Adverse physicochemical properties of tripalmitin in beta cells lead to morphological changes and lipotoxicity in vitro
- First Online:
- Cite this article as:
- Moffitt, J.H., Fielding, B.A., Evershed, R. et al. Diabetologia (2005) 48: 1819. doi:10.1007/s00125-005-1861-9
Long-term exposure of beta cells to lipids, particularly saturated fatty acids in vitro, results in cellular dysfunction and apoptosis (lipotoxicity); this could contribute to obesity-related diabetes. Our aims were to relate cell death to intracellular triglyceride concentration, composition and localisation following incubation of INS1 cells in saturated and unsaturated NEFA in high and low glucose concentrations.
Materials and methods
Insulin-producing INS1 cells were cultured (24 h; 3 and 20 mmol/l glucose) with palmitic, oleic or linoleic acids and the resulting intracellular lipids were analysed by gas chromatography and microscopy. Cell death was determined by quantitative microscopy and 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay, and glucose-stimulated insulin secretion by ELISA.
All NEFA (0.5 mmol/l, 0.5% albumin) inhibited glucose-stimulated (20 mmol/l) insulin secretion. Cytotoxicity was evident only with palmitic acid (p<0.05), in which case intracellular triglyceride consisted largely of tripalmitin in angular-shaped dilated endoplasmic reticulum. Cytotoxicity and morphological disruption were reduced by addition of unsaturated NEFA. Triglyceride content (control cells; 14.5 ng/μg protein) increased up to 10-fold following incubation in NEFA (oleic acid 153.2 ng/μg protein; p<0.05) and triglyceride and phospholipid fractions were both enriched with the specific fatty acid added to the medium (p<0.05).
In INS1 cells, palmitic acid is converted in the endoplasmic reticulum to solid tripalmitin (melting point >65°C), which could induce endoplasmic reticulum stress proteins and signal apoptosis; lipid-induced apoptosis would therefore be a consequence of the physicochemical properties of these triglycerides. Since cellular triglycerides composed of single species of fatty acid are not likely to occur in vivo, destruction of beta cells by saturated fatty acids could be predominantly an in vitro scenario.
KeywordsBeta cell ER stress INS1 Insulin Lipotoxicity Oleate Palmitate Phospholipid Triglyceride Tripalmitin
glucose-stimulated insulin secretion
diacylglycerol acyl transferase
high-temperature gas chromatography
Sustained elevated lipid concentrations result in impaired insulin action and secretion in vivo and in vitro; this has implications for obesity-related diabetes [1, 2, 3]. It is known that increased plasma NEFA concentrations reduce insulin-mediated glucose uptake and metabolism in muscles—the Randle effect —but the exact mechanism of action of NEFA on the beta cell to modulate glucose-stimulated insulin secretion (GSIS) [5, 6] is not known. A complex inter-relationship between the metabolism of glucose and lipids in beta cells has been described , and this is modified by the duration of exposure (acute or chronic) to the lipid metabolites [5, 6, 8].
A variety of mechanisms have been proposed to explain NEFA-induced inhibition of GSIS. In response to prolonged exposure of beta cells to NEFA and hyperglycaemia, fatty acids are deviated from beta-oxidation to alternative pathways, such as storage as triglycerides (TGs), possibly mediated by increased malonyl CoA . These intracellular TGs could provide an energy store, being later hydrolysed by lipases such as hormone-sensitive lipase , and undergo beta-oxidation . Hence, TG accumulation could play a role in inhibition of GSIS. The intermediates of lipid metabolism change expression of beta cell genes and proinsulin biosynthesis [12, 13], and NEFA or their metabolites have been proposed to increase uncoupling protein 2 and modulate insulin secretion through decreased production of ATP [14, 15]. NEFA have also been shown to reduce cellular viability and induce beta-cell apoptosis . NEFA, and particularly saturated NEFA such as palmitic acid [17, 18, 19], have been identified as toxic agents which activate signalling cascades that result in apoptotic beta-cell death [20, 21].
However, the hypothesis of NEFA-induced beta-cell death cannot play a major role in the pathophysiology of human diabetes. Although increased plasma NEFA concentrations and composition occur in obesity in humans , this condition is not invariably linked to the onset of type 2 diabetes; fewer than 12% of obese subjects with elevated lipid profiles have diabetes [23, 24]. Furthermore, the concept of ‘glucolipotoxity’ [25, 26] proposes that, in diabetes, the elevation of glucose and NEFA could contribute to the decrease in pancreatic islet function as a result of both beta-cell apoptotic death and decreased insulin secretion of type 2 diabetes. Our aims were to compare the effects of saturated and unsaturated NEFA in high and low glucose concentrations in INS1 cells. These effects were evaluated by measurement of GSIS, cell death and TG concentration, composition and localisation.
Materials and methods
Cell culture conditions
INS1 beta cells (rat insulinoma, passage 9–19) were cultured in RPMI 1640 (Sigma, Poole, UK), 11 mmol/l glucose, 10% fetal bovine serum (FBS), 2 mmol/l l-glutamine, 50 μmol/l β-mercaptoethanol, 10 mmol/l HEPES, 1 mmol/l sodium pyruvate, 50 U/ml penicillin and 50 μg/ml streptomycin (Gibco, Paisley, UK) in 5% CO2. COS7 cells (monkey kidney fibroblasts) were cultured in RPMI 1640, 11 mmol/l glucose, 10% FBS, 2 mmol/l l-glutamine, 50 U/ml penicillin and 50 μg/ml streptomycin (Gibco) in 5% CO2.
Preparation of fatty acids
Potassium salts of palmitic, oleic and linoleic acids (Sigma) were bound to 10% fatty acid–free BSA in RPMI to make 10 mmol/l stock solutions and were diluted for use to give a final concentration of fatty acid 0.5 mmol/l, BSA 0.5% unless otherwise stated. These fatty acid concentrations were confirmed by enzymatic (Wako, Neuss, Germany) analysis on the IL Monarch automated analyser (Instrumentation Laboratory, Warrington, UK) and composition was confirmed using gas chromatography (described below).
Cell viability: MTT assay
INS1 cells were plated into 96-well plates at 5×104 cells per well. Cells were treated with NEFA as described for light microscopic and insulin secretion experiments. After 24 h, 10 μl of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) at 5.5 mg/ml in PBS was added and the cells were incubated for 2 h. The medium was removed and the cells were lysed with 100 μl propan-2-ol. The absorbance was measured at 550 nm (Bichromatik; Labsystems, Altringham, UK).
Cell viability: quantitative light microscopy
INS1 cells were plated into eight-well Permanox slides (Gibco) and treated for 24 h with 0.1, 0.25, 0.5 and 1 mmol/l of NEFA in 3 or 20 mmol/l glucose. Adjustments to the stock solution were made to ensure a constant albumin concentration. Following incubation, cells were rinsed in PBS, incubated for a further 30 min in PBS containing 10 μmol/l calcein AM (Sigma) at 37°C and fixed in 2.5% paraformaldehyde (PFA) for 30 min. Slides were rinsed twice more in PBS and the slides mounted in Vectorshield containing propidium iodide (Vector Laboratories, Burlingame, CA, USA). Viability was quantified by counting the propidium-iodide-positive nuclei and calcein-AM-positive cells (average cell count=60 cells per field) and determining the percentage of propidium-iodide-positive nuclei in at least three fields of view per condition.
INS1 cells (5×104 per well in 12-well plates) were treated with RPMI containing 3 or 20 mmol/l glucose and 0.5 mmol/l NEFA or a mixture of oleic and palmitic acids in equal ratio, with a total concentration of 0.5 mmol/l. Control wells contained 0.5% fatty acid-free BSA in medium. After 24 h, the cells were washed in PBS and preincubated for 2 h in Krebs–Ringer HEPES (KRH) buffer containing 3 mmol/l glucose (119 mmol/l NaCl, 4.75 mmol/l KCl, 5 mmol/l NaHCO3, 1.2 mmol/l MgSO4, 1.18 mmol/l KH2PO4, 20 mmol/l HEPES, pH 7.4, 2.54 mmol/l CaCl2). Glucose responsiveness was evaluated by an initial 30-min incubation in KRH containing 3 mmol/l glucose followed by incubation for 30 min in KRH containing 20 mmol/l glucose. Samples were collected for insulin measurement using ELISA (Diagenics, Milton Keynes, UK); CV 4.5% and minimal detection 0.15 μg/l.
Triacylglycerol and phospholipid content and composition
INS1 or COS7 cells were seeded into six-well plates at 2×105 cells per well and treated for 24 h with 0.5 mmol/l specific fatty acids as above. Total lipids were extracted from cells in 2:1 chloroform:methanol and lipid fractions were separated using solid-phase extraction . The triacylglycerol and phospholipid fractions were collected and fatty acid methyl esters prepared using methanolic sulphuric acid. The fatty acid methyl esters were separated by gas chromatography using an Agilent 6890 GC equipped with a DB-Wax 30 m capillary column coated with a polyethylene glycol stationary phase (internal diameter 0.25 mm, film thickness 0.25 μm; Agilent Technologies, South Queensferry, UK). Fatty acids were separated according to fatty acyl carbon chain length and degree of saturation. Intact TG analyses were performed by high-temperature gas chromatography (HTGC) on a Hewlett Packard 5890 series II GC with a DB1-HT 15 m fused silica capillary column (internal diameter 0.32 mm, film thickness 0.1 mm; J & W Scientific, Stockport, UK). Hydrogen was the carrier gas and a flame ionisation detector was employed to monitor column eluent. HTGC analyses separated the TGs according to their carbon number. Baseline resolution, however, could not be achieved to determine the degree of unsaturation. The absolute concentrations of individual fatty acids and TGs were determined by reference to internal standards (heptadecanoic acid [17:0] for NEFA; tripentadecanoin [15:0] for triacylglycerol; phosphatidylcholine, diheptadecanoyl [17:0] for phospholipid).
INS1 and COS7 cells were incubated in eight-well Permanox or glass slides (Gibco) for 24 h in 3 or 20 mmol/l glucose with and without NEFA. Cells were washed in PBS and fixed in 2.5% PFA then stained for lipids. A stock solution of 0.5% (saturated) oil red O in propan-2-ol was freshly diluted (three parts stock solution in two parts distilled water) and filtered immediately before use. Cells were stained for 30 min at 37°C then briefly destained in 60% isopropanol and washed in water before mounting in PBS/glycerol.
A freshly prepared 1% Nile blue solution in distilled water was filtered and an aliquot was further diluted in water to 0.02%. Fixed cells were washed twice in PBS. Nile blue solution (1%) was added to the cells for 10 min at 65 or 70°C, after which the cells were washed in water and destained in 1% acetic acid (65°C) for 30 s. Cells were restained for 15 min in 0.02% Nile blue at 65 or 70°C before being washed in water and mounted in PBS/glycerol.
INS1 and COS7 cells for morphological analysis were incubated in 25 cm2 flasks for 24 h as described above. Cells were removed and gently prepared as a pellet, fixed in 2.5% glutaraldehyde (minimum 1 h) at room temperature (or for a 10-min period at 70°C), postfixed in 2% osmium, dehydrated and embedded in Spurr’s resin. Cells for immunoelectron microscopy were fixed in ice-cold 4% PFA (10 min) followed by 8% PFA at room temperature (50 min), washed in PBS, dehydrated in methanol and embedded in LR Gold resin. Ultrathin sections were cut onto nickel grids and contrast-enhanced with uranyl acetate and lead citrate. Immunolabelling for protein disulphide isomerase used methods described previously  and 15 nm protein A gold (Biocell, Cardiff, UK). Sections were examined in the Joel 1010 microscope with an accelerating voltage of 80 kV.
All results are expressed as means±SEM of three or more experiments. Statistical significance was calculated using ANOVA followed by post-hoc analysis; a p value <0.05 was considered significant.
Reduced cellular viability is more pronounced with palmitic acid than with oleic or linoleic acids
NEFA inhibits glucose stimulated insulin secretion
Triglyceride content and composition
The TG content of cells exposed to palmitic acid was increased when oleic acid was substituted in the culture medium to make mixtures of oleic and palmitic acids. This increased TG had a higher palmitate content despite the lower concentration of palmitic acid (<0.05 mmol/l) in the culture medium (Fig. 3b) compared with palmitic acid alone, suggesting that greater uptake and storage of saturated lipid is facilitated by the presence of unsaturated lipid.
Following incubation in high glucose and palmitic acid, 75% of the TG stored in INS1 cells was found by HTGC to contain 48 acyl carbon atoms, e.g. 3×16-carbon fatty acids (Fig. 3c). These TGs are likely to be mostly tripalmitin since gas-chromatographic analysis of fatty acid methyl esters, prepared after hydrolysis, revealed negligible amounts of palmitoleic (16:1 n-7) and myristic acids (14:0). In mixtures of 1:1 oleic:palmitic acid, the relative proportions of TG species were consistent with random esterification of available fatty acids; the percentages of tripalmitin (PPP, Fig. 3c) and triolein (OOO, Fig. 3c) stored in the cells were both exactly 13.8%. The proportion of dipalmitoyl oleoyl glycerol (PPO/POP) and dioleoyl palmitoyl glycerol (OOP/OPO) were 35 and 33%, respectively.
The TG content of COS7 cells was higher than that of INS1 cells (Fig. 3d). Following incubation with 0.5 mmol/l oleic or palmitic acid, intracellular TG content increased 15-fold (from 3.1 to 49.1 μg) and 10-fold (from 3.1 to 31.4 μg) respectively. In COS7 cells incubated in oleic acid, formation of triolein accounted for 73% of total TGs; when incubated in palmitic acid, tripalmitin accounted for 76% of total TGs.
Phospholipid fatty acid composition
Morphology of intracellular TG
Currently there is major interest in cell death following chronic exposure to lipids; this concept of NEFA-induced beta-cell death has been extended to the pathophysiology of type 2 diabetes, in which circulating NEFA levels are high [34, 35]. In vitro studies on islets and clonal beta cells have shown that saturated fatty acids, such as palmitic and stearic acids, are particularly cytotoxic compared with unsaturated oleic or linoleic acid , although unsaturated NEFA are not entirely without cytotoxic properties . Nevertheless, palmitic acid is the NEFA of choice for many islet and cell experiments. In agreement with these findings, cytotoxicity was greatest in the presence of palmitic acid and elevated glucose and lowest with oleic acid in INS1 cells. However, we found that inhibition of glucose-stimulated insulin secretion by chronic exposure to NEFA occurs with both saturated and unsaturated lipids, as previously described . We used new approaches to determine the relationship between inhibition of GSIS and the composition and localisation of intracellular TGs, demonstrating that, under in vitro conditions, palmitic-acid-induced lipotoxicity results from the formation of insoluble tripalmitin which accumulates in the ER of beta cells.
Our data show that incubation of INS1 cells with single specific fatty acids resulted in the accumulation of cellular TG with a distinct composition that reflected the fatty acid species. This suggested that there was no mixing of the exogenous fatty acids with the endogenous pool.
Moreover, our data indicate that cell death, which was highest with palmitic acid, was related to the formation of TG molecules incorporating palmitic acid and, in particular, tripalmitoyl glycerol (tripalmitin); these TGs are relatively solid at 37°C, which would have implications for the cellular handling of TGs. Oil red O and Nile blue staining demonstrated a unique distribution of neutral lipid in cells loaded with palmitate; lipid was present in vacuoles throughout the cytoplasm rather than in droplets (Fig. 5d,g); and these were identified by immunoelectron microscopy to be dilated ER. This suggests that, as in hepatocytes , TGs are assembled in the ER of beta cells. These vacuoles were not specific to beta cells since a similar localisation was noted in kidney COS7 cells. When heated to 70°C to allow increased fluidity of the TGs (Figs. 6h, 7h), lipid droplets were formed, as demonstrated by light and electron microscopy. These observations are consistent with non-physiological retention of tripalmitin in the endoplasmic reticulum, with severe disruption of the cellular architecture that would contribute to cell death. In contrast, in oleate-treated cells, which had much higher accumulation of lipid, largely in the form of triolein (melting temperature 5.5°C) , lipid was present in droplets situated in the cytoplasm, yet cell viability was 100% at 0.5 mmol/l NEFA and 20 mmol/l glucose. Cell viability was therefore unrelated to the amount of TG stored. The concentration of albumin has been shown to be important in NEFA experiments [18, 39]; 0.5% was selected since this had no effect in itself on insulin secretion but effectively bound the NEFA up to 0.5 mmol/l concentrations . At 1.0 mmol/l NEFA, 0.5% albumin there was an increase in cytotoxicity with all fatty acids, suggesting inadequate albumin binding.
Addition of unsaturated fatty acids protects cells from deleterious effects of palmitic acid in vitro . This is supported by our data (Fig. 1b). Cytotoxicity was reduced from 40 to 0% by reducing the palmitate fraction in the incubation medium from 100 to 66%. The incorporation of unsaturated fatty acids into otherwise palmitic-acid-rich TGs (Fig. 3b,c) would lower the molecular melting temperature and introduce more fluidity to the TG structure . It is evident from the data that a proportion of tripalmitin of 20% or less does not induce beta cell cytotoxicity, whereas if tripalmitin forms greater than 60% of the TG, more than 20% cell death occurs (Figs. 1 and 3). The appearance of small droplets in the distended ER (Fig. 7h) of cells incubated in a mixture of oleic and palmitic acids supports this hypothesis. The TGs formed in cells exposed to 1:1 mixtures of palmitic and oleic acids indicated no selectivity of fatty acid acylation. Thus, we propose that the ‘protection’ in this type of in vitro study is simply and primarily due to the physicochemical properties of the stored TGs rather than any other major pathophysiological effect. Palmitic-acid-induced IL-6 production from myocytes in vitro was similarly prevented by addition of unsaturated fatty acids and was independent of beta-oxidation . Both INS1 and COS7 cell lines showed a reduction in cell viability. Taken together, these data suggest that similar in vitro effects of palmitic acid are found in many cell types.
The exact mechanism of the cytotoxic effect of NEFA is unclear. It has been suggested that formation of ceramides , free radicals  and cytokines  and the induction of apoptotic pathways  could be involved. NEFA induce ER stress response genes (nuclear factor κB [NFκB]) in beta cells  in a similar, but not identical, way to that of cytokines (IL-1β, TNF-α) ; these effects were shown to be more pronounced in cells exposed to palmitic than oleic acid [44, 45]. TGs that are rich in saturated fatty acids remain in the ER of cells and could trigger induction of ER stress response proteins, leading to apoptotic death . Recently it has been shown that factors related to oxidative stress (reactive oxygen species and nitric oxide) and palmitate-derived ceramides are not causative factors for loss of GSIS following chronic exposure to NEFA . These data, together with the observations that non-toxic concentrations of unsaturated NEFA result in loss of GSIS, suggest that the effects of lipids on insulin secretion and cell viability are unrelated.
Our results show that within 24 h the composition of the cellular phospholipid fraction is modified by the NEFA in the incubate. This is in agreement with previous studies on beta cells . Changes in the composition of phospholipids, particularly incorporation of saturated NEFA, has implications for cell membrane rigidity and fluidity which could affect membrane signalling, insulin secretion by granule trafficking, fusion of secretory granules to the cell membrane during exocytosis, and protein processing in the endoplasmic reticulum [47, 48]. In addition, the composition of phospholipids in lipid rafts has been shown to be affected by dietary lipids , which has implications for cellular functions such as the recycling of receptors and endocytosis.
Although subtle changes occur in specific plasma fatty acid ratios in obesity, a mixture of saturated and unsaturated NEFA still circulates . Some reports of in vitro experiments using palmitic acid as a surrogate for in vivo conditions of elevated saturated lipids have clouded the view of the putative cytoxicity of fatty acids. Whereas isolated cells are able to take up and metabolise saturated lipids , in the absence of unsaturated NEFA they are unable to form a normal mixture of TG stores. This results in cell death via the morphological changes we have shown. The behaviour of INS1 cells and primary islets may differ in this respect. These marked changes in beta cell morphology with concomitant destruction of cell integrity are likely to mask the more subtle metabolic changes occurring due to physiological changes in fatty acid concentration and composition. The hypothesis that lipid-induced beta-cell death (lipoapoptosis) is a major aetiological factor in type 2 diabetes has evolved partly from such in vitro investigations. Our data would suggest that decreased GSIS derives from an imbalance in the sensitive interaction between glucose and lipid metabolism  coupled with important changes in phospholipids in beta cells. We believe that ‘lipotoxicity’ describes a metabolic dysregulation rather than a cellular insult and should therefore not be studied in cell models supplying a single species of saturated fatty acid.
We are grateful to the Wellcome Trust for funding (J. H. Moffitt, B. A. Fielding, A. Clark); the Trust also provided funding for the electron microscope in the Department of Human Anatomy and Genetics, Oxford, and for the gas chromatograph (Oxford Centre for Diabetes, Endocrinology and Metabolism). Glucose-sensitive INS1 cells were kindly provided by Prof. G. Hitman (London). We thank Sarah Rodgers for her technical expertise with electron microscopical techniques, and Prof. John Morris (Department of Human Anatomy and Genetics), Dr. Fredrik Karpe and Prof. Keith Frayn (Oxford Centre for Diabetes, Endocrinology and Metabolism) for invaluable support and advice. We thank Prof. David Vaux (Dunn School of Pathology, Oxford) for his advice and for the antibody to protein disulphide isomerase. The use of the NERC mass spectrometry facility (grant no. F14/6/13) is gratefully acknowledged.