Lateral Segregation of Palmitoyl Ceramide-1-Phosphate in Simple and Complex Bilayers

Abstract

Ceramide-1-phosphate is a minor sphingolipid with important functions in cell signaling. In this study, we examined the propensity of palmitoyl ceramide-1-phosphate (Cer-1P) to segregate laterally into ordered domains in different bilayer compositions at 23 and 37°C and compared this with segregation of palmitoyl ceramide (PCer) and palmitoyl sphingomyelin (PSM). The ordered-domain formation in the fluid phosphatidylcholine bilayers was determined using the emission lifetime changes of trans-parinaric acid and from differential scanning calorimetry thermograms. The lateral segregation of Cer-1P was examined when hydrated to bilayers in Tris buffer (50 mM Tris, 140 mM NaCl (pH 7.4)). At this pH, Cer-1P was negatively charged. The lateral segregation propensity of Cer-1P in 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) bilayers was intermediate between PCer and PSM. Based on differential scanning calorimetry analysis, we observed that the gel domains formed by Cer-1P in POPC bilayers (POPC:Cer-1P 70:30 by mol) were less stable (melting interval 16–37°C) than the corresponding POPC and PCer gel domains at equal composition (melting interval 20–55°C). The gel-phase melting enthalpy was also much lower in Cer-1P (1.5 kcal/mol) than in the PCer-containing POPC bilayers (9 kcal/mol). Cer-1P appeared to be at least partially miscible with PCer domains in POPC bilayers. Cer-1P domains were stabilized in the presence of PSM (POPC:PSM 85:15), similarly as seen with PCer-rich domains. In bilayers at 37°C, with an approximate outer-leaflet cell membrane composition (sphingomyelin and cholesterol enriched, aminophospholipid poor), Cer-1P segregation did not lead to the formation of ordered domains, at least when compared with PCer segregation. In bilayers with an approximate inner-leaflet composition (sphingomyelin poor, cholesterol and aminophospholipid enriched), Cer-1P also failed to form ordered domains. PCer segregated into ordered domains only after the PCer/cholesterol ratio exceeded an approximate equimolar ratio.

Significance

Ceramide 1-phosphate (Cer-1P) is a sphingolipid with known functions in cell signaling. Knowledge about its biophysical properties in bilayers is limited. In our work, we have examined the lateral distribution of Cer-1P in fluid phosphatidylcholine bilayers and compared its behavior with palmitoyl ceramide and palmitoyl sphingomyelin. Our findings confirm in part and contradict in part findings from previous studies on the subject. Our new information is helpful for better understanding of how physical properties relate to biological functions of Cer-1P.

Introduction

Sphingolipids constitute a large group of biologically active membrane lipids (1, 2), and their biological importance largely derives from their enormous possibilities for structural variation, including variations that produce a multitude of specific hydrogen-bonding functional groups (3, 4). As a result, many sphingolipids have been shown to be not only binding targets for various toxins but also interaction partners with membrane-bound receptors to initiate important signaling pathways in cells (5, 6, 7, 8, 9). The biosynthesis of sphingolipids is initiated with the coupling of L-serine with palmitoyl CoA, which yields 3-ketosphinganine. The rate-limiting enzyme is serine-palmitoyltransferase (10), and 3-ketosphinganine is subsequently reduced and acylated to yield dihydroceramide, which can be further converted to ceramide by dihydroceramide desaturase (11); however, not all dihydroceramide is desaturated. Ceramide is the basic intermediate in the biosynthesis of more complex sphingolipids, such as sphingomyelins, cerebrosides, and gangliosides. Ceramides can also be phosphorylated by a specific ceramide kinase (CerK) to yield ceramide-1-phosphate (Cer-1P) (12).

Cer-1P is a bioactive sphingolipid with known effects on the promotion of cell growth and survival and cell migration (8, 13, 14) and is also believed to have antiinflammatory properties (8, 15). Some of the biological effects of Cer-1P are likely to arise from its activation of cytosolic phospholipase A₂ (16), which may release arachidonic acid from the sn-2 position of membrane phospholipids for the subsequent production of eicosanoids (15). Cer-1P is believed to be dominantly produced by phosphorylation of ceramide by CerK, although other pathways may also produce Cer-1P because CerK^−/− knockout mice contain close to normal levels of Cer-1P (17). The most prevalent species of Cer-1P contains saturated acyl-chain residues in the N-linked position (18), which are common acyl chains of most sphingolipids (19). Cer-1P’s saturated nature indicates that Cer-1P may segregate laterally into domains that are more ordered than the surrounding fluid phospholipid bilayers. However, it has been suggested that palmitoyl Cer-1P does not segregate to ordered domains in fluid 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) bilayers (20, 21).

The biophysical properties of Cer-1P have been examined in only a few studies. Kooijman and co-workers described the properties of the monolayers of palmitoyl Cer-1P at the air-aqueous interface (22). In pure water, Cer-1P formed a monolayer that was more condensed than the monolayer formed on a buffer with a pH of 7.2 at otherwise equal conditions (22). Cer-1P monolayers are also known to be condensed considerably when the subphase contains Ca²⁺ (22, 23). These findings were interpreted to suggest that the repulsive headgroup charge in Cer-1P monolayers affected lateral packing. Interestingly, palmitoyl Cer-1P has been shown to form a bilayer despite its small headgroup. The bilayers are crystalline at low temperatures but transform to a gel phase at 45°C and convert to a liquid crystalline bilayer at 65°C (24). It also appears that the rather small headgroup of Cer-1P may impose negative curvature stress in the bilayers, as Cer-1P was shown to stabilize hexagonal phase transitions in 1,2-dielaidoyl-sn-glycero-3-phosphoethanolamine-mixed bilayers (24). The miscibility of palmitoyl Cer-1P in 1,2,dipalmitoyl-sn-glycero-3-phosphocholine bilayers was found to be significantly better than the corresponding miscibility of palmitoyl ceramide (PCer) at equal conditions (21), which illustrated the role of headgroup size and charge (phosphate versus hydroxyl) for colipid interactions involving PCer or Cer-1P. Based on the excimer/monomer (E/M) emission ratio reported by 1-palmitoyl-2-[(pyrene-1-yl)]decanoyl-sn-glycero-3-phosphocholine in bilayers containing POPC and increasing amounts of either Cer-1P or PCer, it was observed that the E/M ratio increased much more steeply in bilayers with an increasing concentration of PCer when compared to the E/M ratio from POPC bilayers containing Cer-1P instead (21). This was interpreted to show that although PCer segregated laterally and formed ordered domains, Cer-1P failed to do so. First-moment analysis of ²H NMR spectra obtained from perdeuterated POPC in the presence or absence of either PCer or Cer-1P supported this interpretation (21).

The aim of our study was to further examine the lateral segregation of palmitoyl Cer-1P (unless otherwise stated, Cer-1P hereafter always denotes the palmitoyl analog) and to better understand how segregation at 23°C is affected by colipids (POPC, DOPC, palmitoyl sphingomyelin (PSM), PCer, and cholesterol). We also examined the lateral segregation of both PCer and Cer-1P in complex bilayers (at 37°C) with lipid compositions resembling either the outer- or inner-leaflet lipid composition of biological membranes. We observed that the lateral segregation and subsequent ordered domain formation by Cer-1P in POPC bilayers (23°C) was intermediate compared to that seen with PCer or PSM. In cell-membrane-mimicking bilayers at 37°C, Cer-1P failed to form ordered domains in outer-leaflet-mimicking bilayers, whereas PCer segregated readily into ordered domains. In inner-leaflet-mimicking bilayers, Cer-1P did not form ordered domains, and the ordered-domain formation of PCer was hampered by the presence of cholesterol.

Materials and Methods

Materials

POPC, PCer, Cer-1P, egg sphingomyelin (ESM), egg phosphatidylcholine (EPC), 1,2-dioleoyl-sn-glycero-phosphocholine (DOPC), 1-palmitoyl-2-linoleoyl-sn-glycero-3-phosphoethanolamine (PLPE), and 1-palmitoyl-2-linoleyl-sn-glycero-3-phosphoserine were obtained from Avanti Polar Lipids (Alabaster, AL). Cholesterol and 1,6-diphenyl-1,3,5-hexatriene (DPH) were obtained from Merck (Darmstadt, Germany). Trans-parinaric acid (tPA) was prepared by chemical synthesis from the methyl ester of α-linolenic acid (Merck) according to published procedures (25), and the product was purified by crystallization from hexane at −80°C to 99% fatty acid purity. PSM was isolated and purified from ESM by preparative high-performance liquid chromatography (HPLC), as described previously (26).

Steady-state anisotropy measurements

Multilamellar vesicles (MLVs) of the desired composition were prepared by dispensing the appropriate lipids from the methanol stock solutions into disposable borosilicate glass tubes. After the solvent was evaporated from all samples, the dry lipid films were further dissolved in chloroform, dried, and exposed to a high vacuum for at least 1 h before hydration was performed. All samples were hydrated in a Tris-based buffer (50 mM Tris, 140 mM NaCl (pH 7.4); for pH 8.3, a similar Tris buffer was used; for pH 4, the buffer was prepared from 50 mM sodium acetate and HAc and also contained 140 mM NaCl) for 60 min at 70°C and then sonicated in a bath of the same temperature for 5 min. DPH was included at 1 mol%. The final total lipid concentration was 100 μM. The steady-state anisotropy of DPH in MLVs was measured with a QuantaMaster 1 instrument (Photon Technology International, Edison, NJ). The excitation polarizer was in the vertical position (V), and the emission polarizers were switched between the vertical (V) and horizontal (H) positions for each measurement point. The G-factor (the ratio of detection system sensitivities for vertically and horizontally polarized light) was determined using the excitation polarizer in the horizontal position (H). The anisotropy (r) was calculated according to (27)

$$\mathrm{r=}\left({\mathrm{I}}_{\mathrm{VV}}\mathrm{–G}{\mathrm{I}}_{\mathrm{VH}}\right)/\left({\mathrm{I}}_{\mathrm{VV}}\mathrm{+2G}{\mathrm{I}}_{\mathrm{HV}}\right)\text{,}$$ (1)

where I is the intensity measured with a vertical (V) or horizontal (H) polarizer plane (the first letter is for the excitation polarizer and the second for the emission polarizer). The anisotropy of samples containing the indicated lipids was recorded in the indicated temperature range using a temperature ramp of 5°C/min. The excitation and emission wavelengths were 360 and 430 nm, respectively.

Time-resolved fluorescence spectroscopy

The MLVs with indicated lipid compositions were prepared as described in the previous section. The MLVs were prepared in Tris buffer. The fluorescence decays of tPA emission were recorded at 23°C with a FluoTime 100 spectrofluorimeter with a TimeHarp 260 Pico, time-correlated, single photon-counting module (PicoQuant, Berlin, Germany). A 297-nm light-emitting diode laser source (PLS300; PicoQuant) was used to excite tPA, and the emission was collected through a long-pass filter with a 395 nm cutoff. The samples, containing 1 mol% tPA, were prepared and hydrated as described in the previous paragraph, thermostatted, and under constant stirring during the measurements. This amount of tPA is known not to affect colipid thermotropic properties in measurable ways (28). The data were analyzed using FluoFit Pro software obtained from PicoQuant. The decay was described by the sum of the exponentials, where α_i was the normalized pre-exponential and τ_i was the lifetime of the i^th decay component. The intensity-weighted average lifetime (27) is given by

$$\langle \tau \rangle ={\sum }_i{\alpha }_i{\tau }_i^2/{\sum }_i{\alpha }_i{\tau }_i.$$ (2)

Differential scanning calorimetry

MLVs containing POPC with either PCer, Cer-1P, or both were prepared by hydration of dried lipid films in glass tubes. The lipids were hydrated for 1 h at 70°C in Tris buffer before being loaded into the sample cell of a MicroCal VP-DSC instrument (MicroCal, Northampton, MA). The concentration of PCer or Cer-1P was 0.4 mM in all samples, and the colipids were adjusted to the indicated final proportions. The temperature ramp rate was 1°C/min. Five up- and downscans were obtained, and the fifth upscan was used. The data analysis was performed using Origin software (MicroCal).

Results

Segregation of PCer, Cer-1P, and PSM in POPC and DOPC bilayers

The lateral segregation into ordered domains of three saturated sphingolipids was compared in POPC or DOPC bilayers at 23°C. The segregation into ordered domains was detected from the increase in the average tPA emission lifetime because this probe is very sensitive to the order of its local environment (29, 30) and its partition coefficient is much higher for ordered or gel domains when compared to fluid bilayer domains (31). One should, however, keep in mind that the lateral segregation of lipids without the concomitant formation of ordered domains will not be sensed by tPA because emission lifetimes in those cases are not likely to change significantly. PCer is known to segregate into gel-like domains at fairly low bilayer concentrations (32, 33), and such segregation was indeed seen in our study (see Fig. 1). In the POPC bilayers, PCer above 5 mol% segregated into ordered domains, as shown by the markedly increased tPA emission lifetime when PCer content increased from 5 to 25 mol%. In the DOPC bilayers, the PCer concentration had to reach concentrations above 10 mol% before clear segregation into ordered domains was seen. This difference in the segregation susceptibility of PCer in the POPC and DOPC bilayer was previously reported by us (34). When Cer-1P segregation was examined, we noted that the nature of the host bilayer’s colipids (POPC or DOPC) did not affect the tPA emission lifetime and therefore did not affect the tendency of Cer-1P to segregate laterally to form ordered domains (Fig. 1). It also appears that the concentration dependence of Cer-1P ordered-domain formation changed at 20 mol% because the slope of the increase in PA lifetime became higher above 20 mol%. Comparing PCer and Cer-1P segregation, higher Cer-1P bilayer concentrations were needed to establish the formation of ordered domains with similar tPA emission average lifetimes as those seen in PCer-containing bilayers. This finding is in agreement with similar results presented earlier (21). The lateral segregation into ordered domains by PSM required the highest bilayer concentrations of the ordered lipid, and the average lifetime of tPA emission was slightly higher in POPC bilayers (at 45 mol% PSM) compared to DOPC bilayers (although only at the highest concentration used). The PSM segregation data have previously been published by us and were taken from (35). Based on the results shown in Fig. 1, it appears that headgroup size and maybe the charge (of Cer-1P) affected the segregation propensity of the three sphingolipids in the fluid phosphatidylcholine bilayers.

Figure 1.

Time-resolved tPA emission lifetime in POPC or DOPC bilayers containing increasing concentrations of saturated sphingolipids. The average tPA emission lifetime (ns) is plotted against the concentration of the saturated lipid. Key to symbols: circles = PCer; down triangles = Cer-1P, squares = PSM. Solid symbols are for POPC bilayers, and open symbols for DOPC bilayers. Each value is average ± SD from n = 3. To see this figure in color, go online.

Gel-phase stability of Cer-1P as function of pH

Pure Cer-1P bilayers prepared with buffer at pH 7.4 have been shown to undergo a gel-to-liquid crystalline phase transition at 65°C (24). Our steady-state DPH anisotropy measurement of Cer-1P bilayers at pH 7.4 confirms this phase transition temperature (Fig. 2). When Cer-1P in bilayers were increasingly protonated (pH 4.0), the corresponding phase transition temperature occurred at ∼73°C (Fig. 2). On the other hand, Cer-1P bilayers prepared in a buffer of pH 8.3, in which Cer-1P was more deprotonated, showed the gel-phase transition at ∼62°C (Fig. 2). These results suggest that changes in phosphate protonation affected lateral packing because of variation in charge-induced repulsion. Because all other experiments in this study had Cer-1P at pH 7.4, we can conclude that Cer-1P was partially deprotonated, and thus the phosphate headgroup was negatively charged, in good agreement with previous studies on the same subject (24).

Figure 2.

Steady-state anisotropy of DPH in pure Cer-1P bilayers prepared in buffers with varying pH. Each trace is representative for the indicated pH. To see this figure in color, go online.

Thermostability of PCer and Cer-1P-rich domains

Next, we measured the thermostability of the ordered domains formed by PCer and Cer-1P in POPC bilayers using differential scanning calorimetry (DSC). The molar ratio between POPC and the saturated sphingolipid was 7:3. As shown in Fig. 3 A, the T_m of the PCer endotherm was at a temperature of 45°C, and the gel-phase melting occurred over a rather broad temperature interval (20–55°C; enthalpy 9 kcal/mol). The corresponding Cer-1P gel-phase melting had a T_m of 27°C, and the melting covered a temperature interval of 16–37°C (enthalpy 1.5 kcal/mol). The POPC bilayers that contained both PCer and Cer-1P (at 1:1 molar ratio) showed a gel-phase melting with T_m of 39°C and a melting range of 26–48°C (enthalpy 4.1 kcal/mol). As a comparison, the gel-phase melting in POPC/PSM (60:30 by mol) had a T_m close to 20°C and a melting range between 10 and 30°C (36). At a POPC/PSM ratio of 70:30, the gel-phase melting would shift to slightly lower temperatures.

Figure 3.

Properties of segregated PCer and Cer-1P gel phases and miscibility of PCer and Cer-1P in ordered phases. DSC (A) was conducted on POPC bilayers containing either PCer, Cer-1P, or both. The lipid molar ratio was 70:30 POPC/PCer or Cer-1P or 70:15:15 POPC/PCer/Cer-1P. The saturated sphingolipid concentration was 0.4 mM, and POPC was included to give the indicated molar ratio. The fifth upscan is shown (it represents a stable and reproducible thermogram). In (B), the lateral segregation of binary PCer/Cer-1P (equimolar ratio) in the POPC bilayers was compared at 23°C with segregation of either PCer or Cer-1P in the POPC bilayers. In (C), the segregation of PCer in bilayers containing POPC and Cer-1P (80:20 by mol) is shown and compared with segregation of PCer and Cer-1P in POPC bilayers at 23°C. Each value is average ± SD for n = 3. To see this figure in color, go online.

Our Cer-1P results suggest that Cer-1P was miscible with the POPC and PCer gel phase and destabilized the PCer-rich phase (or, conversely, PCer stabilized the Cer-1P domain in the POPC bilayer). The DSC endotherms also show that the Cer-1P gel phase is ∼35–40% melted at an experimental temperature of 23°C. Because pure POPC does not display a noticeable endotherm above 0°C (Fig. 3 A; POPC gel-phase melting temperature is slightly below 0°C (37)) but we see the end part of an endotherm when Cer-1P is present (at 30 mol%), it suggests that Cer-1P is able to interact with POPC and stabilize its gel phase (i.e., no phase separation of POPC and Cer-1P gel phases).

Further support for the miscibility of Cer-1P with PCer in POPC is shown in Fig. 3 B, which shows the average tPA emission lifetime in POPC bilayers containing an equimolar ratio of PCer/Cer-1P, in comparison to PCer/POPC and Cer-1P/POPC binary bilayer systems. The segregation of PCer and Cer-1P occurs at a lower bilayer concentration as compared to Cer-1P only (Fig. 3 B). This observation that PCer can stabilize the Cer-1P-rich ordered domain or that Cer-1P can destabilize the PCer-rich domain in POPC bilayers (Fig. 3 B) is consistent with the DSC data shown in Fig. 3 A. Finally, when PCer segregation into ordered domains was measured in POPC bilayers also containing Cer-1P (POPC/Cer-1P 80:20 by mol), PCer segregation was markedly facilitated (Fig. 3 C). This shows that PCer and Cer-1P together formed the ordered domain, in which some POPC may also have been included.

Effect of PSM and cholesterol on ordered domains formed by Cer-1P and PCer

PCer is known to have rather good miscibility with PSM, and consequently, PCer has been shown to stabilize PSM-rich ordered domains (32). On the other hand, cholesterol can destabilize ordered domains formed by PCer in a concentration-dependent manner (38, 39). We now show (Fig. 4 A) that the inclusion of PSM in POPC bilayers (15:85 PSM/POPC) stabilized Cer-1P ordered domains because less Cer-1P was needed to form the ordered domains in the presence of PSM in the POPC bilayers. This finding is consistent with a previous study in which only a single composition of Cer-1P in POPC/PSM was examined (40). PCer segregation into ordered domains was also markedly enhanced in the presence of PSM in the POPC bilayer (Fig. 4 B), in agreement with previous observations (32, 40).

Figure 4.

Lateral segregation of Cer-1P and PCer in POPC bilayers and effects of PSM and cholesterol. When PSM was included, the POPC/PSM molar ratio was 85:15. With cholesterol, the ratio was POPC/Chol 4:1. (A) shows data for Cer-1P, and (B) shows data for PCer. This figure includes more data points for Cer-1P and allows us to better demonstrate the onset concentration of ordered-domain formation and the effects of cholesterol. Values are averages ± SD of n = 3. To see this figure in color, go online.

In Fig. 4 A, we have indicated the two different slopes for tPA emission increase as the Cer-1P concentration in the bilayer increased. The increase in tPA emission lifetime (and hence the degree of lateral order) appeared to change dramatically above 20 mol%. This suggest that the ordered domains started to form at and above 20 mol% Cer-1P. For PSM, the ordered-domain onset concentration is known to be 30 mol% in POPC bilayers at 23°C, whereas the onset concentration for liquid-ordered domain formation (in bilayers with 20 mol% cholesterol) is at 20 mol% PSM (41). The shift in onset concentration has been suggested to relate to the affinity cholesterol has for PSM (41). For Cer-1P, the presence of 20 mol% cholesterol did not appear to change the onset concentration of the ordered-domain formation (still seen at 20 mol% Cer-1P; Fig. 4 A), suggesting that cholesterol did not have a high affinity for Cer-1P. However, the average tPA emission lifetime in Cer-1P and POPC bilayers was affected by cholesterol, suggesting that lateral order in the Cer-1P-rich environment changed in the presence of cholesterol.

The lateral segregation of PCer in the POPC bilayers was affected by the presence of cholesterol, as higher concentrations of PCer were needed to achieve tPA emission lifetimes similar to those seen in the absence of cholesterol (Fig. 4 B). This finding supports previous observations that cholesterol may (partially) solubilize PCer-rich domains in fluid phospholipid bilayers (38).

Segregation of PCer or Cer-1P in complex bilayers that mimic the lipid compositions of either the outer or inner plasma membrane leaflet

Because both SM and cholesterol influence the lateral segregation of both Cer-1P and PCer, we wanted to measure their segregation at 37°C in complex bilayers with lipid compositions somewhat similar to those of the plasma membranes of eukaryotic cells, which are rich in both SM and cholesterol (42, 43). For the bilayers that mimicked the outer-leaflet compositions, we used the following lipid compositions: EPC, ESM, PLPE, and cholesterol in 40:20:5:33 molar proportions, respectively. For the bilayers that mimicked the inner-leaflet lipid composition, we used the following lipids: EPC, ESM, PLPE, 1-palmitoyl-2-linoleyl-sn-glycero-3-phosphoserine, and cholesterol in 18:10:26:11:29 molar proportions, respectively. Our selected compositions are similar but not completely identical to the outer- and inner-leaflet compositions reported in (44).

The segregation results at 37°C in the complex bilayers that mimicked the outer leaflet of plasma membranes are shown in Fig. 5 A. Based on tPA fluorescence lifetime analysis, we observed that the average lifetime increased curvilinearly with PCer concentration and reached 20 ns at 30 mol% PCer (Fig. 5 A). The corresponding increase in average tPA emission lifetime was much lower for the Cer-1P-containing membranes and only reached values of ∼11–12 ns at 30 mol% (Fig. 5 A).

Figure 5.

Lateral segregation and formation of ordered domains by PCer and Cer-1P in bilayers that mimicked either the outer (A) or the inner leaflet (B) lipid composition of eukaryotic cell membranes. The formation of ordered domains was examined at 37°C. Values are averages ± SD for n = 3. To see this figure in color, go online.

In the complex bilayer that mimicked the inner-leaflet lipid composition, the ability of PCer to form the gel phase was markedly attenuated compared to the outer-leaflet lipid composition system, in the concentration range 5–25 mol% (Fig. 5). Above this concentration range, PCer-rich domains were clearly and efficiently formed. At around 25 mol% PCer, the bilayer PCer/Chol molar ratio was equimolar, and PCer was in excess of cholesterol at PCer concentrations higher than this. Cer-1P did not affect average tPA emission lifetime markedly, even up to a concentration of 30 mol% (Fig. 5 B), which suggested that Cer-1P did not form gel-like ordered domains in this environment. However, we cannot exclude the possibility that the Cer-1P domains contained cholesterol and were more liquid-ordered in nature.

Discussion

Sphingolipids typically segregate laterally into domains because they often have saturated acyl chains (both the long-chain base and the N-linked acyl chain), which contrasts with the frequently mono- and poly-saturated nature of the acyl-chain composition of their glycerophospholipid colipids (45). However, segregation is also enhanced by the extensive hydrogen-bonding network that sphingolipids can form, which is different from the hydrogen-bonding properties that are typical of glycerophospholipids and cholesterol (4). The segregation of sphingolipids can further be enhanced if they lack a large polar headgroup, as is the case with ceramide (46, 47, 48). Because saturated sphingolipids increase acyl-chain order in the domains they form, among themselves and together with glycerophospholipids and cholesterol, methods that can detect changes in acyl-chain order are often used to detect the lateral segregation of ordered domains. For bilayer membranes, DSC (49), ²H NMR (50, 51), fluorescence spectroscopy (30, 31, 35, 52, 53), and microscopy (54, 55) have been successfully used to detect the lateral segregation of ordered domains.

The fluorescent probe tPa is excellent for detecting the lateral segregation of sphingolipid-rich domains, which have an increased acyl-chain order when compared to the surrounding fluid glycerophospholipid bilayer. tPA prefers to partition into ordered phases (K_p^so/ld values between 2 and 6.3 (47, 56, 57)), and its excited state is differently stabilized in ordered and disordered environments (58, 59). Therefore, time-resolved analysis of tPA emission lifetimes provide a valuable approach to detect the formation of ordered domains in a fluid bilayer. The information obtained from tPA emission lifetime studies agree well with applicable results from DSC or ²H NMR analysis (34, 53).

Segregation of Cer-1P in binary bilayers

When the lateral segregation of Cer-1P was compared with the segregation of PCer and PSM, based on the amount of sphingolipid required in POPC or DOPC bilayers to form an ordered phase with similar tPA emission lifetimes (Fig. 1), Cer-1P was intermediate between PCer and PSM in segregation propensity. PCer required the lowest bilayer concentration to form ordered domains, and PSM required the highest, whereas Cer-1P was in between (Fig. 1). Comparing the sphingolipids, there was a correlation between lateral segregation into ordered domains and headgroup size. Because PCer lacks a large headgroup, it can pack closer to adjacent lipids and thus more easily forms ordered domains (60), at least when compared to larger headgroup sphingolipids (Cer-1P and PSM). PSM has the largest headgroup and is zwitterionic in nature, whereas the headgroup of Cer-1P is significantly smaller than the phosphocholine moiety and has at least one negative net charge at pH 7.4 (Fig. 2; (24)). The segregation of neither Cer-1P nor PSM was markedly affected by the acyl-chain composition of the fluid phosphatidylcholine (POPC or DOPC). However, PCer showed different interaction with POPC and DOPC because the PCer concentration needed for gel ordered-domain onset was lower in the POPC than in the DOPC bilayers (Fig. 1; (34)). PCer needs to interact with a large headgroup colipid to avoid unfavorable exposure to interfacial bulk water. Apparently, the enhanced segregation of PCer in POPC compared to DOPC relates to the saturated palmitoyl residue on the sn-1 of POPC because interactions between saturated PCer and saturated acyl chains are favored over interactions with unsaturated oleoyl acyl chains at the sn-1 of DOPC. Because both Cer-1P and PSM have intermediate and large headgroups, respectively, their interaction with the glycerophospholipids is potentially influenced by steric hindrance in the headgroup region, and thus the acyl-chain nature of the colipid is less important than it is for PCer (35).

The ordered domains formed by Cer-1P in the POPC bilayers were less thermostable compared to the ordered domains formed by PCer in the POPC bilayers (Fig. 3 A). This can in part be understood to relate to more steric hindrance in Cer-1P and POPC interaction compared to interactions in the PCer and POPC system. However, because Cer-1P had a better monomer solubility in POPC compared to PCer, it could also order the POPC gel phase, as seen in the DSC thermogram (Fig. 3 A). Clearly, steric interactions in the headgroup region could not completely hinder such interactions. The enthalpy of gel-phase melting was much lower for Cer-1P than for PCer at the concentration examined, in good agreement with the lower thermostability of Cer-1P ordered domains. In the ternary bilayer system, where Cer-1P and PCer ordered domains were compared to pure Cer-1P or PCer domains at equal total sphingolipid concentration, it was apparent that Cer-1P was miscible with the PCer-rich domain because it destabilized the PCer-rich domains (lower T_m, lower enthalpy, Fig. 3 A; reduced tPA emission lifetime, Fig. 3 B). This interpretation is further supported by the facilitated segregation of PCer in POPC bilayers also containing Cer-1P (POPC/Cer-1P 80:20 by mol; Fig. 3 C).

Segregation of Cer-1P and PCer in more complex bilayers

Sphingomyelin and saturated phosphatidylcholines are known to stabilize PCer-rich ordered domains and to facilitate the lateral segregation of PCer in binary or ternary phospholipid bilayers (32, 33, 46). We observed that PSM also enhanced the lateral segregation of Cer-1P in the POPC bilayers (Fig. 4 A) in a similar fashion as seen for PCer segregation (Fig. 4 C). Such enhancement by PSM of the lateral segregation of Cer-1P is not surprising because the saturated Cer-1P is expected to favor interactions with the saturated PSM over the monounsaturated POPC. Hydrogen bonding between Cer-1P and PSM is also likely to stabilize their mutual interactions (4) and may involve the NH function of the sphingolipid as a hydrogen-bond donor (61, 62). Cholesterol, on the other hand, also prefers to interact with saturated (phospho)lipids over more disordered unsaturated (phospho)lipids (26). Cholesterol has been shown to destabilize PCer-rich ordered domains formed in the presence of sphingomyelin (38). Although cholesterol and phospholipid interactions are markedly affected by the size and nature of the colipid headgroup (63), cholesterol still appeared to interfere with Cer-1P or Cer-1P and POPC domain formation and with PCer and POPC interactions. This may be surprising because the ordered lipids lack the large headgroup cholesterol prefers (63); therefore, it can be concluded that the POPC headgroup was enough to allow some cholesterol interactions in the ternary complex. However, cholestatrienol is known to have a fairly low affinity for Cer-1P because it only weakly forms domains with Cer-1P, in which cholestatrienol is protected from quenching by a phase-selective quencher (40). Our observation that cholesterol failed to shift the concentration at which more ordered domains formed in Cer-1P bilayers (20 mol%; Fig. 4 A) also suggests that the affinity between cholesterol and Cer-1P is weak at best (41).

Finally, to establish whether lateral segregation into ordered domains was possible for Cer-1P and PCer in complex bilayers at 37°C, we measured ordered-domain formation in the bilayers that mimicked either the outer- or the inner-leaflet lipid composition of cell membranes. The outer leaflet of cell membranes is known to be enriched in saturated phospholipids (SM and other sphingolipids) and cholesterol and to be largely devoid of anionic phospholipids. The inner leaflet, on the other hand, is mostly devoid of sphingolipids, is enriched in phosphatidylethanolamine and phosphatidylserine, and is likely to contain substantial amounts of cholesterol (43, 44, 64, 65, 66). Compared to the efficient segregation of PCer into ordered domains in the outer-leaflet-mimicking bilayers, Cer-1P segregation into ordered domains was much more attenuated, as indicated by the differences in the tPA emission lifetimes in the two systems (Fig. 5 A). However, Cer-1P still increased the tPA emission lifetime (over the zero Cer-1P level), which suggests some ordering of the bilayer by Cer-1P. The finding that the segregation of PCer into ordered domains was fairly efficient in the outer-leaflet bilayer suggests that domain formation by PCer was more efficiently supported by the presence of ESM (33) than it was hindered by the presence of cholesterol (38). Clearly, ESM was not efficient enough to support the formation of Cer-1P-rich ordered domains in the presence of substantial amounts of cholesterol.

Interestingly, cholesterol apparently prevented PCer from forming ordered domains in the inner-leaflet-mimicking bilayers because these had a very limited amount of SM relative to cholesterol. However, when the bilayer contained PCer in excess of cholesterol (at 25 mol% and higher, Fig. 5 B)), PCer was able to efficiently segregate into ordered domains. Interestingly, Ali and co-workers (67) have shown that bilayer solubility of cholesterol is affected by ceramide, according to an equimolar stoichiometric relationship. Our finding on the segregation of PCer in the presence of cholesterol (Fig. 5 B) also suggests an equimolar stoichiometric relationship for PCer segregation in the presence of cholesterol. Cer-1P failed to form ordered domains in the inner-leaflet-mimicking bilayers, apparently because of the high cholesterol content and the highly unsaturated nature of the aminophospholipids.

Conclusions

Taken together, our results show that Cer-1P at 23°C can segregate laterally into domains that are more ordered than the surrounding Cer-1P-poor bilayer. This was true for both binary and ternary bilayers prepared from POPC or DOPC. In bilayers mimicking the outer or inner leaflet of cell membranes, Cer-1P failed to form ordered domains at 37°C, whereas PCer was able to form ordered domains in bilayers mimicking both leaflet compositions. Even though Cer-1P failed to form ordered domains under physiologically relevant conditions, it is likely that Cer-1P segregated into molecular clusters whose degree of acyl-chain order was higher than the surrounding lipid bilayer, especially in bilayers containing significant amounts of SM (Fig. 5 A). Cer-1P clusters may be important for efficient recognition by proteins, whose activity can be modulated by Cer-1P in cell-signaling pathways, although the physiologically relevant Cer-1P concentrations are much lower (∼0.1–45 pmol/mg cell protein (68, 69)) than those used in this study. Over the total cell phospholipid content, such amounts are low (1 mol% or less); however, locally in the membrane, the Cer-1P concentration can potentially reach significantly higher concentrations.

Author Contributions

All authors performed experiments and contributed to data analysis. J.P.S. wrote the manuscript with contributions from all authors. All authors have read and accepted the submission file.

Editor: Sarah Veatch.