Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Sphingolipids Are Required for Efficient Triacylglycerol Loss in Conjugated Linoleic Acid Treated Adipocytes

  • Wei Wang,

    Affiliation Department of Animal Science, University of Nebraska, Lincoln, Nebraska, United States of America

  • Michael Fromm

    mfromm2@unl.edu

    Affiliation Center for Biotechnology, University of Nebraska, Lincoln, Nebraska, United States of America

Sphingolipids Are Required for Efficient Triacylglycerol Loss in Conjugated Linoleic Acid Treated Adipocytes

  • Wei Wang, 
  • Michael Fromm
PLOS
x

Abstract

Conjugated linoleic acid (CLA) reduces adiposity in human and mouse adipocytes. This outcome is achieved through a variety of biological responses including increased energy expenditure and fatty acid oxidation, increased inflammation, repression of fatty acid biosynthesis, attenuated glucose transport, and apoptosis. In the current study, profiling of 261 metabolites was conducted to gain new insights into the biological pathways responding to CLA in 3T3-L1 adipocytes. Sphinganine and sphingosine levels were observed to be highly elevated in CLA treated adipocytes. Exogenous chemicals that increased endogenous ceramide levels decreased lipid levels in adipocytes, and activated AMP-activated protein kinase (AMPK) as well as NF-κB, both of which are typically activated in CLA treated adipocytes. Concurrent inhibition of ceramide de novo biosynthesis and recycling from existing sphingolipid pools attenuated the lipid lowering effect normally associated with responses to CLA, implicating ceramides as an important component of the lipid lowering response in CLA treated adipocytes.

Introduction

The trans-10, cis-12 isomer of CLA (t10c12 CLA) is capable of reducing adiposity in human and mouse adipocytes [15]. Molecular responses to t10c12 CLA include activation of AMP-activated protein kinase (AMPK) [6], sirtuin 1 [7], and attenuation of peroxisome proliferator-activated receptor γ (PPARγ) protein levels [3]. Treatment with t10c12 CLA requires nuclear factor-kappa B (NF-κB) for an inflammatory response [3,4,811] that includes increased prostaglandin biosynthesis in human adipocytes [12], in mouse white adipose tissue [13], and in 3T3-L1 adipocytes [14]. Despite this progress in understanding the pathways involved in the early perception of t10c12 CLA and the complex regulation of the subsequent response, much remains unknown in this process.

Sphingolipids are a complex class of lipids containing the sphingoid backbones of sphinganine (dihydrosphingosine) and sphingosine. Attaching sphinganine or sphingosine to a fatty acid via an amide linkage forms ceramide, the simplest sphingolipid that is the starting point for more complex sphingomyelins, cerebrosides, and gangliosides [15]. Sphingolipids have structural roles in lipid membranes and can serve as sources of signaling molecules as well. Signaling pathways affected by sphingolipids or molecules derived from them, include apoptosis, inflammation, mitochondrial function, and general cell metabolism [16,17].

The role of sphingolipids in adipocytes and obesity is still emerging. A study of genetically obese mice (ob/ob) and high-fat diet-induced obese mice found inhibiting ceramide de novo synthesis attenuated obesity symptoms, facilitating weight reduction, better energy metabolism, and improved insulin sensitivity [18]. In 3T3-L1 adipocytes, ceramide levels decreased as pre-adipocytes differentiated into adipocytes, and increased when adipocytes were treated with epigallocatechin gallate, which reduced adipocyte fat content [19].

In this study we performed metabolic profiling of t10c12 CLA treated adipocytes to gain insight into responding pathways. Our analysis of adipocytes after 12 h of exposure to t10c12 CLA indicated that in addition to the expected changes in lipid pathways, there were significantly increased sphinganine and sphingosine levels in the ceramide pathway. This study focused on the ceramide pathway due to its important roles in many aspects of cell biology and cell signaling [1517]. Ceramide levels were experimentally perturbed to determine ceramide’s functional contribution to lowering triacylglycerol in t10c12 CLA treated adipocytes.

Materials and Methods

Reagents

Insulin (I1882), isobutyl-methylxanthine (I5879), dexamethasone (D4902), bovine serum (A2153), calf serum (12138C), N-acetyl-D-sphingosine (C2 ceramide, A7191), monoclonal anti-ceramide antibody (C8104), myriocin (M1177), fumonisin B1(F1147) and SKI-II (S5696) were from Sigma (St. Louis, MO). AMPK inhibitor compound C (17126) was from Calbiochem (San Diego, CA). Antibodies to β-actin (SC-30656), NF-κB p65 (sc-372), sphingosine-1-phosphate phosphatase 1(Sppase, sc-55306) and serine palmitoyltransferase (Sptlc, sc-136076) were from Santa Cruz Biotechnology (Santa Cruz, CA). Antibodies to phospho-AMPKα pThr172 (2531) and AMPK (2532) were purchased from Cell Signaling (Beverly, MA). T7 RNA polymerase (P2075), RNase-Free DNaseI (M6101) and rNTPs (E6000) were purchased from Promega (Madison, WI). DharmaFECT Duo transfection reagent (T2010) was from Dharmacon (Thermo Fisher Scientific, Boulder, CO). Chemicals were dissolved in DMSO, with the exception that C2-ceramide was dissolved in ethanol, and fumonisin B1 was dissolved in water. The lowest effective chemical concentrations were chosen by testing a range of initial concentration and were as follows: 20 μmol/L fumonisin B1, 50 μmol/L myriocin, 10 μg/mL SKI-II, 30 μmol/L C2-ceramide, and were added 1 h before adding fatty acids. Chemical stock solutions were added directly to media at ≤0.2% of the final volume of the media.

3T3-L1 cell culture and differentiation

3T3-L1 fibroblasts [20] were cultured in Dulbecco's modified Eagle's medium (DMEM; Invitrogen, Carlsbad, CA) containing 10% bovine calf serum (Fisher, Pittsburgh, PA). Cells were seeded at a concentration of 1 x 104 cells/cm2 and differentiated as described [7], reaching a density of ~4.5 x 104 cells/cm2. The percentage of adipocytes was predominantly 90% of the cells after differentiation, although higher passage cells occasionally produced only 60% of the cells as adipocytes after differentiation. Our earlier whole genome microarray analyses of non-differentiated pre-adipocytes indicated they demonstrate very small responses to t10c12 CLA [14], suggesting the majority of the cellular responses to t10c12 CLA in a mixed population after differentiation are from adipocytes.

Fatty acid preparation

Fatty acids (>99%, Nu-check Prep, Elysian, MN), either linoleic acid (LA) or trans-10, cis-12 CLA, were dissolved in 0.1 M KOH, diluted into fatty acid free (>99%) bovine serum albumin (BSA) in phosphate buffered saline at a 1:1 ratio (2 mmol/L BSA: 2 mmol/L fatty acid), pH adjusted to 7.4, and added to the cultures containing 4 to 6 d post-differentiated 3T3-L1 adipocytes [7]. All media contained 100,000 U/L penicillin and 172 μmol/L streptomycin (Invitrogen, Carlsbad, CA). 50 μM t10c12 CLA was used if assaying chemicals that increased triacylglycerol loss as this facilitated observing triacylglycerol loss, but otherwise 100 μM t10c12 CLA was generally used.

Metabolic profiling

Metabolite profiling of 3T3-L1 adipocytes was performed by Metabolon (Durham, NC). Six replicates of 3T3-L1 adipocytes, treated with 100 μM LA or t10c12 CLA for 0.5 h or 12 h, were collected, and frozen at -80°C for later analysis. Samples were solvent extracted and analyzed by GC/MS and LC/MS or LC/MS/MS platforms for 261 metabolites (S1 Dataset). Compounds were identified by known standards or prior characterizations of metabolite libraries. Welch’s two–sample t-tests were used to identify biochemicals that differed significantly between LA and t10c12 CLA treatments.

RNA interference

Four to five days post differentiation 3T3-L1 adipocytes were transfected by DharmaFECT Duo transfection reagent as described [21]. The final transfection reagent concentration was 1.4 μl/cm2. DharmaFECT Duo and 100 nmol/L shRNA were added 24h before adding fatty acids. For drug and shRNA combination assays, drugs were added to media 1h before fatty acid treatments.

shRNA preparation

Initial sequence against targets were obtained from literature [22,23] or online tools (http://sirna.wi.mit.edu/, http://www.thermoscientificbio.com/design-center/?redirect=true). Five potential sequences for each target were chosen and prepared as described to determine the most effective ones for shRNA knockdowns [24]. In brief, shRNA sequences were synthesized using the T7 polymerase (Promega, Madison, WI) templates and purified by 12–20% denaturing polyacrylamide gel. The shRNA sequences used were: scramble (sh-Non): 5′-AAC AGU CGC GUU UGC GAC UGG UCU CUU GAA CCA GUC GCA AAC GCG ACU GCC UAU AGU GAG UCG UAU UA-3′; ShRNA-sphingosine-1-phosphate phosphatase 1 (sh-S1PP): 5′-AAU CAU CAA GCU GGA GGU CUU CUU CUC UUG AAA GAA GAC CUC CAG CUU GAU GAC CUA UAG UGA GUC GUA UUA-3′; or ShRNA-serine palmitoyltransferase long chain subunit 1 (sh-SPTLC): 5′-AAG CCA UCA UUU ACU CGU AUG UCU CUU GCA CAU ACG AGU AAA UGA UGG CCC UAU AGU GAG UCG UAU UA-3′.

Western blot

Nuclear and cytosolic extracts were isolated using a nuclear extract kit (Active Motif, Carlsbad, CA). Equal amounts of proteins were separated by SDS-PAGE, transferred to Immun-blot PVDF membrane (Bio-Rad Laboratories, Hercules, CA), probed with the indicated primary antibodies, and detected with secondary antibodies. Enhanced chemiluminescence (Pierce, Rockford, IL) was used for detection. Band intensities were determined from digital images from exposures in the linear range using software (Quantity One, Biorad, Hercules, CA). All western blot analyses were repeated at least three times.

Ceramide assay

Immunodot blots for ceramide levels were performed as described with modifications [2527]. Equal amount of cells (~4.5x105) in 6-well plates from each treatment were collected by spinning down at 100 x g for 5 min. Cells were resuspended in 500 μl ice-cold acidic methanol (acetic acid: methanol 1:50). Lipids were extracted by addition of 500 μl chloroform and 500 μl H2O. The organic lower-phase was collected and evaporated with nitrogen gas. After adding 100 μl of chloroform/methanol (2:1, v/v) to each tube, 2 μl of each sample was spotted on PVDF membrane and allowed to dry for 30 min. The PVDF membrane was blocked with 5% dry milk in PBS at room temperature for 1 hour, and standard immunoblot procedures followed to detect ceramide levels using a monoclonal antibody that detects ceramides (Sigma C8104 clone MID 15B4, Saint Louis, MO, USA).

Quantification of cellular triacylglycerol amounts

Cells (~4.5x105) were washed twice with 1 ml PBS, collected in 1 ml PBS by scraping with a rubber policeman spatula, dispersed by pipetting, and aliquoted for triacylglycerol (0.2 ml), DNA (0.1 ml), and immunoblot assays (0.7 ml). A 0.2 ml aliquot of cells was centrifuged at 200 x g for 5 min, the cell pellet was re-suspended in 0.2 ml Triglyceride Working Reagent (TR0100; Sigma, St. Louis, MO) and sonicated for 6s to solubilize the cells and oil bodies. The resulting solution was incubated at 37°C for a 5 min reaction and then cooled. The absorbance at 540 nm was measured and the triacylglycerol level was determined against a standard curve. Values were normalized to the cellular DNA content for the corresponding well. Triacylglycerol data are expressed as nmol of triacylglycerol per μg of DNA. A comparison of this sonication method with a method using isopropanol-hexane extraction [28] prior to the triacylglycerol assay indicated the two methods gave similar linear responses (S1 Fig).

Quantification of DNA content

The DNA assay was performed using Hoechst 33342 (Sigma B2261) and a standard protocol (http://www.piercenet.com/instructions/2162245.pdf) with the following modifications. Cells were harvested and sonicated as described above in the triacylglycerol assay. A 10μl aliquot of the sonicated lysate was added to 200μl of a 1 μg/ml Hoechst 33342 working solution. The mixed solution was incubated at room temperature in the dark for 10 min and then a 350 nm excitation was used for measuring the fluorescence emission at 460 nm. A standard curve of diluted DNA (Sigma D1626) was used to determine the DNA concentration of the samples.

Statistical Analysis

One or two-way ANOVA was used to analyze the data. Posthoc pairwise comparisons were calculated using Tukey’s test and were considered significant for p ≤ 0.05. All analyses were performed using SAS software (SAS, Cary, NC).

Results

Metabolite profiling of t10c12 CLA treated 3T3-L1 adipocytes

A metabolic profiling analysis of 3T3-L1 adipocytes treated with either 100 μM LA or t10c12 CLA for 0.5 h or 12 h was conducted for the purpose of identifying early changes in metabolites in these cells. A total of 261 metabolites were surveyed in this analysis (S1 Dataset). At 0.5 h, the main change was the large increase in t10c12 CLA relative to the LA controls, indicating rapid uptake of t10c12 CLA by the adipocytes. At 12 h, there were changes in various lipid levels, consistent with the ability of t10c12 CLA to perturb lipid metabolism (S1 Dataset). The most striking change and focus of the present study was a 9.8 fold increase in sphinganine and a 3.6 fold increase in sphingosine (Table 1). This indicates the sphingolipid pathway is highly elevated 12 h after adding t10c12 CLA. Sphinganine is produced early in the biosynthetic pathway for ceramides, while sphingosine is derived from ceramide or from recycling of sphingolipids (Fig 1). Ceramide levels were not measured in this metabolite profiling analysis, and were measured separately as described below.

thumbnail
Fig 1. Biosynthetic pathway of ceramide and sphingolipids.

Entry into the pathway occurs by biosynthesis via serine palmitoyltransferase or dephosphorylation of sphingosine-1-phosphate by sphingosine-1-phosphate phosphatase. The enzyme targets of myriocin, fumonisin B1, and sphingosine kinase inhibitor SKI-II are also shown as are the targets of shRNA used in this study.

https://doi.org/10.1371/journal.pone.0119005.g001

thumbnail
Table 1. Relative sphingolipid levels in 3T3-L1 adipocytes after t10c12 CLA treatment.

https://doi.org/10.1371/journal.pone.0119005.t001

Ceramide levels increase in t10c12 CLA treated adipocytes and cause triacylglycerol reductions

An immunoblot analysis of ceramide levels confirmed that they are elevated in adipocytes treated with 50 μM t10c12 CLA, and are present at higher levels if treated with 100 μM t10c12 CLA (Fig 2A). Thus, t10c12 CLA strongly increases ceramide/sphingolipid pools in adipocytes. To test the hypothesis that ceramides and/or sphingolipids are functionally involved in t10c12 CLA-mediated triacylglycerol loss in adipocytes, we tested two chemicals that enhance this pathway. C2 ceramide is a short chain ceramide that can mimic some ceramide responses [29]. SKI-II (sphingosine kinase inhibitor 2) is a chemical that inhibits sphingosine kinase 1, which normally depletes the sphingosine pool by producing sphingosine-1-phosphate (S1P, Fig 1). Addition of 30 μM C2 ceramide to adipocytes increased ceramide levels (Fig 2A). As the immunological assay used does not detect C2 ceramide (S2 Fig), this result indicates exogenous C2 ceramide increases endogenous ceramide levels). Thirty μM C2 ceramide resulted in strong reductions in triacylglycerol levels in the presence of LA, without further reductions in triacylglycerol levels in the presence of 50 μM t10c12 CLA (Fig 2B). Note this study uses 50 μM t10c12 CLA treatments when looking for effects that lower triacylglycerol levels. This is because 100 μM t10c12 CLA depletes triacylglycerol levels, making further reductions difficult to measure. Conversely, we use 100 μM t10c12 CLA when analyzing treatments that increase triacylglycerol levels, as having a lower triacylglycerol level as a reference point provides a larger range of measurable triacylglycerol increases.

thumbnail
Fig 2. Ceramide and triacylglycerol levels in treated 3T3-L1 adipocytes.

(A) Ceramide levels were measured with a monoclonal antibody assay in extracts from adipocytes 24 h after addition of 50 μM (gray bars) or 100 μM (white bars) LA or t10c12 CLA (CLA) with or without 30μM C2 ceramide (C2) or 10μg/ml of an inhibitor of sphingosine kinase (SKI-II), which is expected to increase ceramide levels. (B) Triacylglycerol (TG) levels 24 h after 3T3-L1 adipocytes were treated with 50 μM LA or t10c12 CLA (CLA), with or without 30 μM C2 ceramide or 10 μg/ml SKI-II. The control sample contained 0.2% DMSO, which produced the same values as a control sample containing 0.2% ethanol in separate experiments (data not shown). (C) Triacylglycerol levels 24 h after 3T3-L1 adipocytes were treated 100 μM LA or t10c12 CLA (CLA), with or without 20 μM fumonisin B1 (FB1) or 50 μM myriocin (Myri.). (D) Ceramide levels were detected and quantitated as in A for adipocytes treated for 24 h with 100 μM LA or t10c12 CLA, with or without 20 μM FB1 or 50 μM myriocin. Each bar represents the mean ± SEM (n = 3), and a representative experiment of at least three independent experiments is shown. Means not sharing a common letter differ, p≤0.05.

https://doi.org/10.1371/journal.pone.0119005.g002

Addition of SKI-II, which is expected to increase ceramide and sphingosine levels by inhibiting production of S1P from sphingosine (Fig 1), gives a similar result: ceramide levels increase in the presence of SKI-II and LA or t10c12 CLA (Fig 2A), and there is a strong loss of triacylglycerol levels adipocytes with these treatments (Fig 2B). The results that either C2 ceramide or SKI-II can mediate triacylglycerol loss in the absence of t10c12 CLA (Fig 2B) supports the premise that increased ceramide/sphingolipid levels are functionally involved in the response to t10c12 CLA. Surprisingly, there is not an additive effect on lowering triacylglycerol levels when these chemicals are combined with 50 μM t10c12 CLA.

A remaining question is whether ceramide/sphingolipid pathways are required for triacylglycerol loss in t10c12 CLA treated adipocytes. The combination of de novo biosynthesis and bioconversions between existing sphingolipid pools makes this a difficult question to answer. For example, adding fumonisin B1 or myriocin, both inhibitors of ceramide de novo biosynthesis (Fig 1), had little effect on the triacylglycerol levels in the presence of 100 μM t10c12 CLA (Fig 2C). Analysis of ceramide levels in adipocytes treated with 100 μM t10c12 CLA and these chemicals indicates that ceramide levels remained high in these cells despite the presence of these inhibitors (Fig 2D). This result suggests the ceramides are likely to be derived from both de novo biosynthesis and established pools within these adipocytes. Importantly, for interpreting subsequent experiments, these results indicate that these inhibitors did not have measurable cell toxicity or effects on triacylglycerol levels.

To gain more experimental control over inhibiting ceramide production, we developed short hairpin RNA (shRNA) knockdowns for serine palmitoyltransferase long chain subunit 1 (SPTLC) and S1P phosphatase (S1PP). Serine palmitoyltransferase (SPT1) catalyzes the initial step in ceramide biosynthesis and S1PP produces sphingosine from existing pools of S1P (Fig 1). Efficient shRNA knockdown of SPTLC protein levels was achieved as measured by immunoblot analysis (Table 2), but this knockdown increased triacylglycerol levels in t10c12 CLA treated adipocytes (Table 3), without decreasing ceramide levels (Table 3), making interpretation difficult. ShRNA knockdown of S1PP was also efficient as measured by immunoblot analysis (Table 2), and did not affect triacylglycerol levels in t10c12 CLA treated adipocytes (Table 3). Ceramide levels were still quite high when measured for the S1PP shRNA knockdowns in t10c12 CLA treated adipocytes (Table 3). Thus, shRNA knockdown of either of the SPTLC or S1PP enzymes was not sufficient to diminish ceramide levels.

We next investigated whether ceramide levels could be reduced by concurrently inhibiting biosynthetic and S1P entry points into the pathway. A combination of myriocin, a SPT inhibitor, and shRNA against S1PP, is expected to inhibit both de novo biosynthesis of ceramides and their production from S1P (Fig 1). This combination was sufficient to strongly diminish ceramide levels in 100 μM t10c12 CLA treated adipocytes (Fig 3A). Further, this combination had higher triacylglycerol levels in the presence of 100 μM t10c12 CLA, relative to 100 μM t10c12 CLA alone, indicating it was effective in attenuating the ability of t10c12 CLA to cause triacylglycerol loss (Fig 3B).

thumbnail
Fig 3. Inhibition of both biosynthetic entry points into ceramides affects triacylglycerol and ceramide levels.

(A) Immunoblot analysis of ceramide levels 24 h after 3T3-L1 adipocytes were treated with 100 μM LA or t10c12 CLA in the presence of 100 nM shRNA against non-target (sh-Non) or S1PP (sh-S1PP), with or without 20 μM Fumonisin B1 (FB1) or 50 μM Myriocin (Myri). (B) Triacylglycerol (TG) levels after 3T3-L1 adipocytes were treated as in A. A representative experiment of at least three independent experiments is shown for each panel. Each bar represents the mean ± SEM (n = 3). Means not sharing a common letter differ, p≤0.05.

https://doi.org/10.1371/journal.pone.0119005.g003

A similar analysis was performed using a combination of shRNA against S1PP and fumonisin B1. Fumonisin B1 is an inhibitor of sphinganine N-acyltransferase (Fig 1), and thereby should inhibit ceramide biosynthesis. In combination with a shRNA knockdown S1PP, this should block both biosynthesis of ceramide and its generation from S1P (Fig 1). This combination was sufficient to strongly diminish ceramide levels in 100 μM t10c12 CLA treated adipocytes (Fig 3A). These fumonisin B1 and S1PP shRNA/(LA or t10c12 CLA) combination treatments had a higher ratio of triacylglycerolCLA/ triacylglycerolLA than the control lacking fumonisin (Fig 3B), in agreement with the similar myriocin experiment described above (Fig 3B). However, there was a tendency for the fumonisin B1 and S1PP shRNA/LA combination treatment to have lower triacylglycerol levels than the non-specific shRNA and LA combination control lacking fumonisin (Fig 3B). As this suggests some possible off-target effects with fumonisin B1, our analyses rely primarily on the results with myriocin and S1PP shRNA to support the hypothesis that ceramides mediate much of the triacylglycerol lowering effects in t10c12 CLA treated adipocytes.

Ceramide signaling is downstream of AMPK and can increase AMPK activity

AMPK is a key component of responses in t10c12 CLA treated adipocytes, as its inhibition by compound C abolishes the effects of t10c12 CLA [6]. As demonstrated above (Fig 2B), either C2 or SKI-II is sufficient to mediate triacylglycerol loss in adipocytes in the absence of t10c12 CLA, raising the question of whether ceramide-mediated triacylglycerol loss requires activated AMPK. To answer this, adipocytes were treated with C2 or SKI-II in the presence or absence of 10 μM of AMPK inhibitor compound C (Table 4). For comparison, the ability of compound C to interfere with triacylglycerol loss in the presence of 100 μM t10c12 CLA is also shown. A considerable reduction in triacylglycerol levels occurs in the presence of C2 or SKI-II, but this response is not affected by the presence of compound C (Table 4). Therefore, we conclude that C2 and SKI-II act downstream or independently of AMPK to mediate triacylglycerol loss in adipocytes.

thumbnail
Table 4. Triacylglycerol levels in 3T3-L1 adipocytes treated with C2-ceramide, SKI-II, or t10c12 CLA, with or without compound C.

https://doi.org/10.1371/journal.pone.0119005.t004

We also measured the activity of AMPK in the presence of these compounds. The active form of AMPK can be detected with an antibody specific for its phosphorylated active state at AMPKα pThr172 [30]. 50 μM t10c12 CLA activates p-AMPK, relative to p-AMPK levels in LA treated adipocytes (Table 5). Surprisingly, both C2 and SKI-II activated AMPK in 50 μM LA or t10c12 CLA to equivalent p-AMPK levels (Table 5). Thus, although AMPK is not required for triacylglycerol loss in C2 or SKI-II treated adipocytes as indicated by the similar losses with or without the presence of compound C, these compounds activate AMPK.

C2 and SKI-II activate NF-κB independently of AMPK

Activation of NF-κB has an important functional role in mediating triacylglycerol loss in t10c12 CLA treated adipocytes [3,14]. As an additional comparison between the response of adipocytes to t10c12 CLA and their response to C2 or SKI-II, we examined the amount of active NF-κB present in the nucleus. The ability of 100 μM t10c12 CLA treatment to activate NF-κB is partially dependent on AMPK as compound C attenuates this response (Table 6). Treatment with either C2 or SKI-II activates NF-κB in adipocytes in the presence of either LA or t10c12 CLA (Table 7). NF-κB activation occurs equally in C2 or SKI-II treated adipocytes whether compound C is present or absent (Table 8). These results indicate that C2 or SKI-II act downstream or independently of AMPK to activate NF-κB.

thumbnail
Table 6. Relative NF-κB nuclear levels in adipocytes treated with LA or t10c12 CLA, with or without Compound C.

https://doi.org/10.1371/journal.pone.0119005.t006

thumbnail
Table 7. Relative NF-κB nuclear levels in adipocytes treated with LA or t10c12 CLA, with or without C2-Ceramide or SKI-II.

https://doi.org/10.1371/journal.pone.0119005.t007

thumbnail
Table 8. Relative NF-κB nuclear levels in adipocytes treated with C2-Ceramide or SKI-II, with or without Compound C.

https://doi.org/10.1371/journal.pone.0119005.t008

Loss of cell adherence in t10c12 CLA treated adipocytes is a ceramide-dependent response

Measurement of the amounts of cellular DNA per well of treated adipocytes supports a role for t10c12 CLA in causing a loss of cell adherence. T10c12 CLA treated adipocytes have less total DNA (Table 6) and more non-attached cells when examined microscopically (data not shown). These detached cells are removed when removing the media and washing the adherent cells prior to harvesting the adherent cells for analysis. Addition of C2 ceramide or SKI-II to LA treated adipocytes caused cell loss similar to those treated with 100 μM t10c12 CLA as indicated by the lower amounts of total cellular DNA (Table 9). Combinations of C2 ceramide or SKI-II with 50 μM LA produced about the same amount of cell loss as treatment with 100 μM t10c12 CLA. Combining either of these drugs with 50 μM t10c12 CLA did not significantly increase the amount of cell loss (Table 9). Inhibition of the ceramide pool levels in t10c12 CLA treated adipocytes by concurrent inhibition of ceramide biosynthesis and entry from S1P by combinations of myriocin or FB1 and knockdown of SIPP by shRNA attenuated cell loss (Table 10). These results suggest that t10c12 CLA-mediated loss of cell adherence in adipocytes is mediated in part or predominantly by ceramides.

thumbnail
Table 9. DNA amounts isolated from adherent cells after various chemical treatments.

https://doi.org/10.1371/journal.pone.0119005.t009

thumbnail
Table 10. DNA amounts isolated from adherent cells after adding LA or t10c12 CLA and various shRNAs.

https://doi.org/10.1371/journal.pone.0119005.t010

Discussion

Our metabolite profiling analysis of adipocytes treated with t10c12 CLA indicated there were increased sphinganine and sphingosine levels in the ceramide pathway after 12 h, but not at the earlier 0.5 h time point. The metabolite profiling did not analyze ceramides, but our antibody-based detection method indicated ceramide levels were increased as well. The experiments blocking both entry points into the ceramide pathway attenuated ceramide levels as measured by the monoclonal antibody assay. As single block experiments did not block ceramide levels according to this assay and thereby serve as controls for the double block experiments, these results indicate the monoclonal antibody against ceramides demonstrates preferential specificity for ceramides over sphingomyelin, cholesterol or other phospholipids. The agreement between the mass spectrometry measurements of increased sphinganine and sphingosine levels and the monoclonal assays of ceramides strongly supports the conclusion that t10c12 CLA treatments increase sphingolipid levels in adipocytes.

An earlier report on the effects of t10c12 CLA in adipocytes observed that the rate of ceramide biosynthesis was attenuated when measured from radiolabeled acetate or pyruvate [31]. This is consistent with the reduced rate of de novo fatty acid biosynthesis observed in t10c12 CLA treated adipocytes [7,31]. However, the abundant pools of pre-existing fatty acids and sphingolipids in adipocytes should allow for ceramide synthesis and salvage from existing molecules even when de novo fatty acid biosynthesis from two-carbon units is attenuated. Therefore, as total ceramide levels were not directly measured in this earlier report, it does not contradict our findings of increased ceramide levels in t10c12 CLA treated adipocytes.

Our inhibitor studies indicated inhibiting ceramide levels required blocking both the de novo ‘sphinganine’ entry point as well as the salvage ‘sphingosine’ entry point (see Fig 1). As both the sphinganine and sphingosine levels were elevated in vivo, this suggests increased flux from the unidirectional sphinganine to ceramide steps. Alternatively, ceramide levels could be increased via both of the sphinganine and sphingosine entry points. Our observation that we needed to block both of these entry points in our double block experiments (myriocin and S1PP shRNA) suggests both entry points are active in producing ceramides in adipocytes. This requirement to block both entry points into ceramide production to block the effects of t10c12 CLA suggests increased flux through the ceramide pathway to produce the more complex sphingomyelins, cerebrosides, and/or gangliosides is not functionally important as a single block of this biosynthetic pathway should have been sufficient for blocking their production.

Effectively blocking ceramide levels attenuated the effects of t10c12 CLA, supporting a functional role for elevated ceramides and/or sphinganine and/or sphingosine in lowering triacylglycerol levels. A functional role is also supported by our result that adding C2 ceramide or the sphingosine kinase inhibitor SKI-II increased endogenous ceramide levels and lowered triacylglycerol levels. The combination of increased ceramide levels in cells, an attenuation of the triacylglycerol lowering response when ceramide levels are blocked, and a triacylglycerol lowering response when ceramides are increased, all support the conclusion that molecules in the ceramide pathway are functionally important in the triacylglycerol lowering response in t10c12 CLA-treated adipocytes.

We discuss some of the known responses in adipocytes treated with t10c12 CLA [5] with regards to possible roles of ceramide or ceramide-related molecules in these responses. T10c12 CLA affects a number of biological processes including: increased energy expenditure and fatty acid oxidation, repression of fatty acid biosynthesis, and activation of AMPK [6,7,31,32]; increased NF-κB and inflammation [3,14]; and apoptosis in adipocytes and adipose tissues [11,33].

Energy metabolism and ceramide activation of AMPK

Activated AMPK is required for triacylglycerol loss in t10c12 CLA-treated adipocytes [6,14]. Activated AMPK increases fatty acid oxidation and energy production while decreasing fatty acid biosynthesis in most cells [34], including t10c12 CLA-treated adipocytes [6,7]. Activation of AMPK occurs as early as 0.5 h after addition of t10c12 CLA [6]. In contrast, sphinganine and sphingosine levels are not significantly elevated at this time (Table 1), suggesting the elevation of these occurs after AMPK activation. Addition of C2 ceramide activated AMPK in adipocytes as described here or in an INS-1 pancreatic beta cell line [35]. Sphingosine, which was elevated 3.6 fold in our t10c12 CLA-treated adipocytes, can be phosphorylated by sphingosine kinase to produce S1P, which activates AMPK in hepatocytes [36]. FTY720, an analog of S1P, can stimulate AMPK activity in adipocytes and is associated with an anti-obesity effect in obese mice [37]. Activated AMPK increases fatty acid oxidation in mitochondria by a number of mechanisms [34]. Additionally, ceramides occur in mitochondria and can affect mitochondrial membrane permeability, respiratory chain function, oxidative phosphorylation, and apoptosis [17,38]. Increased mitochondrial ceramide levels may account in part for some of the increased fatty acid oxidation occurring in t10c12 CLA-treated adipocytes [6,31].

Ceramides activate NF-κB in adipocytes

Increased NF-κB activity is observed in t10c12 CLA treated adipocytes and this increased activity is necessary for the triacylglycerol loss response in adipocytes [3]. NF-κB activity is associated with an inflammatory response, including prostaglandin biosynthesis, which is also required for triacylglycerol loss in t10c12 CLA treated adipocytes [12,14]. Sphingolipid metabolites, including ceramides, sphingosine, and S1P activate NF-κB in a variety of cells including fibroblasts [39], bronchial epithelium cells [40], endothelial cells [41], mouse adipocytes [42], and intestinal cells [43]. Our results demonstrate that t10c12 CLA, C2 ceramide, or SKI-II increased ceramide/sphingolipid levels and activated NF-κB in adipocytes. Taken together with the observation of increased ceramide, sphinganine, and sphingosine levels, these data suggest that increased sphingolipid metabolites could be fully or partially responsible for NF-κB activation in the triacylglycerol loss response of adipocytes treated with t10c12 CLA.

Loss of cell adherence and apoptosis in adipocytes and adipose tissues

T10c12 CLA mediates reductions in mouse white adipose tissue cell numbers in part by apoptosis [11,33]. Ceramides stimulate apoptosis in adipocytes in vitro [44,45]. In our current studies, we did not specifically measure apoptosis, but t10c12 CLA or increased ceramide levels caused about 15% of the adipocytes to become non-adherent, presumably due to their becoming apoptotic. Concurrent inhibition of ceramide de novo biosynthesis and entry from S1P attenuated t10c12 CLA-mediated loss of cell adherence. Our results suggest the ceramide/sphingolipid pathway is a major signaling pathway in t10c12 CLA-mediated loss of cell adherence. Note that as our triacylglycerol measurements are per μg of cellular DNA, this measurement is not affected by the loss of the non-adherent cells. However, apoptosis and loss of adipocytes in mouse white adipose tissues significantly contributes to overall reductions in the mass of these tissues [11,33]. Our findings of increased sphinganine, sphingosine, and ceramide levels in t10c12 CLA treated adipocytes provide a biochemical explanation for the apoptosis and loss of adipocytes observed in white adipose tissues of mice fed t10c12 CLA [11,33].

Effect of sphingolipid metabolites in whole animal studies

The triacylglycerol lowering effects of sphingolipid metabolites in adipocytes in vitro is in agreement with a study utilizing FTY720, an analog of S1P, that lowers triacylglycerol levels in adipocytes in vitro and has an anti-obesity effect in obese mice [37]. Mice fed t10c12 CLA diets exhibit rapid loss of white adipose tissue [1,2,11]. A hypothesis that similar mechanisms occur in animals and in vitro is supported by our earlier studies demonstrating the whole genome transcriptional responses of white adipose tissue of mice fed t10c12 CLA and of 3T3-L1 adipocytes treated with t10c12 CLA are very similar when measured by microarrays [9]. If a hypothesis that the mechanisms are similar is correct, then a corollary is that sphingolipid metabolites are an important component of the fat loss response of white adipose tissue in mice fed t10c12 CLA. However, a general conclusion that treatments that increase sphingolipid metabolites cause fat loss in animals is opposed by an experiment in mice where inhibition of ceramide de novo biosynthesis with myriocin attenuated obesity symptoms, facilitating weight reduction, better energy metabolism, and improved insulin sensitivity [18]. These contradictory results from animal trials suggest a simple generalization of the effects of perturbing the complex ceramide/sphingolipid pathway in whole animal studies may not be possible. However, for in vitro cultures, where adipocytes are often 60 to 90 percent of the cells, ceramides and sphingolipid metabolites are associated with lowering triacylglycerol levels and/or apoptosis [19,37,46,47].

Our results suggest that increased sphingolipid metabolites should be added to the list of mechanisms implicated in lowering triacylglycerol in adipocytes treated with t10c12 CLA, although the mechanism by which t10c12 CLA activates sphingolipid biosynthesis remains unknown. Increased sphingolipid metabolites alter the physical properties of cell membranes where they can create lipid rafts, alter membrane fluidity, and interact with key membrane proteins to activate ceramide/sphingolipid signaling pathways [48].

Supporting Information

S1 Dataset. Metabolite levels of 261 metabolites in LA and t10c12 CLA treated adipocytes.

3T3-L1 adipocytes were treated with 100 μM LA (LA) or t10c12 CLA for 0.5 or 12 h and harvested for analysis of 261 metabolites (Materials and Methods).

https://doi.org/10.1371/journal.pone.0119005.s001

(XLS)

S1 Fig. Comparison of methods of preparing cellular triacylglycerol for enzymatic assay.

A comparison of our ‘whole-cell sonication’ method with an isopropanol-hexane extraction method was performed. The whole cell sonication method is described in the Materials and Methods section. The isopropanol-hexane extraction method was performed as described [28]. A series of different amounts of adipocytes were used in the two methods and the triacylglycerol levels measured as described in Materials and Methods. Both assays were linear over the range of cells measured (Note the range of triacylglycerol levels typically measured is close to the mid-point of the range shown). The SEM of the isopropanol-hexane extraction method is larger than the sonication method, while the linear equation for the sonication method has a smaller slope. These results indicate the sonication method is more reproducible and slightly underestimates differences in triacylglycerol levels. A representative experiment from two independent experiments is shown and each bar represents the mean ± SEM (n = 3).

https://doi.org/10.1371/journal.pone.0119005.s002

(TIF)

S2 Fig. The monoclonal antibody recognizing ceramides does not detect C2 ceramide.

10 and 20 nmol of C2 ceramide were spotted onto PVDF membrane as well as lipid soluble extracts from adipocytes treated with 50 μM or 100 μM t10c12 CLA (CLA) and analyzed by immunoblot analysis with a monoclonal antibody that detects ceramides. Complete recovery of the 60 nmol of C2 ceramide used per well at the standard 30 μM concentration would result in 1.2 nmol of C2 per spot as only 2% of the sample is spotted. The 10 and 20 nmol amounts are approximately 8 fold and 17 fold more than the maximum C2 levels possible in the C2 treated samples. Therefore, we conclude that the monoclonal antibody used in the immunoblot assay does not detect C2 ceramide at the levels added to the cells.

https://doi.org/10.1371/journal.pone.0119005.s003

(TIF)

Author Contributions

Conceived and designed the experiments: WW MF. Performed the experiments: WW. Analyzed the data: WW MF. Contributed reagents/materials/analysis tools: WW. Wrote the paper: WW MF.

References

  1. 1. Park Y, Storkson JM, Albright KJ, Liu W, Pariza MW. Evidence that the trans-10,cis-12 isomer of conjugated linoleic acid induces body composition changes in mice. Lipids. 1999;34: 235–241. pmid:10230716
  2. 2. Park Y, Albright KJ, Liu W, Storkson JM, Cook ME, Pariza MW. Effect of conjugated linoleic acid on body composition in mice. Lipids. 1997;32: 853–858. pmid:9270977
  3. 3. Chung S, Brown JM, Provo JN, Hopkins R, McIntosh MK. Conjugated linoleic acid promotes human adipocyte insulin resistance through NFkappaB-dependent cytokine production. J Biol Chem. 2005;280: 38445–38456. pmid:16155293
  4. 4. House RL, Cassady JP, Eisen EJ, McIntosh MK, Odle J. Conjugated linoleic acid evokes de-lipidation through the regulation of genes controlling lipid metabolism in adipose and liver tissue. Obes Rev. 2005;6: 247–258. pmid:16045640
  5. 5. Kennedy A, Martinez K, Schmidt S, Mandrup S, LaPoint K, McIntosh M. Antiobesity mechanisms of action of conjugated linoleic acid. J Nutr Biochem. 2010;21: 171–179. pmid:19954947
  6. 6. Jiang S, Wang Z, Riethoven JJ, Xia Y, Miner J, Fromm M. Conjugated Linoleic Acid Activates AMP-Activated Protein Kinase and Reduces Adiposity More Effectively When Used with Metformin in Mice. J Nutr. 2009;139: 2244–2251. pmid:19828681
  7. 7. Jiang S, Wang W, Miner J, Fromm M. Cross regulation of sirtuin 1, AMPK, and PPARgamma in conjugated linoleic acid treated adipocytes. PLoS One. 2012;7: e48874. pmid:23155420
  8. 8. House RL, Cassady JP, Eisen EJ, Eling TE, Collins JB, Grissom SF, et al. Functional genomic characterization of delipidation elicited by trans-10, cis-12-conjugated linoleic acid (t10c12-CLA) in a polygenic obese line of mice. Physiol Genomics. 2005;21: 351–361. pmid:15888570
  9. 9. LaRosa PC, Riethoven JJ, Chen H, Xia Y, Zhou Y, Chen M, et al. Trans-10, cis-12 conjugated linoleic acid activates the integrated stress response pathway in adipocytes. Physiol Genomics. 2007;31: 544–553. pmid:17878318
  10. 10. Poirier H, Shapiro JS, Kim RJ, Lazar MA. Nutritional Supplementation With trans-10, cis-12-Conjugated Linoleic Acid Induces Inflammation of White Adipose Tissue. Diabetes. 2006;55: 1634–1641. pmid:16731825
  11. 11. LaRosa PC, Miner J, Xia Y, Zhou Y, Kachman S, Fromm ME. Trans-10, cis-12 conjugated linoleic acid causes inflammation and delipidation of white adipose tissue in mice: a microarray and histological analysis. Physiol Genomics. 2006;27: 282–294. pmid:16868072
  12. 12. Kennedy A, Overman A, Lapoint K, Hopkins R, West T, Chuang CC, et al. Conjugated linoleic acid-mediated inflammation and insulin resistance in human adipocytes are attenuated by resveratrol. J Lipid Res. 2009;50: 225–232. pmid:18776171
  13. 13. Hargrave-Barnes KM, Azain MJ, Miner JL. Conjugated linoleic acid-induced fat loss dependence on Delta6-desaturase or cyclooxygenase. Obesity (Silver Spring). 2008;16: 2245–2252. pmid:18719641
  14. 14. Jiang S, Chen H, Wang Z, Riethoven JJ, Xia Y, Miner J, et al. Activated AMPK and prostaglandins are involved in the response to conjugated linoleic acid and are sufficient to cause lipid reductions in adipocytes. J Nutr Biochem. 2011;22: 656–664. pmid:20965713
  15. 15. Aguilera-Romero A, Gehin C, Riezman H. Sphingolipid homeostasis in the web of metabolic routes. Biochim Biophys Acta. 2013.
  16. 16. Hla T, Dannenberg AJ. Sphingolipid signaling in metabolic disorders. Cell Metab. 2012;16: 420–434. pmid:22982021
  17. 17. Kogot-Levin A, Saada A. Ceramide and the mitochondrial respiratory chain. Biochimie. 2013.
  18. 18. Yang G, Badeanlou L, Bielawski J, Roberts AJ, Hannun YA, Samad F. Central role of ceramide biosynthesis in body weight regulation, energy metabolism, and the metabolic syndrome. Am J Physiol Endocrinol Metab. 2009;297: E211–224. pmid:19435851
  19. 19. Choi KM, Lee YS, Choi MH, Sin DM, Lee S, Ji SY, et al. Inverse relationship between adipocyte differentiation and ceramide level in 3T3-L1 cells. Biol Pharm Bull. 2011;34: 912–916. pmid:21628894
  20. 20. Green H, Kehinde O. An established pre-adipose cell line and its differentiation in culture II. Factors affecting the adipose conversion. Cell. 1975;5: 19–27. pmid:165899
  21. 21. Kilroy G, Burk DH, Floyd E. High Efficiency Lipid-Based siRNA Transfection of Adipocytes in Suspension. PLoS ONE. 2009;4: e6940. pmid:19759827
  22. 22. Jiang ZY, Zhou QL, Coleman KA, Chouinard M, Boese Q, Czech MP. Insulin signaling through Akt/protein kinase B analyzed by small interfering RNA-mediated gene silencing. Proc Natl Acad Sci U S A. 2003;100: 7569–7574. pmid:12808134
  23. 23. Medler TR, Petrusca DN, Lee PJ, Hubbard WC, Berdyshev EV, Skirball J, et al. Apoptotic sphingolipid signaling by ceramides in lung endothelial cells. Am J Respir Cell Mol Biol. 2008;38: 639–646. pmid:18192502
  24. 24. Gondai T, Yamaguchi K, Miyano-Kurosaki N, Habu Y, Takaku H. Short-hairpin RNAs synthesized by T7 phage polymerase do not induce interferon. Nucleic Acids Res. 2008;36: e18. pmid:18208841
  25. 25. Krishnamurthy K, Dasgupta S, Bieberich E. Development and characterization of a novel anti-ceramide antibody. J Lipid Res. 2007;48: 968–975. pmid:17210985
  26. 26. Viehaber G, Brade L, Lindner B, Pfeiffer S, Wepf R, Hintze U, et al. Mouse anti-ceramide antiserum——a specific tool for the detection of endogenous ceramide. Glycobiology. 2001;11: 451–457. pmid:11445550
  27. 27. Gupta V, Patwardhan GA, Zhang QJ, Cabot MC, Jazwinski SM, Liu YY. Direct quantitative determination of ceramide glycosylation in vivo: a new approach to evaluate cellular enzyme activity of glucosylceramide synthase. J Lipid Res. 2010;51: 866–874. pmid:19826105
  28. 28. Schwartz DM, Wolins NE. A simple and rapid method to assay triacylglycerol in cells and tissues. J Lipid Res. 2007;48: 2514–2520. pmid:17717377
  29. 29. Obeid LM, Linardic CM, Karolak LA, Hannun YA. Programmed cell death induced by ceramide. Science. 1993;259: 1769–1771. pmid:8456305
  30. 30. Shaw RJ, Kosmatka M, Bardeesy N, Hurley RL, Witters LA, DePinho RA, et al. The tumor suppressor LKB1 kinase directly activates AMP-activated kinase and regulates apoptosis in response to energy stress. Proc Natl Acad Sci U S A. 2004;101: 3329–3335. pmid:14985505
  31. 31. Obsen T, Faergeman NJ, Chung S, Martinez K, Gobern S, Loreau O, et al. Trans-10, cis-12 conjugated linoleic acid decreases de novo lipid synthesis in human adipocytes. J Nutr Biochem. 2012;23: 580–590. pmid:21775116
  32. 32. Terpstra AH, Beynen AC, Everts H, Kocsis S, Katan MB, Zock PL. The decrease in body fat in mice fed conjugated linoleic acid is due to increases in energy expenditure and energy loss in the excreta. J Nutr. 2002;132: 940–945. pmid:11983818
  33. 33. Tsuboyama-Kasaoka N, Takahashi M, Tanemura K, Kim HJ, Tange T, Okuyama H, et al. Conjugated linoleic acid supplementation reduces adipose tissue by apoptosis and develops lipodystrophy in mice. Diabetes. 2000;49: 1534–1542. pmid:10969838
  34. 34. Hardie DG, Ross FA, Hawley SA. AMPK: a nutrient and energy sensor that maintains energy homeostasis. Nat Rev Mol Cell Biol. 2012;13: 251–262. pmid:22436748
  35. 35. Holland WL, Miller RA, Wang ZV, Sun K, Barth BM, Bui HH, et al. Receptor-mediated activation of ceramidase activity initiates the pleiotropic actions of adiponectin. Nat Med. 2011;17: 55–63. pmid:21186369
  36. 36. Levine YC, Li GK, Michel T. Agonist-modulated regulation of AMP-activated protein kinase (AMPK) in endothelial cells. Evidence for an AMPK-> Rac1 -> Akt-> endothelial nitric-oxide synthase pathway. J Biol Chem. 2007;282: 20351–20364. pmid:17519230
  37. 37. Moon MH, Jeong JK, Lee JH, Park YG, Lee YJ, Seol JW, et al. Antiobesity activity of a sphingosine 1-phosphate analogue FTY720 observed in adipocytes and obese mouse model. Exp Mol Med. 2012;44: 603–614. pmid:22859500
  38. 38. Siskind LJ, Kolesnick RN, Colombini M. Ceramide forms channels in mitochondrial outer membranes at physiologically relevant concentrations. Mitochondrion. 2006;6: 118–125. pmid:16713754
  39. 39. Oh E, Yun M, Kim SK, Seo G, Bae JS, Joo K, et al. Palmitate induces COX-2 expression via the sphingolipid pathway-mediated activation of NF-kappaB, p38, and ERK in human dermal fibroblasts. Arch Dermatol Res. 2014;306: 339–345. pmid:24337700
  40. 40. O'Sullivan MJ, Hirota N, Martin JG. Sphingosine 1-phosphate (S1P) induced interleukin-8 (IL-8) release is mediated by S1P receptor 2 and nuclear factor kappaB in BEAS-2B cells. PLoS One. 2014;9: e95566. pmid:24743449
  41. 41. Zhang W, An J, Jawadi H, Siow DL, Lee JF, Zhao J, et al. Sphingosine-1-phosphate receptor-2 mediated NFkappaB activation contributes to tumor necrosis factor-alpha induced VCAM-1 and ICAM-1 expression in endothelial cells. Prostaglandins Other Lipid Mediat. 2013;106: 62–71. pmid:23770055
  42. 42. Wu D, Ren Z, Pae M, Guo W, Cui X, Merrill AH, et al. Aging up-regulates expression of inflammatory mediators in mouse adipose tissue. J Immunol. 2007;179: 4829–4839. pmid:17878382
  43. 43. Pyne NJ, Pyne S. Sphingosine 1-phosphate is a missing link between chronic inflammation and colon cancer. Cancer Cell. 2013;23: 5–7. pmid:23328479
  44. 44. Kim HS, Hausman DB, Compton MM, Dean RG, Martin RJ, Hausman GJ, et al. Induction of apoptosis by all-trans-retinoic acid and C2-ceramide treatment in rat stromal-vascular cultures. Biochem Biophys Res Commun. 2000;270: 76–80. pmid:10733907
  45. 45. Papineau D, Gagnon A, Sorisky A. Apoptosis of human abdominal preadipocytes before and after differentiation into adipocytes in culture. Metabolism. 2003;52: 987–992. pmid:12898462
  46. 46. Yamashita-Sugahara Y, Tokuzawa Y, Nakachi Y, Kanesaki-Yatsuka Y, Matsumoto M, Mizuno Y, et al. Fam57b (family with sequence similarity 57, member B), a novel peroxisome proliferator-activated receptor gamma target gene that regulates adipogenesis through ceramide synthesis. J Biol Chem. 2013;288: 4522–4537. pmid:23275342
  47. 47. Xu F, Yang CC, Gomillion C, Burg KJ. Effect of ceramide on mesenchymal stem cell differentiation toward adipocytes. Appl Biochem Biotechnol. 2010;160: 197–212. pmid:19165630
  48. 48. Castro BM, Prieto M, Silva LC. Ceramide: A simple sphingolipid with unique biophysical properties. Prog Lipid Res. 2014.