Synthesis and Characterization of Ceramide-Containing Liposomes as Membrane Models for Different T Cell Subpopulations

A fine balance of regulatory (Treg) and conventional CD4+ T cells (Tconv) is required to prevent harmful immune responses, while at the same time ensuring the development of protective immunity against pathogens. As for many cellular processes, sphingolipid metabolism also crucially modulates the Treg/Tconv balance. However, our understanding of how sphingolipid metabolism is involved in T cell biology is still evolving and a better characterization of the tools at hand is required to advance the field. Therefore, we established a reductionist liposomal membrane model system to imitate the plasma membrane of mouse Treg and Tconv with regards to their ceramide content. We found that the capacity of membranes to incorporate externally added azide-functionalized ceramide positively correlated with the ceramide content of the liposomes. Moreover, we studied the impact of the different liposomal preparations on primary mouse splenocytes in vitro. The addition of liposomes to resting, but not activated, splenocytes maintained viability with liposomes containing high amounts of C16-ceramide being most efficient. Our data thus suggest that differences in ceramide post-incorporation into Treg and Tconv reflect differences in the ceramide content of cellular membranes.


Introduction
Besides phosphoglycerides and cholesterol (CHOL), bioactive sphingolipids are the main structural components of the eukaryotic cell membrane [1]. In particular, ceramide, which serves as a building hub for complex sphingolipids, influences the biophysical properties of the plasma membrane and organizes the formation of membrane microdomains that are crucial for receptor clustering [2]. While these processes fundamentally impact the biology of many cell types, sphingolipid metabolism in T cells is of particular importance due to their broad involvement in diseases as different as cancer and infections, including COVID-19, and their accessibility to therapeutic intervention [3].
In T cells, ceramide-enriched microdomains are known to modulate signaling of various cell surface receptors including the T cell receptor complex [4]. Moreover, T cell subpopulations differ regarding their ceramide content. For instance, CD4 + Foxp3 + regulatory T cells (T reg ) protecting the body from autoimmunity and immunopathology have been shown to have a much higher ceramide content than CD4 + Foxp3 − conventional (T conv ) T helper cells [5,6]. In humans, the incorporation of clickable ceramide is positively correlated with effector/memory differentiation of both T reg and T conv [7]. While the determination of cellular ceramide content using mass spectrometry reflects ceramide content at the population level, minimally modified ceramides suitable for click chemistry allow the ceramide content to be determined at the single-cell level using flow cytometry, fluorescence, or confocal microscopy [8,9]. After the incubation of cells with azide-functionalized ceramide, the incorporated ceramide can be visualized with reagents such as the Click-IT Alexa Fluor 488 DIBO Alkyne (DIBO Alexa Fluor 488) [10]. Therefore, the detection of incorporated azide-functionalized ceramide can easily be combined with multi-color flow cytometry, covering more than ten parameters in parallel. The detection of incorporated ceramide in combination with other markers can thus not only be used for research, but potentially also for diagnostic purposes. Currently, it is, however, unclear what varying degrees of incorporation of clickable ceramides actually stand for, i.e., whether they reflect cellular ceramide content or whether the composition of lipid microdomains is the primary determinant of clickable ceramide incorporation.
Over the past decade, several reconstituted membrane models, e.g., planar membranes, as well as solid-supported membrane and liposomes, have been evaluated [11]. Liposomes are "soft" biocompatible, biodegradable, and non-immunogenic nanostructures [12]. These unique features and the possibility of adding targeting moieties to their surfaces make liposomes very attractive candidates as drug delivery vehicles [13]. In addition, due to their phospholipid bilayer, size, and morphology, liposomes are a suitable cell model system [14]. Liposomes have been used, e.g., for the investigation of permeation processes [15] and the estimation of the interaction of new drugs with cell membranes [16]. By integrating various proteins or CHOL into the lipid bilayer, a liposome-based membrane model can mimic different cells and allow, e.g., the measurement of the activity of transmembrane proteins or protein complexes [17]. Through variations of the lipid formulation, vesicle size, or surface charge, the effects of various biophysical parameters on the interaction between liposomal bilayers and low-molecular-weight substances can be investigated [18]. Furthermore, liposomes are suitable for the investigation of membrane fusion processes [19], as well as the transfer of membrane components [20] or membrane protein properties [21]. More specifically, the liposome-based model is suitable for the characterization of domain formation [22] and the investigation of the influence of ceramides on the structure of membrane microdomains [23].
Membranes and cellular content are transferred from one cell to another following the release of exosomes. The activity of the neutral sphingomyelinase 2 (Nsm2) is mandatory to enable cells to form and release exosomes [22]. With its localization in the inner leaflet of the cell membrane, ceramide generation by Nsm2 favors its outward protrusion and thus exosomal release. As a consequence, exosomes contain about three times more ceramide than cellular membranes [24]. For lymphocytes, it has been shown that activation greatly increases exosomal release, with T reg being about twice as efficient as conventional CD4 + T cells and B cells [25].
In this study, a reductionist liposomal membrane [24] model system was established to imitate the plasma membrane of mouse T reg and T conv with regard to their ceramide content. This model system was used to investigate whether the initial ceramide content in the membrane determines the capacity of the post-incorporation of ceramides into the lipid layer. In addition, we studied the impact of liposomes with different ceramide contents on the viability of primary mouse splenocytes in vitro.

Animals
C57BL/6 mice were bred under specific pathogen-free conditions and in accordance with German law in the animal facility of the Institute for Virology and Immunobiology of the University of Würzburg. In compliance with the 3R principles, particularly a reduction in animals used for scientific purposes, mice were used which were phenotypically wildtype, i.e., were Thy1.1-congenic or carried loxP sites without Cre recombinase expression or expressed a tamoxifen-inducible Cre recombinase without, however, receiving tamoxifen. The license to breed and house mice at the Institute for Virology and Immunobiology of the University of Würzburg for scientific purposes was obtained from the City of Würzburg (FB VVL 568/300-1870/13) on 29 June 2016.

General Liposome Preparation Procedure
For the synthesis of liposomes, several stock lipid solutions were prepared. Table 1 gives an overview of the lipids used for liposome preparation. The concentration of the stock solutions in CHCl 3 was 5 mM. In order to investigate the post-synthetic incorporation of the ceramide derivatives, additional stock solutions of the azide-modified ceramides in EtOH with a concentration of 25 mM were prepared. To dissolve ω-N 3 -C 16 -Cer, its ethanolic solution was additionally heated up to 50 • C. All stock solutions were stored at −25 • C. Ceramide-free and ceramide-containing liposomes were prepared using a lipid film hydration method following freeze-thawing. For the multi-component lipid formulation, the stock solutions in CHCl 3 were combined in a certain vol% ratio and evaporated using a rotary evaporator. The total volume before evaporation was always 1 mL. The resulting dry lipid film was additionally dried for 15 min under a pressure of 10-18 mbar and rehydrated with 1 mL of PBS. Ceramide-free and C 6 -ceramide-containing lipid formulations (C 6 -Cer, ω-N 3 -C 6 -Cer) were incubated for 1 h at 30 • C in a water bath. C 16 -ceramide-containing formulations (C 16 -Cer, ω-N 3 -C 16 -Cer) were incubated for 30 min at 50 • C in a water bath. The resulting mixture was homogenized for 30 s at 3000 rpm using a vortexer, freezethawed five times, and extruded once through a PC membrane with a pore size of 400 nm and 15 times with a pore size of 100 nm. Resulting liposomes were purified by a GPC. The chromatography column was equilibrated with 10 mL of PBS. The liposomal dispersion was diluted up to a total volume of 2.5 mL of PBS and applied to the column. The liposomes were eluted with 2 mL of PBS. For the transmission electron microscopy (TEM), liposomes were eluted with 2 mL of HEPES buffer.

Post-Insertion of Azide-Modified Ceramides and Click Reaction
Next, 25 mM stock solutions of azide-modified ceramides in EtOH (see Table 1) were diluted 1:1000 with PBS. The final concentration of all solutions was 25 µM. In each case, 500 µL of ceramide-modified liposome dispersion was incubated with 500 µL of diluted azide-modified ceramide solution for 12 h at 25 • C (ω-N 3 -C 6 -Cer) or 65 • C (ω-N 3 -C 16 -Cer) and 300 rpm in a thermoshaker. As a control, liposomes were incubated with 500 µL of EtOH:PBS 1:1000 mixture at 25 and 65 • C, respectively. After incubation, the dispersions were purified by GPC (see section before). Subsequently, 80 µL of the DIBO Alexa Fluor 488 (25 µM) was added. All dispersions were incubated for 12 h at RT and 300 rpm in a thermoshaker and purified by GPC again.
To quantify the amount of integrated azide-modified ceramides, 1 mL of each liposome dispersion was incubated with 80 µL of DIBO Alexa Fluor 488 for 12 h at RT. Liposomes were extruded through a 100 nm PC membrane and purified by GPC.
Primary mouse lymph node cells (1 × 10 7 ) from wildtype C57BL/6J mice (bred at the Institute for Virology and Immunobiology, see below) were incubated for 30 min at room temperature with N 3 -C 6 -Cer (25 µM) in HBSS medium containing Pluronic F-127 (20% in DMSO). Controls were incubated in HBSS medium only. After washing, Alexa Fluor 488 DIBO Alkyne (20 µM; Life Technologies) was added for 10 min to react with incorporated N 3 -C 6 -Cer in a click reaction. Anti-CD25 APC was added in parallel to Alexa Fluor 488 DIBO Alkyne. After three washings with HBSS, cells were transferred to 12-well slides for microscopy. For confocal microscopy, an LSM 780 confocal microscope from Zeiss (Jena, DE) equipped with a 40× Plan-Apochromat oil objective and 404 nm, 488 nm, and 633 nm lasers were used. Pictures were acquired and processed using ZEN2012 software (Zeiss). For FACS analysis, the cells were stained with anti-CD4 brilliant violet 421, anti-CD25 APC, and anti-CD44 PE. To discriminate between dead cells which very readily incorporate Alexa Fluor 488 DIBO Alkyne and live cells, the cells were incubated with Viability Dye eFluor ™ 780 from eBioscience (Carlsbad, CA).

Transmission Electron Microscopy (TEM)
TEM images of the liposomes were obtained on a JEM-1400Flash microscope (JEOL, Akishima, Japan) equipped with a Matataki Flash camera system. For negative contrast, carbon-coated 300-mesh copper grids were processed by a glow discharge. Subsequently, 20 µL of the liposome dispersion were dripped onto a grid. For the better contrast, 7 µL of 2% aq. uranyl acetate (UA) was added to the copper grid for 5 min. UA solution was removed with a filter paper and the grid was dried on the air for 10 min. TEM images were analyzed with the software Image J [26]. In total, 100 liposomes of each sample were analyzed ( Figure S7).

Dynamic Light Scattering (DLS)
For the detection of hydrodynamic diameter, the liposome dispersions were diluted 1:50 with PBS. Then, 1.5 mL of diluted solution was transferred to a polystyrene cuvette. DLS measurement was performed on a Zetasizer ZS (Malvern Instruments, Malvern, UK).

Nanoparticle Tracking Analysis (NTA)
The particle size, size distribution, and particle concentration of liposomes were determined by NTA on a NanoSight NS 300 device from NanoSight (Malvern Panalytical, Malvern, UK). For the NTA measurement, the liposome dispersions were diluted 1:200 with PBS. The examined volume was 1 mL. The analysis of measurement data was performed with the software NanoSight NTA 3.2.

Fluorescence Spectroscopy
The emission spectra were recorded with the Infinite M1000 Pro microplate reader (TECAN, Männedorf, Switzerland). Following this, 200 µL of the undiluted dispersions was transferred into a black 96-well microplate. Spectra were recorded in the range of 500 to 650 nm at the excitation wavelength of 488 nm. The measurement evaluation was carried out with the software TECAN I-Control Version 1.12.4.0.

Culture and Analysis of Mouse Splenocytes
In line with the 3R principle, C57BL/6J mice, i.e., phenotypically wildtype mice ('wrong genotype') from different strains of genetically modified mice, were used to obtain cells for our in vitro experiments. Mice were bred and maintained in the specific pathogenfree animal facility of the Institute of Virology and Immunobiology of the University of Würzburg in accordance with German law. Animals used for the experiments were aged between 23 and 31 weeks.
To obtain a single-cell suspension, the spleen was extracted and homogenized through 70 µm cell strainers in BSS/BSA. After centrifugation (1600 rpm, 5 min), erythrocyte lysis was performed to the cell suspension while vortexing using equal volumes of NaCl (1.8%) and ddH 2 O, followed by 10 min incubation on ice. The supernatant was washed in BSS/BSA and the live cells were counted using Trypan blue dead cell exclusion.
Next, 2 × 10 5 cells of splenocytes were incubated in a RPMI 1640 medium in the presence of Con A (5 µg/mL) with or without different liposomes with various C 16 -Cer (C 16 high, C 16 low, and C 16 free) and different liposome to cell ratio (10:1 to 10,000:1) in 96-well round-bottom plate for 48 h at 37 • C. As a negative control, cells were cultured alone without Con A or liposome.

T reg and Effector/Memory T conv of Mice Incorporated More Ceramide into Their Cell Membrane Than Naïve T conv
For mouse T reg , it has been shown that their membranes contain more ceramide than T conv [5,6]. In line with this, mouse T reg as well as effector/memory cells such as T conv incorporated more of the di-4-ANEPPDHQ (ANE) dye into the cell membrane, indicating a lower order of lipids to which a higher ceramide content contributes [6,28,29]. For human T reg and T conv , there was a positive correlation between the incorporation of the ANE dye and the cumulation of clickable ceramide in the cell membrane [7]. To test whether this also holds true for mouse T reg , we analyzed the incorporation of ω-N 3 -functionalized C 6ceramide into the cell membrane using confocal microscopy ( Figure 1A). Flow cytometric analysis of ceramide incorporation comparing naïve T conv , effector/memory T conv , and T reg cells further showed that both T reg and effector/memory T conv cells incorporated more ceramide than naïve T conv ( Figure 1B,C). Therefore, for both human and mouse CD4 + T cells, higher ANE and C 6 -ceramide incorporation were positively correlated.

Generation of Liposomes with High and Low Ceramide Content
To investigate whether the initial ceramide content in the membrane determines the capacity of post-incorporation of ceramides into the lipid layer, azide-modified ceramides (ω-N 3 -Cer) were synthesized and used for the preparation of liposomes. This enabled the covalent bonding of fluorescent dye molecules using a bio-orthogonal click reaction and subsequent fluorescence spectroscopic and microscopic analyses [30]. To generate liposomal membranes as reductionist model systems for T reg and T conv , liposomes with different ceramide contents and carbon chain lengths (C 6 -Cer and C 16 -Cer) were prepared using the film method followed by freeze-thaw processes and extrusion through a polycarbonate (PC) membrane [31]. Lipid vesicles with the size of 100 nm were made of 1-palmitoyl-2-oleoyl-snglycero-3-phosphocholine (POPC) and CHOL. The amount of CHOL was varied between 10 and 50 volume percent (vol%). The content of POPC was either equal or higher than the ceramide content but the amount of ceramide was, under all conditions, at least 20 vol%. As a reference, ceramide-free liposomes of the composition POPC:CHOL = 75:25 vol% were synthesized. Stable ceramide-modified liposomes (C 6 -Cer, C 16 -Cer) with a size of about 100 nm and narrow particle size distribution were obtained in the composition of Cer:POPC:CHOL = 30:45:25 vol%. Therefore, we used this composition as a membrane model for ceramide-rich mouse T reg . For relatively ceramide-poor T conv , a lipid composition of Cer:POPC:CHOL = 15:35:50 vol% was used [6]. The tested liposome compositions and determined particle sizes and concentrations are shown in Table 2. Figure 1. T reg and effector/memory T conv incorporate more ceramide than naive T conv . (A) Purified CD4 + lymph node cells of C57BL/6 mice were incubated with ω-azide-functionalized C 6 -ceramide and then reacted with DIBO 488 in a click reaction (green, top row). CD25 was used as a marker for T reg (red). The figure shows confocal microscopy images of DIBO 488-labeled cells (left column) and the overlay of DIBO 488, CD25, and the light microscopy images (right column). Cells with a brightly green cytoplasm are dead cells. Lower row: Negative controls without ω-azide-functionalized C 6 -ceramide and anti-CD25 mAb incubation. Original magnification: 40× plan apochromat objective lenses. The program Zen2012 for confocal microscopy was used to analyze and generate the images. (B) Gating strategy to detect T reg , naïve, and effector/memory T conv using the markers CD25 and CD44 (dot plots). The histogram overlays show representative stainings for DIBO 488 after incubation with (red line) or without ω-N 3 -C 6 -Cer incubation (green line). (C) Summary graph depicting C 6ceramide incorporation into T reg , CD44 -CD25naïve, and CD44 + CD25effector/memory T conv . Means + SD of n = 4 mice are shown. The dashed line indicates no specific binding of the ω-N 3 -C 6 -Cer. An unpaired two-sided t-test was used to assess statistical significance (* p < 0.05).
Regardless of the ceramide content, all liposome samples had a spherical unilamellar structure and showed no significant morphological differences (Figure 2).
High-performance liquid chromatography-tandem mass spectrometry (HPLC-MS/MS) was applied to quantify the exact amount of ceramide incorporated into the membrane of the liposomes. The amounts of ceramide measured per sample were normalized to the size of the particles and their concentration as determined by NTA (Table 3). Table 2. Lipid compositions, particle sizes, and concentrations of liposomes with different ceramide contents. The hydrodynamic diameter and the PDI represent the average of 12 measurements. * particle size determined by dynamic light scattering (DLS); ** polydispersity index; *** mean size and liposome concentration determined by nanoparticle tracing analysis (NTA).   The normalized ceramide contents showed that the quantity of ceramide incorporated into the liposomal membranes matched the vol% ratios used during liposome synthesis (Figure 3).

Lipid
To determine whether differences in the ceramide content can also be visualized using azide-modified ceramides, liposomes containing azide-modified ceramide were prepared. For the synthesis of these liposomes, the same ratios of azide-modified ceramides, POPC, and CHOL were used as for non-functionalized ceramides. Subsequently, the fluorescent dye DIBO Alexa Fluor 488 was covalently bound to the azido-ceramide-modified liposomes via a click reaction. Free dye was removed by gel permeation chromatography (GPC). The binding capacity was analyzed by fluorescence spectroscopy in comparison to pure dye (10 µM solution in PBS) and a liposome preparation without azide-modified ceramide. The fluorescence spectra of liposomes after the click reaction with the fluorescent dye are shown in Figure 4.  The pure dye solution had the strongest emission intensity, while the samples without azide-modified ceramide showed almost no signal. This confirms that there is no unspecific binding of DIBO Alexa Fluor 488 to the liposomal membrane. Furthermore, both ceramidecontaining liposome samples synthesized with a vol% ratio of 30:45:25 (ω-N 3 -C 6 /C 16 -Cer:POPC:CHOL) showed about twice the fluorescence intensity signal as the vesicles prepared with a vol% ratio of 15:35:50 (ω-N 3 -C 6 /C 16 -Cer:POPC:CHOL). At the same N 3 -Cer:POPC:CHOL ratio, there were no significant differences between the samples with different chain lengths of the fatty acid contained in the ceramide. This applied to vesicles for which 15 and 30 vol% of azide-modified ceramide had been used. However, it could be seen that the ω-N 3 -C 6 -Cer-containing samples had slightly higher intensities compared to the ω-N 3 -C 16 -Cer samples, which is in line with the quantifications obtained by HPLC-MS/MS (Figure 3).

Ceramide Content of Liposomes Is Positively Correlated with Incorporation of Clickable Ceramides
Different amounts of ceramide were used for synthesis, which led to the generation of liposomes with a two-fold difference in ceramide content as intended. We used these liposomes to investigate the influence of ceramide content on the incorporation of clickable ceramides. For this purpose, all liposome formulations (ceramide-free and ceramidecontaining) were incubated with azide-modified ceramide compounds (Table 2). Liposomes containing C 6 -Cer were incubated with ω-N 3 -C 6 -Cer and samples containing C 16 -Cer with ω-N 3 -C 16 -Cer, respectively. As a negative control, liposome samples were incubated with an EtOH:PBS mixture. Subsequently, all liposome samples were incubated with the fluorescent dye DIBO Alexa Fluor 488. The fluorescence spectra of liposomes after the click reaction with the fluorescent dye are shown in Figure 5. The emission intensities were normalized to the particle concentration determined by the NTA ( Table 2). The emission spectra showed that ceramide-containing liposomes incorporated more azido-modified ceramides than ceramide-free samples.
These results show that liposome samples with a 50 vol% CHOL have lower emission intensities than those with 25 vol%. However, the incorporation of ω-N 3 -C 16 -Cer into C 16 -Cer-containing liposomes led to more intense signals than the incorporation of ω-N 3 -C 6 -Cer into C 6 -Cer-containing vesicles (Figures 5 and 6). In addition, while the fluorescence signals stemming from clicked ω-N 3 -C 6 -Cer were proportional to the C 6 -ceramide content of the liposomes, this was less clear for ω-N 3 -C 16 -Cer-and C 16 -Cer-containing liposomes ( Figure 5). Therefore, the incorporation of ω-N 3 -C 6 -Cer allowed liposomal ceramide content to be quantified, while the incorporation of ω-N 3 -C 16 -Cer offered a higher degree of sensitivity at the cost of poor quantification.

Ceramide-Containing Liposomes Maintain Splenocyte Viability
Liposomes are widely used as nanocarriers for drug delivery including RNA-based vaccines such as BNT162b2 (Comirnaty ® ) [32]. To investigate the impact of liposomes without and with high versus low content of C 16 -Cer, we co-cultured primary mouse splenocytes in the presence of different amounts of liposomes for up to three days (Figure 7). While we did not observe activation or proliferation of B and T cells as read out by the markers CD69, CD25, and Ki-67 (unpublished data) in the presence of liposomes, we observed a clearly beneficial effect on the viability of B and T cells as well as monocytes (Figure 7). Comparing the different liposomal preparations, C 16 -Cer-containing liposomes were slightly more efficient than Cer-free liposomes in maintaining splenocyte viability.
The activation of B and T cells, particularly of T reg , greatly increases exosome release [25], which might provide a natural source of 'external' phospho-and sphingolipids similar to the liposomes (Figure 7). Therefore, we cultured mouse splenocytes in the presence of concanavalin A (Con A) to stimulate both B and T cells and followed the viability, activation, and proliferation of the cells for up to two days ( Figure 8). Compared to unstimulated splenocytes, the addition of liposomes had less of an effect on the viability of cells. Only the C 16 -high liposomes exerted a dose-dependent effect on viability, i.e., high concentrations reduced the viability of splenocytes, while the lowest ratio of liposomes to cells increased the viability of splenocytes (data not shown). Regarding lymphocyte activation, B cell as well as CD4 + and CD8 + T cell activation were reduced in the presence of liposomes with a low content of C 16 -Cer ( Figure 9A-C, left column). This was, however, only the case at a liposome to a cell ratio of 100:1. In line with reduced activation, the proliferation of CD4 + and CD8 + T cells was significantly reduced under these conditions and there was the same trend for B cells ( Figure 9A-C, right column).

Figure 7.
Mouse splenocytes were cultured for up to three days in the presence of different amounts of C 16 -Cer-free or C 16 -Cer-containing liposomes amounting to liposome: cell ratios of 10:1 to 10,000:1. (A) Viability and subpopulation composition were determined on a daily basis by flow cytometry using a viability dye (VD). (B) Means + SD of three independent experiments are shown. Two-way ANOVA followed by Tukey's test. * p < 0.5, ** p < 0.01, *** p < 0.001, **** p < 0.0001. p < 0.05 was considered statistically significant.  Values were normalized to stimulated cells in the absence of liposomes (=100%, black columns). Means + SD of n = 4 independent experiments. Two-way ANOVA followed by Tukey's test. * p < 0.5, ** p < 0.01, *** p < 0.001, **** p < 0.0001. p < 0.05 was considered statistically significant. Comparisons to unstimulated cells rendered significant results throughout (p < 0.05) and are not shown for clarity.
Ceramide-containing liposomes were thus superior to ceramide-free liposomes in maintaining the viability of splenocytes in vitro, especially at low liposome to cell ratios. Moreover, C 16 -low liposomes partially reduced lymphocyte activation and proliferation.

Discussion
Measuring the ceramide content of the cell membrane provides crucial insights into cellular activation and differentiation states. This is particularly true for lymphocytes which rely on sensing signals from the environment for mediating their function. Azide-functionalized ceramides offer the chance to assess ceramide content in the cell membrane of lymphocytes. However, so far, it has been unclear whether differences in the azidefunctionalized ceramide incorporation into the membranes of T reg and T conv actually reflect differences in the ceramide content in the cell membrane of T reg and T conv .
Liposomes constitute reductionist models of spherical lipid bilayers which we used to determine the contribution of the ceramide content of membranes to the incorporation of azide-functionalized ceramide. For this, we established a protocol to synthesize liposomes containing different amounts of ceramide. We could thus use liposomes with a high ceramide content (vol% ratio of 30:45:25 (C 6 / 16 -Cer:POPC:CHOL)) as membrane models of T reg , while liposomes with only half the ceramide content (vol% ratio of 15:35:50 (C 6 / 16 -Cer:POPC:CHOL)) served as a model for the plasma membrane of T conv [6].
In general, liposomes with a higher ceramide content (30:45:25 vol% (C 6 / 16 -Cer:POPC: CHOL)) incorporated more clickable azide-functionalized ceramides than those with a lower ceramide content (15:35:50 vol% (C 6 / 16 -Cer:POPC:CHOL)) (Figures 4 and 5). The differences between the degrees of post-incorporation into liposomes with high and low vol% ceramide were more pronounced for C 6 -than for C 16 -Cer, while the overall emission intensity was higher for C 16 -than for C 6 -Cer (Figures 4 and 5). This suggests that the post-incorporation of azido-modified C 16 -Cer goes into saturation at lower C 16 -Cer than C 6 -Cer concentrations in membranes. The negative controls confirmed that no unspecific binding of the fluorescent dye to the lipid layer had occurred.
The investigation of CHOL influence on the post-insertion of ceramides shows that liposome samples with 50 vol% CHOL have lower emission intensities than those with 25 vol% (Figure 6). This indicates that vesicles with a higher CHOL content incorporated less clickable ceramides via post-incorporation. The same tendency was observed for the ceramide-free, C 6 -Cer-and C 16 -Cer-containing samples (unpublished data). As CHOL increases the packing density of the phospholipids, a high CHOL content causes a higher density of the phospholipid double layer. It is also known that CHOL is displaced from domains rich in ceramide [33]. If CHOL increases the packing density of the phospholipids around the ceramide microdomains, this influences the lateral diffusion of the ceramide molecules and prevents the incorporation of azido-modified ceramides. Ceramides prefer to accumulate in ceramide-rich membrane regions [10]. This hypothesis was confirmed by lower ceramide post-incorporation into ceramide-free liposomes into which the ceramide derivatives are incorporated more slowly and in lower quantities ( Figure 5 and unpublished data). Samples with a ratio of 30:45:25 vol% (C 6 -Cer:POPC:CHOL) incorporated almost double the amount of azido-modified ceramides compared to vesicles with a ratio of 15:35:50 vol% (C 6 -Cer:POPC:CHOL) ( Figure 5A). C 16 -Cer-modified samples incorporated relatively more azido-functionalized ceramide than C 6 -Cer-containing liposomes with an overproportioned incorporation into liposomes with low C 16 -Cer content ( Figure 5B). A possible reason for this observation is that the liposomes were incubated with C 16 -Cer at 65 • C compared to 25 • C for C 6 -Cer to ensure post-incorporation of the C 16 -Cer. Generally, long-chain ceramides have a stronger influence on membrane fluidity than shortchain ceramides. The former can increase the membrane order and the phase transition temperature at which there is a shift from the gel to the liquid crystalline phase [34]. While both phases coexist in membranes, lipid monomers preferentially settle in the liquid crystalline phase due to its higher fluidity. Therefore, we incubated the liposomes with azido-modified C 16 -Cer at 65 • C instead of 25 • C. Additional experimentation is required to determine the conditions needed to achieve incorporation of C 16 -Cer into liposomes, and cells, in proportion to the ceramide content of the membrane. For the post-incorporation of ω-N 3 -C 6 -Cer, our data, however, clearly show that its degree reflected the Cer content of the liposomal membrane and thus might be suitable for quantification on a single-cell basis ( Figure 1) [3,7]. To determine the precise relationship of membrane ceramide content and ω-N 3 -C 6 -Cer post-incorporation, more variations regarding the Cer content of the liposomes studied by us would have been necessary. However, only the two ratios of C 6 -Cer/C 16 -Cer:POPC:CHOL reported here resulted in stable liposomes with a diameter of about 100 nm suitable for further investigations.
A further limit of our study is that we did not compare the effect of ceramide on ceramide post-incorporation to that of other lipids, apart from, of course, CHOL. In particular, it will thus be interesting in future experiments to study how, e.g., saturated lipids such as dipalmitoyl glycerol, known to affect the biophysical properties of liposomal membranes [35], might impact ceramide post-incorporation.
Cellular membranes contain acidic phospholipids including phosphatidylserine and phosphatidic acid which, in viable cells, are primarily localized in the inner leaflet of the plasma membrane [1]. Therefore, we assume that they are not a crucial determinant for the incorporation of extracellularly added ceramides into the outer leaflet of the plasma membrane of T cells (Figure 1). Still, also for these lipid species, their impact on the capacity of liposomes to incorporate azido-modified ceramides should be analyzed in future experiments.
As liposomes are widely used for drug formulations including mRNA-based vaccine preparations [32], we studied the impact of the different liposomal preparations on splenocyte survival and activation in vitro. Independent of their ceramide content, we observed that the addition of liposomes enhanced splenocyte survival. In contrast to its apoptosis-inducing activity in solution, liposomal ceramide even enhanced the beneficial effect of the liposomes on cell survival (Figure 7) [36]. Currently, it is unclear how liposomes maintain viability of splenocytes in vitro. One possibility is that cells constantly take up exosomes/extracellular vesicles in vivo to replenish their lipid content. Organ isolation and the preparation of single-cell suspensions of splenocytes might cause a lack of exosomes/extracellular vesicles in in vitro cultures. The majority of splenocytes might be unable to compensate for this lack by increasing de novo lipid synthesis. In line with this notion, liposomes had less of an effect on the viability of stimulated concanavalin A (Con A) than on unstimulated splenocytes ( Figure 9). The activation of B and T cells, particularly of T reg , greatly increases exosome release [25], which might compensate for the putative lack of exosomes/extracellular vesicles in the cultures of unstimulated splenocytes.
Ceramide-containing liposomes might thus be used in the future to enhance the yield and quality of difficult-to-culture cells, e.g., tumor-infiltrating lymphocytes used for cancer immunotherapy [37].
In contrast to the positive effect of liposomes, both with and without ceramide, on cell viability, B and T cell activation and proliferation were inhibited in the presence of liposomes with lower ceramide content and a liposome to cell ratio of 100:1. In addition, we noted that freshly prepared liposomes were strikingly more potent than liposomes stored at 4 • C for one week or more (unpublished data). Currently, it is unclear why we observed reduced T and B cell activation and proliferation only in the presence of a certain type of liposome and at a liposome to cell ratio of 100:1. One reason for the less sustained activation of B and T cells in the presence of ceramide-containing liposomes could be the fusion of liposomes and lymphocytes, leading to increases in ceramide content in the cell membrane of the T cells. For T cells, it has been shown that ceramide interferes with T cell receptor nanoclustering, which reduces the sensitivity of the T cells towards antigenic stimulation [35].

Conclusions
In summary, this study showed that ceramide-containing liposomes can be used as cell membrane models to test post-incorporation of ceramides into lipid bilayers. Regardless of the length of the ceramide chain, ceramide-containing vesicles incorporated more azide-functionalized ceramides than ceramide-free vesicles; in other words, a higher initial content of ceramides in the liposome membrane led to a higher post-incorporation capacity. This supports the assumption that the higher incorporation of clickable ceramide into T reg than into T conv actually reflected a higher ceramide content in the plasma membrane of T reg than of T conv , where, indeed, clickable ceramides can be found in ex vivo analyses ( Figure 1) [10,38]. For T cells whose biology and function heavily depend on the stimulation of cell surface receptors such as the T cell receptor and costimulatory molecules such as CD28, being able to quantitate the ceramide content of the cell membrane is crucial. This knowledge will help us obtain a better understanding of how the sphingolipid composition of the cell membrane influences T cell subsets and responses. Here, measuring the incorporation of clickable ceramides has clear advantages over the mass spectrometry of whole cell lysates, reflecting overall ceramide content.