CHO/LY-B Cell Growth under Limiting Sphingolipid Supply: Correlation between Lipid Composition and Biophysical Properties of Sphingolipid-Restricted Cell Membranes

The FASEB Journal. 2021;35:e21657. | 1 of 23 https://doi.org/10.1096/fj.202001879RR wileyonlinelibrary.com/journal/fsb2 Received: 7 August 2020 | Revised: 9 April 2021 | Accepted: 26 April 2021 DOI: 10.1096/fj.202001879RR RESEARCH ARTICLE CHO/LYB cell growth under limiting sphingolipid supply: Correlation between lipid composition and biophysical properties of sphingolipidrestricted cell membranes Bingen G.Monasterio1,2 | NoemiJiménezRojo3 | Aritz B.GarcíaArribas1,2 | HowardRiezman3 | Félix M.Goñi1,2 | AliciaAlonso1,2 This is an open access article under the terms of the Creative Commons AttributionNonCommercialNoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is noncommercial and no modifications or adaptations are made. © 2021 The Authors. The FASEB Journal published by Wiley Periodicals LLC on behalf of Federation of American Societies for Experimental Biology. In memoriam MJO Wakelam (19552020), colleague and friend. Abbreviations: AFM, atomic force microscopy; Cer, ceramide; Chol, cholesterol; FBS, fetal bovine serum; GP, generalized polarization; GPL, glycerophospholipids; GPMV, giant plasma membrane vesicles; GUV, giant unilamellar vesicles; HexCer, hexosyl ceramide; PM, plasma membrane; SL, sphingolipids; SM, sphingomyelin; SPT, serine palmitoyl transferase; SUV, small unilamellar vesicles. 1Instituto Biofisika (CSIC, UPV/EHU), Universidad del País Vasco, Leioa, Spain 2Departamento de Bioquímica, Universidad del País Vasco, Leioa, Spain 3NCCR Chemical Biology, Department of Biochemistry, University of Geneva, Geneva, Switzerland Correspondence Félix M. Goñi, Instituto Biofisika (CSIC, UPV/EHU), Universidad del País Vasco, Leioa 48940, Spain. Email: f[email protected] Funding information Spanish Ministry of Economy, Grant/ Award Number: PGC2018099857BI00; Basque Government, Grant/Award Number: IT126419 and IT127019; Swiss National Science Foundation, Grant/Award Number: 310030184949 Abstract Sphingolipids (SL) are ubiquitous in mammalian cell membranes, yet there is little data on the behavior of cells under SLrestriction conditions. LYB cells derive from a CHO linein whichserine palmitoyl transferase (SPT), thus de novo SL synthesis, is suppressed, while maintaining the capacity of taking up and metabolizing exogenous sphingoid bases from the culture medium. In this study, LYB cells were adapted to grow in a fetal bovine serum (FBS)- deficient medium to avoid external uptake of lipids. The lowest FBS concentration that allowed LYB cell growth, though at a slow rate, under our conditions was 0.04%, that is, 250fold less than the standard (10%) concentration. Cells grown under limiting SL concentrations remained viable for at least 72hours. Enriching with sphingomyelin the SLdeficient medium allowed the recovery of growth rates analogous to those of control LYB cells. Studies including whole cells, plasma membrane preparations, and derived lipid vesicles were carried out. Laurdan fluorescence was recorded to measure membrane molecular order, showing a significant decrease in the rigidity of LYB cells, not only in plasma membrane but also in whole cell lipid extract, as a result of SL limitation in the growth medium. Plasma membrane preparations and whole cell lipid extracts were also studied using atomic force microscopy in the force spectroscopy mode. Force measurements demonstrated that lower breakthrough forces were required to penetrate samples obtained from SLpoor LYB cells than those obtained from control cells. Massspectroscopic analysis was also a helpful tool to understand the rearrangement undergone by the LYB cell lipid metabolism. The most abundant SL in LYB cells, sphingomyelin, decreased by about 85% as a result of SL limitation in the medium, the bioactive lipid ceramide and the ganglioside precursor hexosylceramide decreased similarly, together with cholesterol. Quantitative SL analysis showed 2 of 23 | MONASTERIO ET Al. 1 | INTRODUCTION Sphingolipids (SL) are characterized by a sphingoid structural backbone, sphingosine being the most abundant base in mammals.1 In vivo studies of the SL roles are hampered, among other reasons, by the fact that they can be either synthesized de novo or taken up from the diet. A plausible approach would be to investigate mutant cells containing the smallest possible amounts of SL, or even none at all, but this is not a straightforward procedure because SL appear to be essential for cell growth and survival.24 In particular, SL are considered as instrumental in the architecture of eukaryotic cell membranes. Aside from stabilizing the lamellar structure and helping to maintain its asymmetry, the tendency of certain SL to undergo lateral phase separation to form micro or nanodomains has been characterized.58 Lipids and proteins colocalize with these domains, in which sphingomyelin (SM) is the most abundant SL,9 and cholesterol (Chol) is often present.7,8 The sphingoid base provides SL with hydrogenbonding acceptors and donors (amide and free hydroxyl groups, respectively) that are rare in glycerophospholipids, leading to a dense intermolecular hydrogenbonding network.10,11 Hydrogen bonding allows SM to interact preferentially with Chol12,13 and with ceramide (Cer),6 and recent studies have found that sphingomyelin (SM), Cer and Chol are able to coexist in a single ternary gel phase (at a 54:23:23mol ratio) with intermediate properties between SMCer enriched gel domains and Choldriven liquidordered phases.14 The above data have been obtained mostly from model membrane studies. Investigations at the cellular level have allowed to assign a wide variety of functions to SL, including apoptosis, cell growth, cell membrane function, tumor formation, drug resistance, degranulation, and phagocytosis, among others.1518 Alterations in the normal activities of SMcycle enzymes have been linked to many central nervous systemrelated pathologies such as Alzheimer's, Parkinson's, ischemia/hypoxia, depression, schizophrenia, or NiemannPick diseases.19 Specific SL functions have often been characterized in cells with decreased amounts of SL, using either SLdegrading enzymes (eg, sphingomyelinases or ceramidases),20 or specific enzyme inhibitors,15,2123 of which the SPT inhibitor myriocin is a good example.2427 Procedures to decrease the SL contents of cells constitute good tools to determine their potential functions in vivo. De novo SL biosynthesis is initiated by the condensation of Lserine with palmitoyl CoA. This reaction is catalyzed by SPT to generate 3ketodihydrosphingosine. 3Ketodihydrosphingosine is then converted to dihydrosphingosine, which is Nacylated and (most of it) dehydrogenated at the endoplasmic reticulum to form Cer. After moving to the Golgi apparatus, Cer is converted to SM or glycosphingolipids.6,28 Finally, these complex SL are translocated to the PM.29,30 Thus, SPT is a key enzyme for the regulation of cellular SL content.9,29 Using a genetic selection method in CHO cells,31 the Hanada lab isolated the defective LYB cell line, which had a loss of function ofserine palmitoyltransferase (SPT) enzyme activity through a defective SPTLC1 subunit. The mutant cells maintained the ability to take up and metabolize exogenous sphingoid bases from the culture medium.31 Mutant LYB and wildtype CHO cells could be comparatively studied to determine the effect of SL depletion on the biophysical properties of cell membranes. LYB cells have been used in multiple studies exploring SL effects, and their interaction with glycerophospholipid metabolism.3235 SM synthases, which use Cer and phosphatidylcholine as substrates to produce SM and diacylglyceride, are used in the de novo synthesis pathway, sometimes also involved in reutilization of ceramide30,36 from exogenous or endogenous sources. In a previous work,37 plasma membrane (PM) preparations from CHO cells and model membranes prepared from their lipid extracts had been characterized. In the present study, we have applied those methods to the study of SLsynthesisdeficient LYB cells to determine the role of SL on cell growth, membrane physical properties, and composition. that a 250fold reduction in sphingolipid supply to LYB cells led only to a sixfold decrease in membrane sphingolipids, underlining the resistance to changes in composition of these cells. Plasma membrane compositions exhibited similar changes, at least qualitatively, as the whole cells with SL restriction. A linear correlation was observed between the sphingomyelin concentration in the membranes, the degree of lipid order as measured by laurdan fluorescence, and membrane breakthrough forces assessed by atomic force microscopy. Smaller, though significant, changes were also detected in glycerophospholipids under SLrestriction conditions. KEYWORDS AFM, CHO, Laurdan, lipidomics, LYB, massspectroscopy, membrane fluidity, plasma membrane, sphingolipids, sphingomyelin | 3 of 23 MONASTERIO ET Al. 2 | MATERIALS AND METHODS 2.1 | Cell growth Wildtype CHO (ATCC, Manassas, Virginia, US) and a serineSPTdeficient CHO cell line, known as LYB31 (RIKEN BioResource Research Center, Koyadai, Japan), were used in this study. Unless otherwise mentioned, cells were grown on DMEM:F12 (Dulbecco's Modified Eagle Medium: Nutrient Mixture F12) medium containing 10% FBS (Fetal Bovine Serum), 100 U/mL penicillin, 100 U/ mL streptomycin, and 6mM glutamine (GlutaMax supplemented) at 37°C and 5%CO2 humidified atmosphere. All cell culture products were purchased from Thermofisher (Waltham, MA). 2.1.1 | Cell adaptation: Standard vs. deficient (lowFBS) medium CHO and LYB cells were adapted to growth in deficient (lowFBS) medium. For this purpose, cells were first seeded in DMEM:F12 medium containing 10% FBS, 100 U/ml penicillin and 100 U/mL streptomycin, and 6mM glutamine (this medium will be referred to as ”standard medium”). After 24hours cell growth, when a 15%- 25% confluence was reached, the standard medium was discarded, cells were washed with PBS buffer (137mM NaCl, 3mM KCl, 80mM Na2HPO4, 7mM KH2PO4), and DMEM:F12 medium containing 0.04% FBS, 100 U/mL penicillin, 100 U/mL streptomycin and 6mM glutamine was added (this medium will be named “FBSdeficient” or ”SLdeficient medium”). Cells were grown in the appropriate medium for 24, 48, 72, or 96hours. Other lowFBS media (5%, 2.5%, or 1.25%) were also used in some specific cases. 2.2 | Growth rate and viability tests 2.2.1 | Cell growth 2.65×105 cells were seeded in 25cm2 flasks in standard medium and grown for 24hours until 1525% confluence. Then, the standard medium was discarded, cells were washed twice with PBS, and the appropriate medium (standard or deficient) was added. Cells were grown for 24, 48, 72, or 96hours. Cell growth quantification was performed by cell counting with a hemocytometer (BioRad TC20 Automated Cell Counter, Hercules, CA). Protein was assayed with the colorimetric Pierce BCA Protein Assay Kit (Thermofisher, Waltham, MA). It was tested whether LYB cells were able to reach the full growth rates when the SLdeficient medium was supplemented with SL. SM (0.2mg/5mL; ≈80µM SM), sphinganine (0.0125mg/5mL ≈8µM), cerebroside (at 0.2mg/5mL, ≈55 µM ), or sphingosine (0.02 mg/5 mL, ≈13 µM) were used for this purpose. 2.2.2 | Viability test Flow cytometry was performed to evaluate how the decreased FBS concentration in the medium affected cell viability.38 Cells were stained with AnnexinVFITC and propidium iodide as indicated in the manual of the annexin VFITC detection kit (CalbioChem, Darmstadt, Germany), and fluorescence was measured using a FACS Caliburflow cytometer (BectonDickinson, Franklin Lakes, NJ) as in Ahyayauch et al.39 Annexin VFITC fluorescence intensity was measured in fluorescence channel FL1 with λex = 488nm and λem = 530nm, while FL3 was used for propidium iodide detection, with λex = 532nm and λem = 561nm. All measurements were performed in triplicate. Data analysis was performed using Flowing Software 2. 2.3 | Sample preparation Intact cells (whole cells), two different PM preparations (giant plasma membrane vesicles, known as GPMV or blebs, and PM patches) and SUV or GUV formed with wholecell or PM lipid extracts were used. 2.3.1 | PM preparations PM preparations were obtained as described in Monasterio et al.37 Briefly, GPMV formation was induced adding the GPMV formation reagent [freshly prepared 2 mM dithiothreitol, 25mM paraformaldehyde in GPMV buffer (2mM CaCl2, 10mM HEPES, 150mM NaCl, pH 7.4)] to T25 flasks with cells at confluence. Cells were incubated for 1hour at 37°C. After incubation, the GPMVcontaining GPMV reagent was collected from the flasks and centrifuged at 14000g for 20minutes. The supernatant was discarded, the pellet was resuspended in GPMV buffer, and the sample was centrifuged at 14000g for 20minutes. The procedure was repeated twice to remove traces of dithiothreitol and paraformaldehyde. Finally, the GPMV were resuspended in 500 µL GMPV buffer.40 PM patches were isolated by a modification37 of the protocol described by Bezrukov et al.41 In summary, cells were seeded at approximately 50% confluence and incubated for 2 hso that they adhered to the support. After incubation, 2 washing steps were performed using cold TBS (Tris Buffer Saline: 150 mM NaCl, 25 mM TrisHCl, 2 mM KCl) to discard nonattached cells. Then, cold distilled water was added for 2 minutes to induce cell swelling. Mechanical 4 of 23 | MONASTERIO ET Al. cell disruption was achieved using a pressure stream from a 20mL syringe coupled to a 19X11/2(TW)A needle. In the process, intracellular contents were released, while PM stayed attached to the support. Several washing steps were performed to discard the released intracellular contents. Purification quality was checked using Di4ANEPPDHQ (λex = 465 nm, λem = 635 nm) as a general fluorescent staining, together with organellespecific fluorophores as described in Monasterio et al.37 Images were taken in a Leica TCS SP5 II microscope (Leica Microsystems GmbH, Wetzlar, Germany) at room temperature with ImageJ software. The fluorescence intensities of the various markers were comparatively measured in PM patches and intact cells, so that specific organelle contamination could be estimated. 2.3.2 | Whole cell lipid extract Lipid extraction was performed following the method used in Ahyayauch et al.39 Briefly, cell pellets were first dispersed in aqueous perchloric acid (60% v/v), then centrifuged at 14000g for 15minutes, and the supernatant was discarded. Pellets were resuspended in 2.5 mL chloroform:methanol (2:1, v/v) and samples were mixed for 15 minutes. Then, 5mL cold 0.1mM HCl was added to the mixture. After homogenizing, samples were centrifuged at 1,700g for 20minutes. Supernatants were discarded while the lipidcontaining organic phase remained in the bottom layer. Phospholipid concentration was assayed as inorganic phosphorus after acid digestion. 2.3.3 | PM patch lipid extraction PM patches were formed following the above mentioned protocol.37,41 Chloroform:methanol (2:1) organic solvent was used to recover the lipid fraction of the attached PM patches. 2.3.4 | GUV formation GUV were formed in a PRETGUV 4 chamber supplied by Industrias Técnicas ITC (Bilbao, Spain) using the modified electroformation method42 first developed by Angelova and Dimitrov.43 2.3.5 | SUV formation The sample was kept under vacuum for 2 hours to remove solvent traces and the lipids were swollen in PBS buffer. SUV were obtained by sonication of the swollen lipid suspensions with a probetype Soniprep 150 sonicator (MSK, London, UK) for 10minutes, in 10s on, 10s off intervals. 2.4 | SM quantification with lysenin 2.4.1 | LyseninmCherry expression and purification The nontoxic monomeric Cterminal domain of the SMspecific toxin, NTlysenin, was expressed and purified as described by Carquin et al.44 Briefly, the expression plasmid pET28/lysenin encoded NTlysenin as a fusion protein with an Nterminal 6xHistag followed by the monomeric red fluorescent protein mCherry. The plasmid was expanded in Escherichia coli BL21 (DE3) and the recombinant protein was expressed in lysogeny broth (LB) medium at 16°C for 72hours in the presence of 0.4mM isopropyl βDthiogalactoside. Bacterial extracts were prepared as described45 and the recombinant protein was purified using an NiNTA Superflow cartridge (Qiagen, Hilden, Germany) and eluted with imidazol.46 Fraction analysis by SDSPAGE revealed recombinant NTlysenin with the expected size (45kDa). The most enriched fractions were pooled, concentrated, and desalted. The aliquots were stored in 20mM NaCl and 25mM Hepes pH 7.2 and 5% glycerol at −80°C. Protein concentration was calculated by measuring absorbance at 280nm. 2.4.2 | SM staining and quantification with lyseninmCherry Whole cells and PM patches were stained with lyseninmCherry(for SM) and NBDPE [(N- (7nitrobenz2oxa1,3diazol4yl)- 1,2dihexadecanoylsnglycero3phospho ethanolamine,triethylammonium salt] fluorophore (the latter a general membrane stain) and samples were visualized using a confocal microscopy Nikon DECLIPSE C1 (Nikon, Melville, NY). In whole cells, the mCherry signal was also quantified using a FL3 FACS Calibur flow cytometer (BectonDickinson, Franklin Lakes, NJ) with λex = 532nm and λem = 561nm. For sample visualization, cells were seeded in glassbottom dishes and grown as above. Cells were first stained with 100 µM NBDPE as a control for general membrane staining. A washing step was performed with PBS, and lyseninmCherry was added at 100µM. Forflow cytometry analysis, cells were stained in suspension at a final concentration of 100µM lyseninmCherry. | 5 of 23 MONASTERIO ET Al. 2.5 | Laurdan general polarization Laurdan is a fluorescence polarity probe whose emission undergoes a spectral shift due to the reorientation of water molecules in the glycerol backbone region of the membrane, and this shift can be correlated to the lipid phase.47 In the gel phase, when little water is present, laurdan maximum emission is around 440nm, whereas in the liquid crystalline phase the spectrum is redshifted to around 490 nm. Intact cells, PM preparations and model membranes formed with lipid extracts have been used to compare the laurdan fluorescence of CHO and LYB cells grown in standard and deficient media. Both fluorescence microscopy imaging and spectrofluorometric analysis have been performed for laurdan fluorescence characterization. 2.5.1 | Confocal microscopy Intact cells grown in glassbottom dishes were stained with laurdan (Molecular Probes, Eugene, OR) as follows: First, cell culture medium was discarded and two PBS washing steps were performed. Then, for laurdan staining, laurdan dissolved in DMSO was added to a final concentration of 5µM and cells were incubated at 37°C for 5minutes. Finally, a washing step with PBS was conducted prior to cell visualization. Similarly, PM patches formed as described above were stained with 5µM laurdan. GPMV were stained with laurdan by adding 5µM (final concentration) of the fluorophore dissolved in DMSO. GPMV were transferred to polylysinecoated glassbottom dishes (MatTek, Ashland, OR) and vesicles were left to sediment for 3hours before visualization.40 For GUV, 0.2mM lipid extracts in chloroform:methanol (2:1, v/v) were mixed with 0.01 mM laurdan. A quantity of 3 µL of the lipid stocks were added onto the surface of Pt electrodes and solvent traces were removed under high vacuum for at least 2hours. The Pt electrodes were then covered with 400μL of 300mM sucrose buffer and the Pt wires were connected to an electric wave generator (TG330 function generator, Thurlby Thandar Instruments, Huntington, UK) under alternating current field conditions (10 Hz, 2.5 VRMS for 2hours) at 37°C. After GUV formation, the chamber was placed on an inverted confocal fluorescence microscope for GUV visualization. 2.5.2 | Image acquisition and analysis A Leica TCS SP5 II microscope (Leica Microsystems GmbH, Wetzlar, Germany) was used for image acquisition. A 63x waterimmersion objective (numerical aperture NA =1.2) was used and samples were imaged at 512 × 512 pixel and 400Hz per scanning line. Equatorial planes were imaged to avoid photoselection effects. A pulsed titaniumsapphire (MaiTai Deepsee, SpectraPhysics) laser tuned at 780 nm was used for twophoton imaging of laurdanlabeled samples. Fluorescence emission was collected by nondescanned (NDD) hybrid detectors, as they offer higher sensitivity compared to descanned photomultipliers. The blue edge of the emission spectrum was collected by NDD 1 at 435±20nm and the red edge by NDD 2 at 500 ± 10 nm. Irradiance at the sample plane was ≈500 GW·cm– 2 for twophoton excitation.48 Generalized polarization (GP) value of samples was calculated using a MATLAB (MathWorks, Natick, MA)- based software. Images were smooth in each channel with 2 pixel averaging, and the GP value was calculated using the following Equation49: where IB is the intensity collected by NDD 1, IR is the intensity collected by NDD 2, and Gis the correction factor. The G factor is calculated measuring the GP value of the same fluorophore concentration used in sample staining, dissolved in this case in pure DMSO.50 In whole cell images, the region of interest, that is, the PM, was selected when required. 2.5.3 | Fluorescence spectroscopic analysis PM preparations, and SUV formed with whole cell or PM lipid extracts were measured in a spectrofluorometer. Samples (82.5µM lipid concentration) were labeled with 0.75µM laurdan. For this purpose, lipid extracts in chloroform:methanol (2:1) were mixed with laurdan and the solvent was evaporated to dryness under a stream of N2. Then, the sample was kept under vacuum for 2hours to remove solvent traces and the lipids were swollen in buffer (NaCl 150mM, Hepes 25mM, pH 7.4). Sonicated SUV were obtained as described above and fluorescence measurements were performed using a QuantaMaster 40 spectrofluorometer (Photon Technology International, Lawrenceville, NJ).51 2.6 | AFM Contact mode AFM imaging has been used to study bilayer topography, looking at possible lateral segregation effects through bilayer thickness analysis. A NanoWizard II AFM (JPKInstruments, Berlin, Germany) was used to perform topographic measurements under contact mode scanning (constant vertical deflection). For measurements, the AFM GP = I B− G ⋅ IR I B +G⋅I R 6 of 23 | MONASTERIO ET Al. was coupled to a Leica microscope and mounted onto a Halcyonics Micro 40 antivibration table (Halcyonics, Inc, Menlo Park, CA) and inside an acoustic enclosure (JPK Instruments, Berlin, Germany).52 Vshaped MLCT Si3N4 cantilevers (Bruker, Billerica, MA) with nominal spring constants of 0.1 or 0.5N/m were used. The sample thickness was estimated by crosssection height analysis.53 2.6.1 | Force spectroscopy Vshaped MLCT Si3N4 cantilevers (Bruker, Billerica, MA) with nominal spring constants of 0.1 or 0.5 N/m were individually calibrated in a lipidfree mica substrate in assay buffer using the thermal noise method. After proper bilayer area localization by means of AFM topography and direct epifluorescence microscopy, force spectroscopy was performed at a speed of 1μm/s. Force steps were determined for each of the indentation curves as reproducible jumps within the extended traces. At least three independent sample preparations were scanned for each case and 50100 curves were measured in each sample. Topographic images and force spectroscopy analysis of PM patches and supported planar bilayers (SPB) formed from lipid extracts, and force spectroscopy analysis of GPMV were performed. GPMV topographic observations could not be performed for experimental reasons; these structures would not flatten on the mica for AFM examination. SPB were prepared on high V2 quality scratchfree mica substrates (AshevilleSchoonmaker Mica Co., Newport News, VA). A quantity of 180 μL assay buffer containing 3mM CaCl2 was added onto a 1.2cm2 freshly cleaved mica substrate mounted onto a BioCell (JPK Instruments, Berlin, Germany). Then, 80μL sonicated 0.4mM SUV formed with CHO or LYB lipid extract was added on top of the mica. BioCell temperature was gradually increased (5°C every 5minutes) up to 80°C. Vesicles were let to adsorb and extend for 30minutes keeping the sample temperature at 80°C. Samples were left to equilibrate for 30minutes at room temperature before performing five washing steps with CaCl2free buffer in order to discard nonadsorbed vesicles and remove the remaining Ca2+ cations.52 Isolated PM patches for force spectroscopy were prepared as previously described,37,41 this time using polylysinecoated mica slips instead of glassbottom dishes. GPMV were first stained using Di4ANEPPQHD to allow detection on the mica slip. Then, samples were left for 3hours to sediment over the polylysinecoated mica slip before measurements were performed. 2.7 | Mass spectroscopic analysis Mass spectroscopic analysis was performed essentially as described in Monasterio et al.37 A methodological summary follows. 2.7.1 | Sample treatment Lipid extraction was performed using a modified methyl tertbutyl ether (MTBE) protocol.54 Briefly, cells were washed with cold PBS and scraped off in 500μL cold PBS on ice. The suspension was transferred to a 2ml tube in which it was spun down at 3200rpm for 5minutes at 4°C. After removing the PBS, samples were stored at −20°C or directly used for further extraction. GPMV and PM patch samples were prepared as previously mentioned. Then, 360μL of methanol was added and vortexed. A mixture of lipid standards (see Table1) was included and samples were vortexed for 10minutes at 4 ºC using a Cell Disruptor Genie (Scientific Industries, Inc, Bohemia, NY). MTBE (1.2mL) was then added and the samples were incubated for 1hour at room temperature with shaking (750rpm). Phase separation TABLE 1 MS detection conditions for the different lipid classes Lipid class Standard Polarity Mode m/z ion Collision energy Phosphatidylcholine [M+H]+DLPC + Product ion 184.07 30 Phosphatidylethanolamine [M+H]+PE31:1 + Neutral ion loss 141.02 20 Phosphatidylinositol [MH]- PI31:1 − Product ion 241.01 44 Phosphatidylserine [MH]- PS31:1 − Neutral ion loss 87.03 23 Cardiolipin [M2H]2CL56:0 − Product ion acyl chain 32 Ceramide [M+H]+C17Cer + Product ion 264.34 25 Dihydroceramide [M+H]+C17Cer + Product ion 266.40 25 Hexosylceramide [M+H]+C8GC + Product ion 264.34 30 Hexosyldihydroceramide [M+H]+C8GC + Product ion 266.40 30 Sphingomyelin [M+H]+C12SM + Product ion 184.07 26 | 7 of 23 MONASTERIO ET Al. was induced by adding 200μL H2O. After 10minutes incubation at room temperature, the sample was centrifuged at 1,000 x g for 10minutes. The upper (organic) phase was transferred to a 13mm screwcap glass tube and the lower phase was extracted with 400 μL artificial upper phase (MTBE/methanol/water (10:3:1.5, v/v/v)). The two upper phases were combined and the total lipid extract was divided into three equal aliquots [one for phospholipids (TL), one for sterols (S) in 2mL amber vials, and one for sphingolipid (SL) detection in a 13mm glass tube] and dried in a Centrivap at 50°C or under a nitrogen flow. The SL aliquot was deacylated by methylamine treatment (Clarke method) to remove glycerophospholipids. 0.5 mL monomethylamine reagent [MeOH/H2O/nbutanol/methylamine solution (4:3:1:5 v/v)] was added to the dried lipid, followed by sonication (5minutes). Samples were then mixed and incubated for 1hour at 53°C and dried (as above). The monomethylaminetreated lipids were desalted by nbutanol extraction. 300 μL H2Osaturated nbutanol was added to the dried lipids. The sample was vortexed, sonicated for 5minutes, and 150μL of MSgrade water was added. The mixture was vortexed thoroughly and centrifuged at 3200 x g for 10minutes. The upper phase was transferred to a 2mL amber vial. The lower phase was extracted twice more with 300μL H2Osaturated nbutanol and the upper phases were combined and dried (as above). 2.7.2 | Glycerophospholipid and sphingolipid detection in a Triple Quadrupole Mass Spectrometer TL and SL aliquots were resuspended in 250μL chloroform/ methanol (1:1 v/v) (LCMS/HPLC grade) and sonicated for 5minutes. The samples were pipetted in a 96well plate (final volume =100μL). The TL were diluted 1:4 in negativemode solvent (chloroform/methanol (1:2) + 5mM ammonium acetate) and 1:10 in positivemode solvent (chloroform/methanol/water (2:7:1 v/v) + 5mM ammonium acetate). The SL were diluted 1:10 in positivemode solvent and infused onto the mass spectrometer. Tandem mass spectrometry for the identification and quantification of sphingolipid molecular species was performed using Multiple Reaction Monitoring (MRM) with a TSQ Vantage Triple Stage Quadrupole Mass Spectrometer (Thermofisher Scientific, Waltham, MA) equipped with a robotic nanoflow ion source, Nanomate HD (Advion Biosciences, Ithaca, NY). The collision energy was optimized for each lipid class. The detection conditions for each lipid class are listed in Table1. Cer species were also quantified with a loss of water in the first quadrupole. Each biological replica was read in two technical replicas (TR). Each TR comprised three measurements for each transition. Lipid concentrations were calculated relative to the relevant internal standards and then normalized to the total lipid content of each lipid extract (mol %). 2.8 | Gas chromatographymass spectrometry for cholesterol assay Lipid extracts were analyzed by GCMS as described previously.55 Briefly, samples were injected into a VARIAN CP3800 gas chromatograph equipped with a FactorFour Capillary Column VF5ms 15m×0.32mm i.d. DF =0.10, and analyzed in a Varian 320 MS triple quadrupole with electron energy set to – 70eV at 250°C. Samples were applied to the column oven at 45°C, held for 4minutes, then temperature was raised to 195°C (20°C/min). Sterols were eluted with a linear gradient from 195 to 230°C (4°C/min), followed by rising to 320°C (10°C/min). Cholesterol was identified by its retention time (compared with an ergosterol standard) and fragmentation patterns, which were compared with the NIST library. 2.9 | Quantitation of lipids per cell An estimate of the amounts of lipids per cell, or per weight protein, was obtained as follows. Szeliova et al56 measured the average dry weight of CHO cells as 264pg/cell. Alberts et al57 indicated that the mammalian cell contained 10 dry wt% phospholipids and 7 dry wt% other lipids. Moreover, the average amount of protein per cell was measured experimentally with the BCA protein assay. Cell numbers were counted with a hemocytometer. From the above data, and knowing from MS analysis the concentration of a given lipid in a sample, the amount of such lipid per cell and per wt protein could be estimated. 3 | RESULTS 3.1 | CHOderived mutant cells can grow and survive with extremely low sphingolipid concentrations in the culture medium 3.1.1 | Growth and viability The extent to which CHO (wild type) and LYB (SPTdefective) cell adaptation to a SLdeficient medium affected cell division ratio and integrity was assessed. FBS was the only external source of SL for cell growth, thus FBS in the growth medium was the only SL source for LYB cells. SLdeficient growth media were prepared containing 0.04% FBS, that is, a 250fold decrease with respect to the standard conditions (10% FBS). Cell count measurements were performed using a BioRad TC20 hemocytometer. Figure1A shows a comparison between 8 of 23 | MONASTERIO ET Al. cell growth in standard (containing 10% FBS) or SLdeficient (containing 0.04% FBS) medium. Both cell lines (CHO and LYB) grew steadily for at least 96hours in full medium (10% FBS), and both divided, even if slowly, for the first 72hours in a lowSL medium. After 72hours in the SLlimited medium cell quantity decreased. The difference between CHO and LYB | 9 of 23 MONASTERIO ET Al. cellgrowth ratios was not statistically significant when they were grown in standard medium. Nevertheless, in SLdeficient medium CHO and LYB cells behaved differently. After 72hours, CHO cell number was 47% of control (complete medium), while LYB growth was only 18%. Growth rates of cells treated with the SPT inhibitor myriocin (2.5µM) 58 and grown in deficient medium were also measured as an additional control. CHO cell growth ratio was decreased to the level of LYB cells grown in deficient medium. Moreover, myriocin treatment did not affect the growth of LYB cells (data not shown). The results concur in suggesting that the different behavior of CHO and LYB cells in our study is due to the lack of an active SPT in the latter strain. Ancillary experiments were performed in which cell growth after 72hours in media with different degrees of SLlimitation was measured (FiguresS1A,B). Cell counts were performed as in Figure 1 (Figure S1A) and total cell protein was quantified with a BCA protein assay (FigureS1B). After 72hours, differences in growth were already significant between CHO and LYB cells when medium contained 5% FBS. No LYB cell growth could be reliably measured with FBS concentrations below 0.04%. In a different series of experiments, cell growth was intended on delipidated FBScontaining medium, but no cell division could be observed. To ascertain that the difference between CHO and LYB cell division ratio at low FBS concentrations was indeed due to a lack of SL, we tested whether LYB cells were able to reach the full growth rates when the SLdeficient medium was supplemented with SL. For this purpose, equimolar mixtures of egg PC and the sphingolipid under study were sonicated in buffer and added to the culture flasks in various amounts. The best recoveries were achieved with sphingomyelin (SM) or sphinganine (FigureS1E). Figure1B shows that LYB cells grown in SLdeficient medium for 72hours reached ≈80% of the control growth when 0.2mg SM was added per T25 culture flask (5mL; 80µM SM), while there was no difference in the case of CHO cells. With 0.0125mg sphinganine/flask (8µM), cells grown in SLdeficient medium under the same conditions recovered 81% of the highFBS control value (FigureS1C). Supplementation with pure PC vesicles did not have any effect when compared to the growth of LYB cells in the nonsupplemented, FBSdeficient medium (FigureS1E). Cerebroside (16% recovery at 0.2mg/flask, 55µM) and sphingosine (33% recovery at 0.02mg/ flask, 13µM) supplementations were also tested (FigureS1E). It was concluded that it is the lack of SPT activity what makes the main difference between CHO and LYB cell division ratios in SLdeficient medium. As a further control to ascertain that the main effects of lowering FBS concentration were due to the low supply of SL, the timecourse of cell growth in SLdeficient medium supplemented with SM and treated with sphingomyelinase inhibitors was measured (Figure1©). When cells were treated with5 µM acid sphingomyelinase inhibitorfluphenazine dihydrochloride or 20µM neutral sphingomyelinase inhibitor GW4869, the effect that SM addition had on cell growth recovery was suppressed. There was no statistically significant difference between the growth of cells treated with sphingomyelinase inhibitor grown in SM supplemented deficient medium and the nontreated LYB cells grown in deficient medium (Figure1C). The possible effect of low cell growth/ SLdeficient medium on cell viability was then tested using flow cytometry analysis with AnnexinVFITC and propidium iodide. Flow cytometry analyses demonstrated that, despite the low growth rate, 86% of LYB cells grown in SLdeficient medium for 72hours remained viable (FigureS2E). Ethanoltreated CHO cells (Figure S2A) were used as a positive control for nonviable cells. 95% CHO cells (FigureS2B) and 95% LYB cells (FigureS2D) grown in standard medium were viable. With respect to cells grown in SLdeficient medium, 92% CHO (FigureS2C) and 86% LYB (FigureS2E), as well as 89% LYB grown in SMsupplemented medium (Figure S2F) were viable. These plots are representative data used to quantify results shown in Figure1A and S1. Considering that even LYB cells grown in 0.04 FBS retained a fair viability, these cells were considered as a good tool to obtain reliable information on the putative effects of a defective SPT activity on their biophysical properties. 3.1.2 | Lysenin staining Cells were stained with SMspecific NTlyseninmCherry and visualized with confocal microscopy. LYB cells grown FIGURE 1 LYB cells can grow on very small amounts of sphingolipids. A, LYB (empty symbols) and CHO (filled symbols) cell growth as a function of time in standard (10% FBS) (triangles) and sphingolipiddeficient (0.04% FBS) (circles) medium. Inset: LYB cell growth in sphingolipiddeficient medium, Yaxis expanded (B) LYB and CHO cell growth after 72hours in sphingolipiddeficient medium supplemented with SM (seeded cells: 0.25 × 106; medium volume =5mL)). C, LYB cell growth as a function of time in sphingolipiddeficient (0.04% FBS) medium supplemented with SM (0.2mg ≈ 80µM) (circles) and treated with sphingomyelinase inhibitors, fluphenazine or GW4869 (squares) (seeded cells: 0.25 × 106). In AC, data correspond to average values ± SD (n=3). In A, error bars are smaller than the symbols. Fluorescence images and flow cytometrymediated mCherrylysenin (red) quantification. CHO (D) and LYB (E) cells grown in standard medium. CHO (F) and LYB (G) cells grown in SLdeficient medium. NBDPE (green) was used in fluorescence images for general membrane staining. Measurements are shown after 72h growth. Geometric mean ± SD (n=3). H, Flow cytometry: timedependent mCherrylysenin quantification in LYB cells grown (from left to right) in standard medium, or deficient medium for 24, 48hours, and 72hours. Histograms in black correspond to control cells (without mCherrylysenin staining) and those in red, to the sample of interest (mCherrylysenin signal). Geometric mean ± SD (n=3) 16 of 23 | MONASTERIO ET Al. in medium containing different FBS concentrations. These three specific SL were selected among the lipidomic data respectively because SM is the most abundant SL, Cer is particularly important in cell signaling, and HexCer is at the origin of the biosynthetic pathway leading to the complex glycosphingolipids. Significant differences were seen between SM amounts in CHO and LYB cells (Figure 6A). When cells were grown in standard medium, the SPTdeficient LYB cells contained 43% less SM than CHO cells. When the amount of FBS in the medium was decreased, total SM was also lower in both wild type and mutant cells. Nevertheless, the decrease was greater in LYB than in CHO cells, 90% versus 25% with the lowest FBS concentration (0.04%) (Figure 6A). In cells grown in standard medium, FIGURE 6 Lipidomic analysis of CHO and LYB cells. Total SM (A), Cer (B) and HexCer (C) from CHO (black bars) or LYB (gray bars) whole cells grown in media containing various concentrations of FBS. D, E, A comparison of lipid compositions of whole cell and PM patches of LYB cells grown in standard or sphingolipiddeficient (0.04% FBS) media. Only selected lipids are included in the figure. A comprehensive description of the various lipid compositions can be seen in the Supplementary Material TableS1 and Figure7. (F) Fully saturated (DB0 = no double bonds) and (G) shortchain (3032C) GPL of whole cells treated for GPMV preparation, GPMV, cells treated for PM patch preparation, and PM patches. n=3. Statistical significance was calculated with ANOVA or Student´s ttest, with similar results. Significance: (*) P<.05; (**) P<.01; (***) P<.001. Lipid percentages were computed over the total lipid amount measured with the mass spectrometer | 17 of 23 MONASTERIO ET Al. LYB contained 70% less Cer than CHO cells. When FBS concentration was decreased, the total Cer amount was also diminished in both cell lines. As with SM, the reduction was larger in LYB than in CHO cells, 66% vs. 38% (in 0.04% FBScontaining medium). Figure6C shows the corresponding values for HexCer. LYB grown in 10%- FBS medium contained 70% less HexCer than CHO, in agreement with the Cer data (Figure 6B). When FBS in the medium was decreased, HexCer also decreased by 50% in LYB cells grown in 0.04% FBScontaining medium, but it increased by 85% in CHO (Figure6C). Considering the PM patches, SM, Cer and HexCer were all lower in LYB 0.04 PM patches than in LYB 10 ones. In summary, in LYB cells the three SL under study exhibited a similar decrease (55%- 70%) with the reduction of FBS, at variance with CHO cells, suggesting that, in the latter, an active de novo SL synthesis could occur. FIGURE 7 Lipidomic analysis of whole cells and plasma membrane preparations. SM (A), Cer (B), Hex Cer (C), Chol (D), PC (E), PE (F), PI (G), PS (H), CL (I). JL: GPL saturation distribution (number of double bonds per GPL molecule). No double bonds (DB 0) (J), one double bond per molecule (DB 1) (K), two – six double bonds per molecule (DB 26) (L). MO: GPL length distribution (number of C atoms in the two acyl chains). 3032 C (M), 3440 C (N), 4244 C (O). Mean values ± SD (n=3). Statistical significance was calculated with ANOVA and Student´s ttest: (*) P<.05; (**) P<.01; (***) P<.001. Lipid percentages were computed over the total lipid amount measured with the mass spectrometer 18 of 23 | MONASTERIO ET Al. The situation with GPMV was, however, different, mainly in that Cer and HexCer increased when the cells (either CHO or LYB) were grown in 0.04 FBS medium (Figure7B,C). With lowFBS medium, Chol levels were decreased in both cell lines (Figure7D). Since FBS is a major source of lipids and proteins for cell growth under our conditions, the drastic reduction from 10% to 0.04% in the SLdeficient medium induces partial cell starvation. Chol synthesis is known to decrease in fasting conditions.6769 Moreover, LYB PM preparations had larger amounts of Chol than the whole cell average, for cells grown in highand lowFBS (Figure6D). This had been observed by Monasterio et al37 for the case of CHOcells. Glycerophospholipid (GPL) acyl chain also changed along with SL deprivation. In Figure6E we can see that ether PC was increased in LYB 0.04 cells. GPL acyl chain saturation and length also play an important role in the physical properties of the membrane bilayer, specifically on its disorder/fluidity. Specifically, unsaturated and shorter acylchaincontaining GPL increase membrane fluidity.70 In Figure6F, the distribution of fully saturated GPL of control and two PM preparations (patches and GPMV) is shown. PM preparations had more fully saturated and less polyunsaturated (26 double bounds) GPL (Figure7JL) than their respective controls in all measured samples.37 The differences were larger in the case of GPMV. Comparing CHO and LYB grown in standard medium, LYB had more fully saturated GPL in all cases. In addition, when FBS in the medium was decreased, the saturated GPL increased in LYB, while in CHO cells they remained almost constant (Figure6F). As to the GPL chain length distribution, PM preparations were richer in 3032 C chains than whole cells, the difference being larger for the GPMV (Figure 6G).37 GPL of LYB grown in standard medium contained more 3032 C acyl chains than CHO 10 in all measured samples. This difference was increased when LYB were grown in 0.04% FBS medium. Conversely, CHO 0.04 chainlength values remained constant (Figure6G). In summary LYB cells synthesized shorter and more saturated GPL in their homeostatic response to SM depletion. A comprehensive description of the various lipid compositions can be seen in the TableS1 and Figure7. The homeostatic response undergone by LYB cells grown in deficient medium and supplemented with SM (Figure5E) was also observed. Data are shown as percent increase or decrease when compared to the amounts found in CHO cells grown in deficient medium. SM and Cer levels were totally recovered, becoming even higher than the control values. However, for reasons that remain unexplained, the HexCer CHO amounts were not reached. PC ethers increased when LYB cells were grown in deficient medium; this increase was suppressed with SM supplementation. Fully saturated GPL levels were slightly decreased with SM supplementation (Figure5E). A quantitative estimate of the amount of SM and Chol, two representative lipids in this context, was carried out as described under Methods. The results can be seen in FigureS8. With respect to Chol, the concentration in CHO cells grown in SLmedium was 33% of those grown in standard medium (data in pg Chol/cell), and the corresponding figure for LYB was 45% (FigureS8B). As for SM data, CHO cells grown on medium with 0.04% FBS contained 68% of the SM found in cells grown on 10% FBS (data in pg SM/cell) (FigureS8B), or 53% (in pg SM/ng protein) (FigureS8A). For the SPTdefective LYB cells, the corresponding figures are 15% (FigureS8B), and 16% (FigureS8A). Thus, a 250fold reduction in sphingolipid supply to LYB cells leads to a 6fold decrease in membrane sphingolipids. As a summary of the lipidomic results, SM, Cer, and HexCer concentrations were lower in LYB PM patches than in CHO ones when grown in standard medium. All three SL were similarly decreased (55%- 70%) with the reduction in FBS. LYB cells contained larger amounts of Chol than CHO ones in both standard and SLdeficient media. With respect to the GPL fatty acyl distribution, LYB had more saturated and shorter GPL fatty acids than CHO cells. These groups of fatty acids, together with PC ethers, were increased in LYB and maintained in CHO when FBS concentration was decreased. For LYB in SLdeficient medium, SM decreased both in GPMV and in PM patches but Cer and HexCer were increased with lower FBS concentrations (Figure7AC). 4 | DISCUSSION LYB cells grown in SLdeficient medium were used to understand the effects that a defective SPT activity might have on the biophysical properties of the cell. SPTdefective cells grown with very low SL concentrations were viable (FigureS2E) and they were able to recover the control growth rates when the SLdeficient medium was supplemented with SM (Figure1B) or sphinganine (FigureS1C). 4.1 | CHO and LYB cells grown in standard medium 4.1.1 | Whole cells Comparing SL levels in CHO and LYB cells grown in standard medium, they happened to be markedly lower in the mutant cells. Considering the three most abundant SL, SM was 43% lower in LYB cells (Figure 6A), Cer was 66% lower (Figure 6B), and HexCer was 70% lower (Figure 6C). This indicated that the de novo pathway could be a major SL synthesis source. This result is in agreement with the one published by Ziulkoski et al30 where they used fumonisin B1 and | 19 of 23 MONASTERIO ET Al. βchloroalanine to determine the contribution of the different pathways to the synthesis of SM in Sertoli cells. They found that 40% of 16:0 and 61% of 18:0, 18:1 and 18:3 SM was synthesized by the de novo pathway. They also observed that these values could be increased when the requirement for cell membranes was greater, as in rapidly dividing cells.30 In contrast with the SL results, Chol concentrations were similar in CHO and LYB cells grown in standard medium (Figure7D). In addition, comparison of both kinds of cells in standard medium showed small changes in GPL, LYB contained more fully saturated, and less monounsaturated GPL than CHO, chain length distributions in GPL being virtually the same (Figure6F,G). 4.1.2 | PM preparations Important differences were found between the whole cell and PM lipid compositions, as anticipated from the studies in CHO by Monasterio et al37 Both for CHO and LYB preparations, PM patches contained less SM (about one half), more HexCer and more Chol than the whole cells. GPMV had also less SM but contained higher amounts of HexCer, and particularly of Cer and Chol. Changes in GPL were moderate or low, except for cardiolipin, that was almost absent in the PM preparations (Figure7AD,I). The higher amounts of Chol in PM patches (Figure7D) could be a major factor responsible for compensating PM molecular order even with lower SM levels. Neither laurdan GP of GPMV nor laurdan GP or breakthrough forces of PM patches showed any statistically significant difference between CHO and LYB grown in 10% FBS (Table2, Figure3C). GPMV constitute a frequently used PM preparation.37,40,71,72 However, the lipidomic data showed that their lipid composition departed from those of the whole cells and from other PM preparations (patches). In particular, GPMV were enriched in Cer (Figure7B) and HexCer (Figure7C), and they also exhibited an unusual enrichment in PI (Figure7G). Furthermore, their GPL were enriched in saturated fatty acids (Figure7J), and contained correspondingly less unsaturated chains, specifically with 26 double bonds per GPL molecule (Figure7L). Also, the proportion of mediumlength fatty acids (C3032 per GPL molecule) increased at the expense of the longer ones (C3440) (Figure7M,N). All these are peculiarities of GPMV, in which they differed from all other cell and membrane preparations, with either 10% or 0.04% FBS. The fact that these changes were not modified by SL depletion, and that some of them affected mainly GPL, makes GPMV a less useful membrane preparation in the context of our study. GPMV penetration required consistently higher breakthrough forces than PM patches (FigureS7). This could be related to the enrichment in Cer and HexCer found in GPMV with respect to whole cells (Figure7B,C). Both Cer and HexCer are known to increase membrane lipid order7375 GPMV have been shown to be permeable to hydrophilic macromolecules,72 and this could again be related to the increased Cer, and partly HexCer, since these SL happen to increase membrane permeability.73,74 The observed increases in GP and breakthrough forces could be secondary to the use of dithiothreitol in GPMV formation. As seen in Epstein et al76 dithiothreitol can be responsible for increasing Cer concentrations even in SPTsuppressed cell lines, without altering SM values. Those authors concluded that dithiothreitol could induce the “unfolded protein response” and this would lead to an overexpression of the SPT LCB1 subunit mRNA, partially recovering its activity.76 Dithiothreitol was also shown to affect lipidlipid and lipid– protein interactions and to integrate directly into lipid membranes.77 4.2 | CHO and LYB cells grown in SLdeficient medium When LYB cells were grown in SLdeficient medium, SM and Chol percent levels were markedly decreased, respectively by about fivefold and twofold, with no comparable changes in Cer or HexCer, and the derived PM patches followed parallel trends (Figures6A,D, 7A,D). In CHO cells, the decrease in SM concentration was less clear, and HexCer levels actually increased somewhat, other SL varying as in LYB (Figure6A,C), with the corresponding PM patches showing similar trends (Figure7AD). Growth in SLdeficient medium did not cause any remarkable changes in GPL, nor in their associated fatty acids (Figure7EN), with the exception that the very long fatty acids (C4244 per GPL molecule) whose concentration was in any case very low, were further decreased with the low FBS medium. Note that the largest decrease in SM, the most abundant SL, occurred in LYB cells deprived of SL in the growth medium, thus the two factors appear to be required, lack of SL in the nutrients and lack of capacity to synthesize the sphingosine precursor, to obtain lowSL cells. With respect to the PM preparation, laurdan GP indicated a decreased lipid order (increased bilayer fluidity) in all samples under study (Table2) and breakthrough forces decreased accordingly, more in LYB than in CHO cells and membranes (Figure3B,C). As a result, PM patches from LYB cells were less ordered and more easily penetrable (Table2, Figure3C). The close correlation between decrease in SM concentration in cell membranes, as a result of SL deprivation in the nutrients, decrease in GP values and decrease in breakthrough forces can be seen in Figure3C,D. (Only data from whole cell lipid extracts are included in Figure3D, for simplicity.) 4.3 | Homeostatic adaptations At least some of the observed changes in membrane lipid composition as a result of gene suppression or of changes in 20 of 23 | MONASTERIO ET Al. nutrient media could be explained in terms of homeostatic responses to the novel situations. Perhaps the main observation in terms of adaptation is the remarkable resilience of LYB cells that, when grown under extremely low SL concentrations (250fold below standard conditions), are still able to divide while keeping SL concentrations just sixfold lower than the standard value, and membrane physical properties not far away from the wildtype cells. Examining the data in more detail, and specifically comparing CHO and LYB cells grown in 10% FBS medium, hints on adaptation to lack of de novo SL synthesis could be retrieved. In particular, as described above, the only notable change between the lipidomes of those two cell lines, grown under standard conditions, is the clear decrease in SL as percent total lipids (onehalf on average) in the LYB cells (Figure7), while Chol levels did not vary. The percent concentration of SM, the most abundant sphingolipid, went from 6.7% to 3.8% (Figure 7A). Parallel changes were recorded in PM patches derived from those cells. This was not accompanied by any changes in the measured physical properties of the membranes, laurdan GP (Table2) or AFM breakthrough forces (Figure3C). Perhaps the observed variation in SL concentration was not enough to cause any observable physical changes, and a very minor, or no adaptation was required. The situation was different when cells were grown in SLdeficient medium. In LYB cells SM concentration dropped by one order of magnitude when cells were grown in 0.04% instead of 10% FBS. Other SL, as well as Chol, were decreased in parallel. PM patches underwent similar changes as the whole cell lipids. Perhaps as a consequence of these changes, the membranes became more easily penetrable, and lipids became less ordered (Table2 and Figure3) when FBS concentration was lowered. Simultaneously, fatty acyl unsaturation was decreased (Figure7JL), a phenomenon that could have the effect of increasing lipid order, thus tending to compensate the decrease in bilayerordering SM. Our results regarding the increase in ether lipids in sphingolipiddepleted cells fit previous observations 78 and confirm the coregulation between these two lipid classes. When CHO cells were grown in SLdeficient medium the proportion of HexCer was considerably increased, by about twofold. This increased HexCer synthesis (that could not occur in LYB cells because of their low sphingosine availability, due to the lack of SPT activity), may be one of the homeostatic responses that wildtype cells carry out under starvation. HexCer is at the origin of the complex glycosphingolipid biosynthetic pathway.1 Glycosphingolipids are required for cellular differentiation and there are human diseases resulting from defects in their synthesis.79 This may be one of the reasons for the different dividing ratios of CHO and LYB in SLdeficient medium (Figure 1A). CHO, but not LYB cells, may overexpress the HexCer synthesis to continue cell division in order to buffer the nutrient depletion condition. As discussed above, Chol levels in CHO and LYB grown in standard medium remained invariant, and this could help in maintaining membrane rigidity under conditions of low SM (Figure7D, J). Nevertheless, as Chol synthesis decreases under cell starvation conditions, rigidity cannot be maintained in this manner under SLdeficient conditions. When FBS in the medium was decreased, saturated GPL were increased in LYB (Figure6F), while in CHO cells they remained almost constant. 4.4 | CONCLUDING REMARKS The present study has demonstrated that in cells lacking the SPT activity, SM, Cer and HexCer are markedly decreased in all measured samples (controls and PM preparations). Fully saturated GPL are increased and polyunsaturated ones are decreased. Synthesizing more saturated GPL can be the way that LYB cells have to compensate the low SM. Cholesterol may also have some influence in that response but its effect is minimized when its levels are decreased because of starvation. The SMdepleted cells try to maintain membrane order by undergoing a homeostatic response, although they achieve it only partially as their PM are more fragile when grown in SLdeficient medium. These changes in lipid order and membrane rigidity caused by low SL could be linked to a variety of phenomena in cell physiology and pathology. Alterations in the normal activities of the SMcycle enzymes have been associated to many central nervous system and neurodegenerative diseases.19 Specific SM species have been found to bind membrane proteins thereby modifying their functions.80 The capacity shown by certain cells in this paper to grow under extremely demanding low concentrations of SL opens the way to a variety of functional studies on the role of SL in membranes. ACKNOWLEDGMENTS We thank Dr Alfred Merrill (Georgia Tech) for suggesting that we examine the properties of LYB cells in sphingolipiddepleted medium. The authors are also grateful to Dr Donatienne Tyteca for the kind gift of plasmid pET28/DronpaNTlysenin and to Ms Marina Iriondo for her help in purifying lysenin. This work was supported in part by grants from the Spanish Ministry of Economy (grant FEDER MINECO PGC2018099857BI00) and the Basque Government (grants No. IT126419 and IT127019), as well as Fundación Biofísica Bizkaia and the Basque Excellence Research Centre (BERC) program of the Basque Government, and by the Swiss National Science Foundation (310030184949). | 21 of 23 MONASTERIO ET Al. CONFLICT OF INTEREST The authors have stated explicitly that there are no conflicts of interest in connection with this article. AUTHOR CONTRIBUTIONS FM Goni and A. Alonso designed the research; BG Monasterio, N. JimenezRojo, and A. GarciaArribas performed the research; H. Riezman contributed analytic tools (mass spectrometry); all authors analyzed the data; BG Monasterio and N. JimenezRojo wrote the manuscript, which was edited by FM Goni, H. Riezman, and A. Alonso. REFERENCES 1. Merrill AHJ, Schmelz EM, Dillehay DL, et al. Sphingolipids– the enigmatic lipid class: biochemistry, physiology, and pathophysiology. Toxicol Appl Pharmacol. 1997;142:208222. 2. Nieto FL, Pescio LG, Favale NO, Adamo AM, SterinSpeziale NB. Sphingolipid metabolism is a crucial determinant of cellular fate in nonstimulated proliferating MadinDarby Canine Kidney (MDCK) cells. J Biol Chem. 2008;283:2568225691. 3. AdachiYamada T, Gotoh T, Sugimura I, et al. De novo synthesis of sphingolipids is required for cell survival by downregulating cJun Nterminal kinase in Drosophila imaginal discs. Mol Cel Biol. 2015;19:72767286. 4. Li Z, Kabir I, Tietelman G, et al. Sphingolipid de novo biosynthesis is essential for intestine cell survival and barrier function. Cell Death Dis. 2018;9:113. 5. Chiantia S, London E. Sphingolipids and membrane domains: recent advances. Handb Exp Pharmacol. 2013;215:3355. 6. Goñi F, Alonso A. Effects of ceramide and other simple sphingolipids on membrane lateral structure. Biochim Biophys Acta. 2009;1788:169177. 7. Simons K, Ikonen E. Functional rafts in cell membranes. Nature. 1997;387(6633):569– 572. https://doi.org/10.1038/42408 8. Goñi FM. “Rafts”: a nickname for putative transient nanodomains. Chem Phys Lipids. 2019;218:3439. 9. Yasuda T, Al Sazzad M, Jäntti N, Pentikäinen O, Slotte J. The influence of hydrogen bonding on sphingomyelin/colipid interactions in bilayer membranes. Biophys. 2016;110(2):431– 440. https://doi. org/10.1016/j.bpj.2015.11.3515 10. Boggs JM. Lipid intermolecular hydrogen bonding: influence on structural organization and membrane function. BBA— Rev Biomembr. 1987;906:353404. 11. Schmidt CF, Barenholz Y, Thompson TE. A nuclear magnetic resonance study of sphingomyelin in bilayer systems. J Phys Soc Jpn. 1977;42:719720. 12. Estep T, Mountcastle D, Barenholz Y, Biltonen R, Thompson T. Thermal behavior of synthetic sphingomyelincholesterol dispersions. Biochemistry. 1979;18:21122117. 13. Keyvanloo A, Shaghaghi M, Zuckermann M, Thewalt J. The phase behavior and organization of sphingomyelin/cholesterol membranes: a deuterium NMR study. Biophys J. 2018;114:13441356. 14. Busto J, GarciaArribas A, Sot J, Torrecillas A, GomezFernandez JC, Goni F, Alonso A. Lamellar gel (lbeta) phases of ternary lipid composition containing ceramide and cholesterol. Biophys. 2014;106:621630. 15. Albi E, Magni MV. Sphingolipid metabolism inhibitors and cell function. Open Enz Inhibi J. 2008;1:7279. 16. Kolesnick R. The therapeutic potential of modulating the ceramide/ sphingomyelin pathway. J Clin Invest. 2002;1:38. 17. Spiegel S, Milstien S. Exogenous and intracellularly generated sphingosine 1phosphate can regulate cellular processes by divergent pathways. Biochem Soc Trans. 2003;31:12161219. 18. Luberto C, Kraveka J, Hannun Y. Ceramide regulation of apoptosis versus differentiation: a walk on a fine line. Neurochem Res. 2002;27:609617. 19. Kamil B, Anna F, Anna S, Sławomir P, Halina C. Regulation of sphingomyelin metabolism. Pharmacol Rep. 2016;68:570581. 20. Ziesemer S, Möller N, Nitsch A, Müller C, Beule AG, Hildebrandt JP. Sphingomyelin depletion from plasma membranes of human airway epithelial cells completely abrogates the deleterious actions of S. aureus alphatoxin. Toxins. 2019;11:126. 21. Holleran W, Feingold K, Man M, Gao W, Lee J, Elias P. Regulation of epidermal sphingolipid synthesis by permeability barrier function. J Lipid Res. 1991;32:11511158. 22. Feingold K. The regulation of epidermal lipid synthesis by permeability barrier requirements. Crit Rev Ther Drug Carr Syst. 1991;8:193210. 23. Dougherty AM, McDonald FE, Liotta DC, et al. Synthesis of 1deoxysphingosine derivatives with conformationally restricted pyrrolidinediol head groups. Org Lett. 2006;8:649652. 24. Fujita T, Inoue K, Yamamoto S. A potent immunosuppressive activity found in Isaria sinclairii metabolite. J Antibiot. 1994;47:208215. 25. Huang S, Liu K, Jiang D, Fang D. Codetermination of sphingomyelin and cholesterol in cellular plasma membrane in sphingomyelindepletioninduced cholesterol efflux. Anal Chem. 2019;91:15011506. 26. Mikłosz A, Łukaszuk B, Baranowski M, Górski J, Chabowski A. Effects of inhibition of serine palmitoyl transferase (SPT) and sphingosine kinase 1 (SphK1) on palmitate induced insulin resistance in L6 myotubes. PLoS ONE. 2013;8:19. 27. Li H, Yun HY, Baek KJ, Kwon NS, Park KC, Kim DS. Myriocin, a serine palmitoyltransferase inhibitor, increases melanin synthesis in MelAb cells and a skin equivalent model. Pharmazie. 2014;69:187191. 28. Gault C, Obeid L, Hannun Y. An overview of sphingolipid metabolism: from synthesis to breakdown introduction to sphingolipid metabolism. Adv Exp Med Biol. 2010;688:123. 29. Merrill AH, Wang MD, Park M, Sullards MC. (Glyco)sphingolipidology: an amazing challenge and opportunity for systems biology. Trends Biochem Sci. 2007;32:457468. 30. Ziulkoski AL, Zimmer AR, Guma FCR. De novo synthesis and recycling pathways of sphingomyelin in rat sertoli cells. Biochem Bioph Res Commun. 2001; 281:971975. 31. Hanada K, Hara T, Fukasawa M, Yamaji A, Umeda M, Nishijima M. Mammalian cell mutants resistant to a sphingomyelindirected cytolysin. J Biol Chem. 1998;273:3378733794. 32. Nakamura H, Wakita S, Yasufuku K, et al. Sphingomyelin regulates the activity of secretory phospholipase A2 in the plasma membrane. J Cell Biochem. 2015;116:18981907. 33. Mise K, Akifusa S, Watarai S, Ansai T, Nishihara T, Takehara T. Involvement of ganglioside GM3 in G2/M cell cycle arrest of human monocytic cells induced by Actinobacillus actinomycetemcomitans cytolethal distending toxin. Infect Immun. 2005;73:48464852. 34. Han G, Gupta SD, Gable K, et al. Identification of small subunits of mammalian serine palmitoyltransferase that confer distinct acylCoA substrate specificities. Proc Natl Acad Sci. 2009;106:9931. 22 of 23 | MONASTERIO ET Al. 35. Bejaoui K, Uchida Y, Yasuda S, et al. Hereditary sensory neuropathy type 1 mutations confer dominant negative effects on serine palmitoyltransferase, critical for sphingolipid synthesis. J Clin Invest. 2002;110:13011308. 36. Chakraborty M, Jiang X. Sphingomyelin and its role in cellular signaling. Adv Exp Medi Biol. 2013;991:114. 37. Monasterio BG, JiménezRojo N, GarcíaArribas AB, Riezman H, Goñi FM, Alonso A. Patches and blebs: A comparative study of the composition and biophysical properties of two plasma membrane preparations from CHO cells. Int J Mol Sci. 2020;21:2643. 38. GalisteoGonzález F, Monasterio BG, Gil D, Valle M, Goñi F. Photoacoustic effect applied on model membranes and living cells: direct observation with multiphoton excitation microscopy and longterm viability analysis. Sci Rep. 2020;10:299. 39. Ahyayauch H, GarcíaArribas AB, Sot J, et al. Pb (II) induces scramblase activation and ceramidedomain generation in red blood cells. Sci Rep. 2018;8:117. 40. Manni MM, Sot J, Goñi FM. Interaction of Clostridium perfringens epsilontoxin with biological and model membranes: A putative protein receptor in cells. Biochim Biophys Acta Biomembr. 2015;1848:797804. 41. Bezrukov L, Blank PS, Polozov IV, Zimmerberg J. An adhesionbased method for plasma membrane isolation: Evaluating cholesterol extraction from cells and their membranes. Anal Biochem. 2009;394:171176. 42. Montes L, Ahyayauch H, Ibarguren M, et al. Electroformation of giant unilamellar vesicles from native membranes and organic lipid mixtures for the study of lipid domains under physiological ionicstrength conditions. Methods Mol Biol. 2010;606:105114. 43. Angelova D, Dimitrov M. Liposome electro formation. Faraday Discuss Chem Soc. 1986;81:303311. 44. Carquin M, Pollet H, VeigadaCunha M, et al. Endogenous sphingomyelin segregates into submicrometric domains in the living erythrocyte membrane. J Lipid Res. 2014;55:13311342. 45. Maliekal P, Vertommen D, Delpierre G, Schaftingen EV. Identification of the sequence encoding Nacetylneuraminate9phosphate phosphatase. Glycobiol. 2006;16:165172. 46. Veigadacunha M, Hadi F, Balligand T, Stroobant V, Schaftingen EV. Molecular identification of hydroxylysine kinase and of ammoniophospholyases acting on 5PhosphohydroxyLlysine and phosphoethanolamine. J Biol Chem. 2012;287:72467255. 47. Krasnowska EK, Gratton E, Parasassi T. Prodan as a membrane surface fluorescence probe: Partitioning between water and phospholipid phases. Biophys J. 1998;74:19841993. 48. Parasassi T, Gratton E, Yu WM, Wilson P, Levi M. Twophoton fluorescence microscopy of laurdan generalized polarization domains in model and natural membranes. Biophys J. 1997;72:24132429. 49. Carravilla P, Nieva JL, Goñi FM, Requejoisidro J, Huarte N. Twophoton laurdan studies of the ternary lipid mixture DOPC:SM: Cholesterol reveal a single liquid phase at Sphingomyelin: Cholesterol ratios lower than 1. Langmuir. 2015;31:28082817. 50. Owen DM, Rentero C, Magenau A, AbuSiniyeh A, Gaus K. Quantitative imaging of membrane lipid order in cells and organisms. Nat Prot. 2012;7:2435. 51. Santis AD, Varela Y, Sot J, Errico GD, Goñi FM, Alonso A. Omega3 polyunsaturated fatty acids do not fluidify bilayers in the liquidcrystalline state. Sci Rep. 2018;8:113. 52. Monasterio BG, Alonso B, Sot J, et al. Coating graphene oxide with lipid bilayers greatly decreases its hemolytic properties. Langmuir. 2017;33:81818191. 53. GarcíaArribas AB, Busto JV, Alonso A, Goñi FM. Atomic force microscopy characterization of palmitoylceramide and cholesterol effects on phospholipid bilayers: A topographic and nanomechanical study. Langmuir. 2015;31:31353145. 54. Guri Y, Colombi M, Dazert E, et al. mTORC2 promotes tumorigenesis via lipid synthesis. Cancer Cell. 2017;32:807823. 55. Guan XL, Riezman I, Wenk MR, Riezman H. Yeast lipid analysis and quantification by mass spectrometry. Methods Enzym. 2010;470:369391. 56. Széliová D, Ruckerbauer DE, Galleguillos SN, et al. What CHO is made of: Variations in the biomass composition of Chinese hamster ovary cell lines. Metab Eng. 2020;61:288300. 57. Alberts B, Bray D, Lewis J, Raff M, Roberts K, Watson J. Molecular biology of the cell. Garland Sc; 1994. 58. Capasso S, Sticco L, Rizzo R, et al. Sphingolipid metabolic flow controls phosphoinositide turnover at the transGolgi network. EMBO J. 2017;36:17361754. 59. Frisz JF, Lou K, Klitzing HA, et al. Direct chemical evidence for sphingolipid domains in the plasma membranes of fibroblasts. Proc Natl Acad Sci. 2013;110:E613E622. 60. Frisz JF, Klitzing HA, Lous K, et al. Sphingolipid domains in the plasma membranes of fibroblasts are not enriched with cholesterol. J Biol Chem. 2013;288:1685516861. 61. Nicovich PR, Kwiatek JM, Ma Y, Benda A, Gaus K. FSCS reveals the complexity of lipid domain dynamics in the plasma membrane of live cells. Biophys J. 2018;114:28552864. 62. Agrawal H, Zelisko M, Liu L, Sharma P. Rigid proteins and softening of biological membraneswith application to HIVinduced cell membrane softening. Sci Rep. 2016;6:112. 63. Shchelokovskyy P, TristramNagle S, Dimova R. Effect of the HIV1 fusion peptide on the mechanical properties and leaflet coupling of lipid bilayers. New J Phys. 2011;13:117. 64. Bouvrais H, Méléard P, Pott T, Jensen KJ, Brask J, Ipsen JH. Softening of POPC membranes by magainin. Biophys Chem. 2008;137:712. 65. Tristramnagle S, Chan R, Kooijman E, et al. HIV fusion peptide penetrates, disorders, and softens TCell membrane mimics. J Mol Biol. 2010;402:139153. 66. Umagai IZK, Keda KAI, Obayashi TOK, Ada HIW. Imaging by atomic force microscopy of the plasma membrane of prestintransfected Chinese Hamster Ovary cells. J Assoc Res Oto. 2006;278:267278. 67. Smith WS, Baker EJ, Holmes SE, et al. Membrane cholesterol is essential for triterpenoid saponin augmentation of a saporinbased immunotoxin directed against CD19 on human lymphoma cells. Biochim Biophys Acta Biomembr. 2017;1859:9931007. 68. Kovanen P, Nikkilä E, Miettinen T. Regulation of cholesterol synthesis and storage in fat cells. J Lipid Res. 1975;16:211223. 69. Désert C, Duclos M, Blavy P, et al. Transcriptome profiling of the feedingtofasting transition in chicken liver. BMC Genom. 2008;9:611. 70. van Meer G, Voelker DR, Feigenson GW. Membrane lipids: where they are and how they behave. Nat Rev Mol Cell Biol. 2008;9:112124. 71. Sezgin E, Kaiser H, Baumgart T, Schwille P, Simons K, Levental I. Elucidating membrane structure and protein behavior using giant plasma membrane vesicles. Nat Protoc. 2012;7:10421051. 72. Skinkle A, Levental K, Levental I. Cellderived plasma membrane vesicles are permeable to hydrophilic macromolecules. Biophys J. 2020;118:12921300. | 23 of 23 MONASTERIO ET Al. 73. Alonso A, Goñi F. The physical properties of ceramides in membranes. Annu Rev Biophys. 2018;47:633654. 74. GonzálezRamírez E, Goñi F, Alonso A. Mixing brain cerebrosides with brain ceramides, cholesterol and phospholipids. Sci Rep. 2019;9:13326. 75. Varela A, Ventura A, Carreira A, et al. Pathological levels of glucosylceramide change the biophysical properties of artificial and cell membranes. Phys Chem Chem Phys. 2017;19:340346. 76. Epstein S, Kirkpatrick CL, Castillon GA, et al. Activation of the unfolded protein response pathway causes ceramide accumulation in yeast and INS1E insulinoma cells. J Lipid Res. 2012;53:412420. 77. Gerstle R, Desai R, Veatch S. Dithiothreitol raises transition temperatures in giant plasma membrane vesicles. Biophys J. 2017;112:519a. 78. JiménezRojo N, Leonetti MD, Zoni V, et al. Conserved functions of ether lipids and sphingolipids in the early secretory pathway. Curr Biol. 2020;30:37753787. 79. Meer GV, Wolthoorn J, Degroote S. The fate and function of glycosphingolipid glucosylceramide. Philos T R Soc. 2003;B.358:869873. 80. Contreras FX, Ernst AM, Haberkant P, et al. Molecular recognition of a single sphingolipid species by a protein’s transmembrane domain. Nature. 2012;481:525529. SUPPORTING INFORMATION Additional Supporting Information may be found online in the Supporting Information section. How to cite this article: Monasterio BG, JiménezRojo N, GarcíaArribas AB, Riezman H, Goñi FM, Alonso A. CHO/LYB cell growth under limiting sphingolipid supply: Correlation between lipid composition and biophysical properties of sphingolipidrestricted cell membranes. The FASEB Journal. 2021;35:e21657. https://doi.org/10.1096/ fj.20200 1879RR