activation. Primary T cells were fed with unmodified or functionalized C₆- or C₁₆-ceramides under standard conditions (25 μM, 30 min) and subsequently activated by anti-CD3/CD28 costimulation or PMA/ionomycin for 16 h. Stimulation-induced upregulation of CD69 was unaffected by the functionalized ceramides, indicating that they did not interfere with early T cell activation (Fig. 8A). This was confirmed by experiments addressing the ability of T cells exposed to unmodified or functionalized ceramides to mobilize Ca²⁺ in response to CD3/CD28 ligation (Fig. 8B).

Because ceramides are known to be produced and to redistribute also during T cell activation, we investigated whether the functionalized N₃-C₆-cer would cosegregate with endogenously released ceramides into activation-induced membrane clusters. For this, primary T cells fed with N₃-C₆-cer were kept on ice for 15 min and subsequently stimulated either with anti-CD3, anti-CD28, or anti-CD3/CD28 Abs, which were cross-linked for 15 min by anti-IgG coated onto a planar support. Concomitantly, cells were DIBO488 labeled at 37°C. Each stimulation protocol promoted formation of labeled clusters at or proximal to the plasma membrane (Fig. 9A, and anti-CD28–stimulated cells in the three-dimensional reconstruction, 9C). As also found for Jurkat T cells by SIM analysis (Fig. 6B, Supplemental Fig. 1), the majority of these clusters appeared as membrane-proximal vesicular structures possibly representing ceramide-enriched domains endocytosed from the plasma membrane (Fig. 9C). CD3 and CD3/CD28 costimulation caused a spreading response with formation of lamellar extensions in N₃-C₆-cer–fed cells, again indicating that ceramide feeding does not interfere with early T cell activation. Expectedly, ligation of CD28 alone did not induce cell spreading (Fig. 9A, cell boundaries marked). As revealed by quantitative analysis, frequencies of clicked N₃-C₆-cer clusters increased in stimulated as compared with unstimulated cells, indicating that the functionalized ceramide analogue cosegregated into ceramide-enriched activation clusters. This was particularly evident after ligation of CD28 (Fig. 9B), which promotes ASM activation and ceramide release in the outer leaflet of the plasma membrane (27). Supporting the hypothesis that the N₃-C₆-cer codistributed with endogenous ceramides released upon receptor ligation, pre-exposure of T cells to the ASM inhibitor amitriptyline largely prevented its redistribution into activation clusters after CD28 ligation (Fig. 9D).

To further corroborate that the membrane regions incorporating the N₃-C₆-cer indeed represented ceramide enriched domains, we exposed N₃-C₆-cer–fed T cells to bacterial SMase and detected its redistribution by concomitant click reaction. After bacterial SMase exposure, N₃-C₆-cer was recruited into large clusters, which were also seen for endogenous ceramides by Ab-based detection (Fig. 10A).

Moreover, ceramides have been shown to be excluded from the center of the IS formed between a DC and a T cell in favor of IS membrane-proximal vesicles and the lamellum of the conjugating T cell (29). This pattern was effectively mimicked when Jurkat or primary T cells containing the functionalized ceramide analogue were conjugated to costimulatory beads (pseudo-IS) or DCs (IS) (Fig. 10B, with N₃-C₆-cer–containing vesicles and its IS peripheral accumulation marked by arrows and arrowheads, respectively), revealing that N₃-C₆-cer indeed represents a suitable tool to study rapid ceramide segregation and subcellular trafficking in living cells.

Given the high dynamics of their membrane accumulation and redistribution, versatile, nontoxic tools for rapid detection and tracing of sphingolipids in living cells are urgently needed. For this, azido-functionalized analogues allowing for rapid labeling by bio-orthogonal click reactions in the absence of copper catalyzation appear ideally suited, and also because terminal clickable alkynes are stable and less likely to perturb physicochemical properties of biomolecules (18). With this study we show that our azido-functionalized ceramide N₃-C₆-cer fulfills these criteria in a biologically relevant system. Application of this analogue does not interfere with viability, and even more important, activation of T cells. Moreover, it reveals a subcellular distribution compatible with a bona-fide ceramide and is redistributed, most likely along with endogenous ceramides, upon T cell stimulation in a SMase-dependent manner. In contrast with the tetrazine-functionalized C₆-ceramide analogue recently used to image the Golgi in living HeLa cells (21), labeling and thereby detection of N₃-C₆-cer is rapid, thus rendering it a novel, highly efficient reagent to study ceramide redistribution in living cells early after stimulation. However, the functionalized long-chain ceramide (N₃-C₁₆-cer) proved to be very inefficiently incorporated under the conditions analyzed. Therefore, its ability to retain cell viability or responsiveness as a compound per se could not be evaluated nor could it be used to study stimulation-dependent redistribution (Figs. 2, 4, 5).

We routinely applied N₃-C₆-cer in a concentration of 25 μM and for time intervals indicated (Fig. 2), which, in our hands, was not toxic for Jurkat and primary T cells (unstimulated or preactivated). This concentration approximated levels that have previously been described to cause apoptosis in Molt-4 cells (C₆-ceramide) and in Jurkat T cells (C₂-ceramide), however, only after treatment for 4–6 h, respectively (32, 33). In addition, C₁₆-ceramide when incubated with Jurkat or primary T cells at 50 μM for 2 h did not cause detectable cell death, but a loss of stimulated membrane protrusions was observed (34). Although we cannot rule out interference with T cell viability or activation upon long-term feeding, we did not detect any impact on these parameters with the azide-functionalized analogues within the time frame and concentration used. As indicated by our dilution experiment, these compounds may also be effectively used for potentially more sensitive target cells (Fig. 3).

Although plasma membrane incorporation was still seen, the N₃-C₆-cer also accumulated in an intracellular compartment after prolonged feeding, where it could be codetected with NBD-C₆-cer, which, under the conditions used, marks the Golgi compartment (Fig. 6). In line with previous observations made with other cell types (35), NBD-C₆-cer also proved to be toxic when added to living Jurkat T cells even after 10 min of application, which was not the case for our compounds added at the same concentration (Fig. 2). Morphological changes we saw already with low concentrations of NBD-C₆-cer (Fig. 6A). It is unlikely that observed differences in toxicity depend on the presence or absence of a bulky fluorophore conjugate during the feeding period because preclicking our compounds before application did not substantially increase the percentage of propidium iodide⁺ cells even after a 2-h application period (Fig. 2).

As revealed for NBD-C₆-cer, which directly reaches the Golgi in a variety of cells and is mainly converted into NBD-glucosylceramide, NBD-SM, NBD–ceramide-1-phosphate, and NBD-capronic acid within 2 h, sphingolipid analogues serve as substrate for metabolizing enzymes (35). Because they are minimally modified analogues, alkyne lipids also proved to be reliable substrates for metabolizing enzymes (15); and though we did not address this issue directly, it is more than likely that N₃-C₆-cer is also further metabolized. This metabolization may also involve reacetylation catalyzed by the ceramide synthase, which would lead to the loss of the functionalized fatty acid moiety (Fig. 1). Because this would have to occur in the endoplasmic reticulum, this would not be a fast process and, therefore, most likely does not play a role for the short-term experiments carried out by us. This interpretation is further supported by the similarity in the incorporation pattern of N₃-C₆-cer and N₃-C₁₆-cer, which is not a classical substrate for reacetylation (Fig. 5C). In contrast with reacetylation, potential transformation of N₃-C₆-cer to SM in the lumenal Golgi by SM synthase, as shown to occur for NBD-C₆-cer (36), would not affect the integrity of the functionalized molecule.

We have applied the functionalized ceramide to a highly relevant, established stimulation system known to involve ceramide release in response to SMase activation to verify its suitability for application in cell biology experiments. As seen in Ab-based experiments, the analogue appeared to be recruited into clusters in Jurkat T cells and primary T cells, and it redistributed, most likely along with endogenously generated ceramides, into larger aggregates upon SMase activation induced by TCR ligation or CD28 ligation or upon exposure to bacterial SMase (Figs. 9A, 10A). As evidenced by efficient upregulation of CD69, Ca²⁺ mobilization and the spreading response after TCR stimulation (Figs. 8, 9A), the analogue did not abrogate signaling or cause loss of actin-based lamellar protrusions as known to occur after hyperactivation of SMases (29, 34). The compound appears to be excluded from the center of the IS and pseudo-IS (as seen with DCs and stimulatory beads, respectively; Fig. 10B), which is in agreement with findings made in fixed conjugates by Ab staining (29). As evidenced by the vesicular structures accumulating proximal to the IS (Fig. 7), ceramide exclusion from the IS center may involve endocytosis, which has so far not been shown to occur for a sphingolipid at this interface.

Given the importance of ceramide release and functional consequences thereof in response to ligation of receptors, but also cellular stress conditions, tools to study their dynamic redistribution and interaction with target proteins in living cells are needed that allow their rapid and sensitive detection. It is, however, of utmost importance to reveal that the tools generated truly cosegregate with these endogenous sphingolipids in terms of toxicity, localization to subcellular membrane compartments, and stimulation, but also SMase activation-dependent recruitment into activation clusters and sorting into vesicular compartments. Our azido-functionalized ceramide analogues proved to meet all these criteria using T cell activation as a highly relevant model system, where ceramide release has a significant regulatory role. Evidently, the azido-functionalized ceramides validated in this study can now be widely used as tools in studies addressing sphingolipid dynamics in response to receptor ligation on a variety of cells. With these options, our ceramide analogues will be highly instrumental to experimentally address important questions regarding dynamic reorganization and molecular interactions of ceramides in cellular membranes.

We thank Erich Gulbins, Jürgen Schneider-Schaulies, Niklas Beyersdorf, and Elita Avota for helpful discussions and Charlene Börtlein for excellent technical assistance.

The authors have no financial conflicts of interest.