Ablation of ceramide synthase 2 strongly affects biophysical properties of membranes.

Little is known about the effects of altering sphingolipid (SL) acyl chain structure and composition on the biophysical properties of biological membranes. We explored the biophysical consequences of depleting very long acyl chain (VLC) SLs in membranes prepared from lipid fractions isolated from a ceramide synthase 2 (CerS2)-null mouse, which is unable to synthesize C22-C24 ceramides. We demonstrate that ablation of CerS2 has different effects on liver and brain, causing a significant alteration in the fluidity of the membrane and affecting the type and/or extent of the phases present in the membrane. These changes are a consequence of the depletion of VLC and unsaturated SLs, which occurs to a different extent in liver and brain. In addition, ablation of CerS2 causes changes in intrinsic membrane curvature, leading to strong morphological alterations that promote vesicle adhesion, membrane fusion, and tubule formation. Together, these results show that depletion of VLC-SLs strongly affects membrane biophysical properties, which may compromise cellular processes that critically depend on membrane structure, such as trafficking and sorting.

series 2 spectrofl uorimeter using t-PnA and Rho (at fi nal probe concentrations of 0.7 µM and 0.5 µM, respectively). All measurements were performed at room temperature in 0.5 cm × 0.5 cm quartz cuvettes under magnetic stirring. The excitation ( ex )/ emission ( em ) wavelengths were 320/405 nm for t-PnA and 570/593 nm for Rho.

Sphingolipid composition of liver and brain microsomes
We previously demonstrated that CerS2 null mouse liver and brain exhibit a signifi cantly altered SL composition but no major alterations in phospholipid and cholesterol composition ( 6,8 ). We now characterize the SL profi le of microsomal lipids obtained from liver and brain of control and CerS2 null mice of different ages. Microsomes were used as a source of lipids because they are free from contaminating lipid particles (such as LDL) in the case of liver or myelin in brain.
Microsomal brain lipids are enriched in Cer (Supplementary Fig. S1A) compared with whole brain where SM and hexosylceramide (HexCer), which consists of both glucosylceramide (GlcCer) and galactosylceramide (GalCer), predominate ( 8 ). Because HexCer, particularly GalCer ( 8 ), is the major SL in myelin, the enrichment of Cer and the low levels of HexCer in microsomal brain lipids demonstrate that the extracts have little or no myelin lipid contamination. This was further confi rmed by Western blotting, in which myelin basic protein was barely detectable in the microsomal fractions, which were highly enriched in the ER marker, BiP (data not shown). Changes in the acyl chain composition of Cer, SM, and HexCer were similar in microsomes to whole brain (Supplementary Fig. S1A). Briefl y, C18-SLs were elevated, and VLC-SLs were depleted ( Fig. 1A and Supplementary Fig. S1). In addition, CerS2 null mice displayed a signifi cant reduction in unsaturated SLs and an increase in saturated SLs ( Fig. 1B ). The reduction in unsaturated SLs is a consequence of CerS2 ablation because the major contribution to the total unsaturated SLs in the WT mouse is due to C24:1 species ( ‫ف‬ 14% of the total SLs), with a smaller fraction ( ‫ف‬ 3%) contributed by the C18:1 species. Long chain bases (LCBs) (mainly sphinganine) were also elevated in brain microsomes, but the mol% of LCBs was only 1.0 ± 0.2 of the total sphingolipids in microsomes from WT mice and 3.0 ± 1.0 mol% in microsomes from CerS null mice. Such low amounts are unlikely to affect the biophysical properties of the vesicles prepared from microsomal lipids.
The SL composition of microsomal liver lipids (Supplementary Fig. S2A) was similar to whole liver ( 6 ) (i.e., SM comprised ‫ف‬ 50-60% of the total SLs), and HexCer was a minor component (<10%). A small increase in LCB levels was observed in CerS2 null liver microsomes (4.0 ± 1.3%) compared with 1.5 ± 0.5% in WT mice. As in whole liver, a signifi cant shift in the acyl chain composition from VLC-SLs to LC-SLs ( Fig. 1C ) was observed in the CerS2 null mouse, with C16-SLs signifi cantly elevated (Supplementary Figs. S2B, C). Similar to brain, a signifi cant shift in the ratio of unsaturated to saturated SLs was observed, with the latter the CerS2 null mouse ( 6 ), including increased fl uidity and strong morphological alterations.
We have performed a systematic analysis of the biophysical properties of membrane lipids isolated from the two main tissue types that show pathology in the CerS2 null mouse, namely brain and liver. Membrane biophysical properties were studied by fl uorescence spectroscopy and microscopy. Our results demonstrate that ablation of CerS2 has different effects on different tissues, causing signifi cant alterations in global membrane order through changes in the type and/or extent of the phases present in the membrane and drives changes in membrane morphology, promoting adhesion and fusion of neighboring vesicles. These changes are a consequence of the depletion of very long chain (VLC) and unsaturated SLs, which occurs to a different extent in liver and brain.

Preparation of lipid extracts
Tissue was homogenized in 20 mM HEPES (pH 7.4), 0.2 mM EGTA, 420 mM mannitol, and 150 mM sucrose. Homogenates were centrifuged at 400 g av for 10 min, and the supernatant was collected and centrifuged at 8,500 g av for 10 min. The subsequent supernatant was centrifuged at 100,000 g av for 1 h. The pellet was suspended in the same buffer, and lipids were extracted ( 14 ).

Lipid analysis
SLs were measured by ESI-MS/MS using a PE-Sciex API 3000 triple quadruple mass spectrometer and an ABI 4000 quadruplelinear ion trap mass spectrometer ( 4,6,15 ). SL levels were normalized to pmol/mg of dry weight, and data were plotted as a molar fraction of total SLs using X i = n i / ⌺ n i , where X i is the molar fraction of the lipid i , and n is the number of moles of the lipid i .

Confocal fl uorescence microscopy
Fluorescence microscopy experiments were performed at room temperature using a Leica TCS SP5 (Leica Microsystems CMS GmbH, Mannheim, Germany) inverted microscope (DMI6000) with a 63× water (1.2 numerical aperture) apochromatic objective. Giant unilamelar vesicles were prepared by electroformation ( 16 ) using microsomal lipid extracts from liver or brain. Rho and NBD were used to label the vesicles at a fi nal concentration of 1 M and 5 M, respectively.

Fluorescence measurements
Multilamellar lipid vesicles (MLVs) ( ‫ف‬ 0.3 mM phospholipid in phosphate-buffered saline) were prepared as previously described using microsomal lipid extracts from liver and brain tissue for at least three wild-type (WT) and three CerS2 null mice of different ages ( 6,17 ). Fluorescence anisotropy was measured (at least 10 measurements for each analysis) in a FluoroLog®-3 fl uorimeter (Horiba Jobin-Yvon, NJ) or an SLM Aminco 8100 and composition of CerS2 null mice compromise membrane lipid packing, promoting changes in membrane curvature and enhancing processes such as fusion and budding. Because there are no signifi cant variations in phospholipid and cholesterol content between CerS2 null and control mice ( 6,8 ), we conclude that the changes in biophysical properties are a direct consequence of changes in the SL composition.

Changes in membrane phases and phase separation in CerS2 null mouse brain
Alterations in the type of membrane phases and in the extent of phase separation were observed in lipids obtained from CerS2 null mouse brain. Vesicles were labeled with probes (i.e., NBD and Rho) that act as markers of liquid-ordered (lo) and liquid-disordered (ld) phases, respectively ( 17 ). Vesicles from WT mice (14 and 30 days old; Fig. 3A, B ) were homogenously labeled with both probes (except for small regions enriched in Rho), resulting in a yellow-colored overlay containing small red-colored areas enriched in Rho, which correspond to less-packed lipid domains. The area occupied by these domains changes with the age of the mice, resulting in alterations in global membrane order, as observed by the variation in t-PnA and Rho fl uorescence anisotropy ( Fig. 3D, E ). In contrast, vesicles from CerS2 null mice displayed lo-ld phase separation, resulting in an overlay image with large and distinct green (lo) and red (ld) areas. The extent of ld-lo phase separation also changes with mouse age. The increase in much more prominent ( Fig. 1D ), due to the higher contribution of C24:1-SLs ( ‫ف‬ 30% of total SLs). Thus, ablation of CerS2 causes signifi cant changes in the chain length of liver and brain microsomal SLs and also in their degree of saturation; the former should increase membrane fl uidity ( 18 ), whereas the latter should promote an increase in membrane order ( 19 ). The balance between these changes should dictate the overall biophysical properties of the membranes.

Altered membrane morphology in CerS2 mull mice
Giant unilamelar vesicles (GUVs) were prepared from liver and brain microsomal lipids and examined by microscopy. Control vesicles prepared from WT mice were present as stable isolated vesicles with a round shape typical of GUVs ( Fig. 2A, B ). In contrast, most vesicles prepared from CerS2 null mice showed morphological changes that included nonround-shaped vesicles containing areas of different curvature in the same vesicle and tubule-like structures emerging from a single vesicle ( Fig. 2A , lower  panel) or at contact points between two vesicles ( Fig. 2B ,  lower panel). These vesicles were also highly dynamic, with a strong tendency to adhere and fuse with neighboring vesicles ( Fig. 2A, B ). As a consequence, the vesicles appeared as aggregates or as small vesicles inside bigger vesicles. Similar alterations were observed in brain ( Fig. 2A ) and liver ( Fig. 2B ), although vesicles derived from brain were less affected (see also Figs. 3 and 4 ). These results demonstrate that alterations in the SL acyl chain structure  ( Fig. 3C ), the more homogeneous distribution of Rho results in a slight increase in Rho anisotropy ( Fig. 3F ), which is due to the increased area available for probe distribution. Simultaneously, a decrease in the lo phase results in a decrease in global membrane order. Thus, CerS2 ablation promotes alterations in the type and extent of phase separation of brain lipids, which leads to changes in the local and global order of the membrane.

Increased membrane fl uidity in CerS2 null mouse liver
Fluorescence microscopy of vesicles obtained from WT and CerS2 null liver membranes ( Fig. 4A, B ) demonstrated a relatively homogeneous distribution of NBD and Rho, with small NBD-enriched, Rho-free areas corresponding to more ordered domains. The lo domains were present in WT and CerS2 null membranes irrespective of mouse age. Noteworthy differences were observed in membrane order ( Fig. 4C, D ), with the lower t-PnA anisotropy in CerS2 null mouse consistent with increased membrane fl uidity, as previously demonstrated ( 6 ) and consistent with the higher anisotropy of Rho. These results suggest an overall decrease in membrane order in lipids from CerS2 null mice. The increased fl uidity might be associated with the morphological alterations, consistent with observations showing that the lower bending rigidity of the ld phase facilitates formation of highly curved structures ( 21 ) such as tubules.

DISCUSSION
In the present study we have shown that ablation of CerS2 leads to changes in SL levels that cause alterations in various membrane biophysical properties, such as fl uidity, type of phases, the extent of phase separation, and membrane curvature and morphology. The changes were tissue dependent, which is related to the different changes in SL acyl chain and composition between liver and brain.
In membranes derived from microsomal lipids from liver, fl uidity was increased in the Cer2S null mouse. This change in fl uidity is due to a combination of several factors, including the reduction in VLC-SL and the increase in C16-SL. LC-SLs have a lower melting temperature ( 18,22 ), which might contribute to the higher fl uidity. In contrast, lipids with a higher melting temperature have a strong tendency to segregate into gel domains, which would contribute to the increase in the global order of the membrane ( 23,24 ). In contrast, the increase in saturated SLs, which induce a considerable degree of order, may counteract the effect of decreasing VLC-SLs in such a way that the changes in membrane fl uidity are not as large as would be expected if only one of the parameters had changed. The increase in LCB levels could also cause an increase in membrane order and in stabilization of raft domains, but higher levels are required to cause such effects ( 25,26 ). Irrespective of the extent or the cause of the change in membrane fl uidity, changes in membrane morphology are induced. Low membrane bending rigidity, such as occurs in disordered phases ( 21 ), enhances formation of curved structures. Liver membranes from the lo phase, and therefore in the global order of the membrane, was further detected by the increase in fl uorescence anisotropy of t-PnA at 30 days of age ( Fig. 3D, E ). In older CerS2 null mice, the extent of ld-lo phase separation is lower ( Fig. 3C ), but t-PnA anisotropy shows an increase in membrane order ( Fig. 3F ) due to formation of small gel-like domains that appear as dark areas (areas which exclude NBD and Rho) ( Fig. 3C ). The partitioning of t-PnA into these gel-like domains ( 20 ) results in a signifi cant increase in anisotropy, which refl ects gel domain formation rather than a global increase in membrane order of a fl uid phase membrane. A similar increase in membrane order is observed in 120 day-old WT mice, which is also due to formation of small scattered gel-like domains. Despite the similar increase in membrane order in WT and older CerS2 null mice, fl uorescence microscopy images show distinctive morphological features.
To further monitor changes in membrane order, the fl uorescence anisotropy of Rho was measured. Rho anisotropy is affected by energy homotransfer, causing a decrease in anisotropy, with the process depending on the surface concentration of the probes. This is associated with an increase in membrane order [e.g., formation of lo or gel domains ( 17 )], leading to probe exclusion from the more ordered regions ( Fig. 3A, B ), thereby inducing closer localization in the more fl uid regions as occurs in vesicles from CerS2 null brain in younger mice ( Fig. 3D, E ). In older Fig. 2. Effect of CerS2 ablation on membrane morphology. Confocal fl uorescence microscopy of GUVs prepared from microsomal lipids from (A) brain or (B) liver of control (upper panels) and CerS2 null (lower panels) mice of different ages. GUVs were labeled with Rho. Scale bar, 5 µM.
in increased membrane order, which can be attributed to depletion of unsaturated SLs. The decrease in VLC-SLs in brain membranes was not as large as in liver because brain is enriched in LC-SLs ( 8 ), such that the length of the acyl chains contributes less to the increased membrane fl uidity. Despite the lower fl uidity of CerS2 null brain membranes, altered morphology was also observed. Changes in membrane morphology in liver and brain are probably related to similar changes that occur in each type of tissue, such as depletion of VLC-and unsaturated SLs.
CerS2 null mice had a hi gher tendency to adhere and fuse with neighboring vesicles and/or to bud from bigger vesicles, processes that require changes in membrane curvature, which could be achieved by lipid phase separation where sorting of lipids that promote the formation of curved structures takes place ( 27 ).
One of the most interesting observations in brain membranes was the complete segregation of ld and lo phases in vesicles from CerS2 null mice. Moreover, the extent of the lo phase was increased compared with WT mice, resulting closely dependent on membrane mechanics such as tension, shear stress, hydrostatic pressure, and compression ( 34 ). Cellular processes that depend on membrane mechanics and morphology, such as endocytosis-exocytosis ( 35,36 ), fusion, and budding ( 32,37 ), may be impaired in the CerS2 null mouse. Impairment of these processes has signifi cant implications on the sorting of lipids and proteins and in the development and maintenance of organelles such as the ER and Golgi (38)(39)(40). Thus, alterations in the curvature and fl uidity of the membrane might have important ramifi cations on cell physiology ( 13,41 ). For instance, deficient traffi cking might drive accumulation of lipids in the lysosomes, ultimately leading to lysosomal storage diseases ( 10,42 ). Banana-shaped storage material, similarly to that observed in neurons of GM1 gangliosidosis mouse ( 43 ), was detected in lysosomes of astrocytes from the brain of CerS2 null mouse ( 8 ).
In summary, ablation of CerS2 affects SL metabolism to different extents in liver and brain, causing signifi cant This is one of the fi rst studies to examine differences in membrane biophysical properties in natural membranes enriched in SLs with specifi c acyl chain structures. Such changes as reported herein may infl uence a variety of cellular functions, including activation/inhibition of specifi c signaling proteins that undergo conformational changes upon membrane physical alterations ( 28,29 ). Moreover, the existence of membrane domains with phase properties distinct from the bulk membrane is often associated with entrapment therein of ligands and receptors in order to amplify signal transduction ( 30 ) or to the activation of certain enzymes by facilitating their adsorption on the membrane at the interfacial defects between the two phases ( 31 ). In cell membranes, formation of such domains may occur transiently due to the constant lipid and protein reshuffl ing by cellular processes such as synthesis/degradation, fl ip-fl op, and traffi cking ( 32,33 ).
Of no less importance are the changes that occur in membrane morphology. Changes in membrane morphology are Values are means ± SD (n у 3). Fluorescence anisotropy was measured at least 10 times for each sample. ** P < 0.01. alterations in the global order of the membrane, in the extent of phase separation, and in the number and type of lipid domains formed and causes major changes in membrane morphology. The study of membrane biophysical properties paves the way for understanding the impact of SL acyl chain structure on various membrane properties in general and more specifi cally for defi ning the cellular mechanisms of pathology in CerS2 null mice.