Stigmasterol prevents glucolipotoxicity induced defects in glucose-stimulated insulin secretion

Type 2 diabetes results from defects in both insulin sensitivity and insulin secretion. Elevated cholesterol content within pancreatic β-cells has been shown to reduce β-cell function and increase β-cell apoptosis. Hyperglycemia and dyslipidemia contribute to glucolipotoxicity that leads to type 2 diabetes. Here we examined the capacity of glucolipotoxicity to induce free cholesterol accumulation in human pancreatic islets and the INS-1 insulinoma cell line. Glucolipotoxicity treatment increased free cholesterol in β-cells, which was accompanied by increased reactive oxygen species (ROS) production and decreased insulin secretion. Addition of AAPH, a free radical generator, was able to increase filipin staining indicating a link between ROS production and increased cholesterol in β-cells. We also showed the ability of stigmasterol, a common food-derived phytosterol with anti-atherosclerotic potential, to prevent the increase in both free cholesterol and ROS levels induced by glucolipotoxicity in INS-1 cells. Stigmasterol addition also inhibited early apoptosis, increased total insulin, promoted actin reorganization, and improved insulin secretion in cells exposed to glucolipotoxicity. Overall, these data indicate cholesterol accumulation as an underlying mechanism for glucolipotoxicity-induced defects in insulin secretion and stigmasterol treatment as a potential strategy to protect β-cell function during diabetes progression.

(HMG-CoA reductase, the rate-limiting enzyme for cholesterol synthesis), are best at reducing serum cholesterol in order to prevent cardiovascular diseases, they present a modestly increased risk for the onset of new diabetes cases [16][17][18][19] . Possible mechanisms include side effects that are independent of statin's cholesterol-lowering property, such as a statin-specific immune response leading to insulin resistance 20 . Therefore, alternative ways to reduce cholesterol in β-cells are needed.
Phytosterols are found in plant foods and are analogous to cholesterol in mammals. Phytosterols have been of interest due to their well-documented ability to decrease serum LDL-cholesterol levels in humans 21,22 , with the greatest reductions at 2.5 grams of phytosterols per day 23 . Given that the typical Western diet provides 200-400 milligrams of phytosterols per day 24 , we may not be consuming phytosterols at a level needed for maximum health benefits. Absorption of phytosterols is regulated by small intestinal ATP-binding cassette transporters that export phytosterols into the lumen 25 . Although the amount of phytosterols absorbed into serum is low, it can accumulate in tissues. A study by Alhazzaa 26 showed that feeding rats phytosterols increases accumulation in all the internal organs measured except the brain.
Commonly consumed dietary phytosterols include stigmasterol, campesterol and sitosterol. Stigmasterol has been widely investigated for its anti-atherosclerotic cholesterol-lowering effects. Stigmasterol increases cholesterol efflux and decreases LDL-induced proinflammatory cytokine secretion, whereas campesterol and sitosterol have no beneficial effects 27 . The purpose of this study is to assess the capacity for stigmasterol, a common food-derived phytosterol with anti-atherosclerotic potential, to prevent β-cell dysfunction induced by glucolipotoxicity.

Glucolipotoxicity reduces insulin secretion by inducing cholesterol accumulation.
Cholesterol excess is a possible contributing factor to β-cell failure. To determine if glucolipotoxicity treatment can increase free cholesterol within β-cells, INS-1 insulinoma cells and human islets were treated with HGP (30 mM glucose, 0.5 mM palmitate) or normal growth medium (LG) for 72 h. Filipin is a widely used fluorescent probe to measure free cholesterol in cells 28 . HGP treatment increased filipin staining in INS-1 cells and human islets (Fig. 1a-c). The result from INS-1 cells was consistent with other reports 10 . To our knowledge, this was the first time the effect of glucolipotoxicity on cholesterol in human islet was examined (Fig. 1b,c). Cholesterol accumulation may occur through increased uptake of LDL, increased biosynthesis, or decreased efflux to acceptors such as HDL. Treatment with mevinolin (Mev), which inhibits the rate-limiting enzyme of cholesterol biosynthesis HMG-CoA reductase, dose-dependently reduced cholesterol accumulation in cells exposed to HGP (Fig. 1d). Similarly, removing LDL as a source for cholesterol uptake by using lipoprotein-deficient serum (LPDS) in place of FBS or adding HDL to LPDS growth medium also inhibited HGP-induced cholesterol accumulation in INS-1 cells (Fig. 1e).
The increase in cellular free cholesterol was accompanied by reduced glucose-stimulated insulin secretion (GSIS) in both INS-1 cells and human islets treated with HGP (Fig. 1f). Consistent with an inhibitory role of elevated cholesterol in GSIS, GSIS was normalized when HGP-induced cholesterol accumulation was blocked by mevinolin (Fig. 1g). The effect of mevinolin treatment was accomplished through normalization of basal insulin secretion (Fig. 1h), with little effect on stimulated insulin secretion (Fig. 1i).

Oxidative stress increases β-cell cholesterol
Pancreatic β-cells are particularly susceptible to oxidative stress because they express low levels of intracellular antioxidants 29 . ROS has been linked to cholesterol accumulation in neurons, vascular smooth muscle cells, and hepatocytes [30][31][32] . To determine if ROS may be the underlying cause for HGP-induced cholesterol accumulation, INS-1 cells were incubated with HGP for time periods up to 48 h. As shown in Fig. 2a, HGP increased ROS production as estimated by carboxy-H 2 DFFDA staining. Carboxy-H 2 DFFDA becomes fluorescent when oxidized and is used as a general marker of ROS 33 . The increase in ROS corresponded to a rise in filipin staining (Fig. 2b). The most significant increase in ROS (Fig. 2c) and free cholesterol (Fig. 2d) both took place between 6 and 24 h after HGP treatment, suggesting that the rise in ROS and cholesterol may occur with similar time course.
To determine if oxidative stress alone increases cholesterol accumulation in β-cells, INS-1 cells were incubated with AAPH, a water-soluble free radical generator used to produce ROS 34,35 (Supplementary Fig. 1). Cholesterol increased in INS-1 cells treated with varying concentrations of AAPH for 4 h in serum-free media (Fig. 3a). AAPH increased oxidation (H 2 DFFDA) and filipin staining with a similar temporal pattern, with the most significant rise occurring between one and two hours after AAPH addition (Fig. 3b). At the higher concentration range (5 mg/ml), AAPH was sufficient to increase free cholesterol labeled by filipin staining within 1 h (Fig. 3c). Esterified cholesterol labeled by lipidtox was also increased by AAPH during the same time frame ( Supplementary  Fig. 2). These data indicate that cholesterol accumulation in β-cells is sensitive to free radical generation. To further determine if ROS is the underlying cause for HGP-induced cholesterol increase in β-cells (Fig. 2), we treated INS-1 cells exposed to HGP with the antioxidant N-acetylcysteine (NAC) to prevent HGP-induced ROS production. Mevinolin (Mev) was included as a positive control of cholesterol depletion. As shown in Fig. 3d, NAC prevented an increase in filipin staining in HGP-treated INS-1 cells. Together with the results from AAPH, these data suggest that ROS stress can raise free cholesterol within β-cells, and the presence of antioxidants can prevent HGP-induced cholesterol increase.
Stigmasterol reduces free but not esterified cholesterol. Because glucolipotoxicity increases β-cell cholesterol, which is at least partially responsible for reduced GSIS observed in HGP-treated cells, we next tested whether including stigmasterol with HGP could rescue β-cell function. While there are studies of stigmasterol's effects on lowering plasma cholesterol levels, how stigmasterol impacts cholesterol pools in cells is not well investigated. Cholesterol is found in either esterified or non-esterified (free) pools within the cell. We utilized filipin to measure free cholesterol and lipidtox to measure neutral lipids in INS-1 cells. As shown in Fig. 4a and b, both filipin and lipidtox staining increased with HGP treatment (HGP vs LG), indicating increased free cholesterol accumulation and conversion to cholesteryl esters. Stigmasterol decreased filipin staining due to HGP treatment (Fig. 4a, HGP + Stig vs HGP). The effect appeared to be specific to free cholesterol, as lipidtox staining, a measure of esterified cholesterol, did not change (Fig. 4b). LPDS treatment, on the hand, reduced both forms of cholesterol in INS-1 cells (Supplementary Fig. 3). At a high concentration, stigmasterol was as effective as LPDS at reducing free cholesterol, but remained ineffective at reducing esterified cholesterol ( Supplementary Fig. 3).

Stigmasterol decreases glucolipotoxicity-induced ROS production.
Stigmasterol is reported to increase glutathione content in mice, indicating an antioxidative property for stigmasterol 36 . Because β-cells naturally have low levels of glutathione 29 , stigmasterol treatment could act by reducing the ROS that accompanies glucolipotoxicity. Increased carboxy-H 2 DFFDA fluorescence, an indication of elevated ROS, was found in INS-1 cells treated with HGP (Figs 2 and 4c, HGP vs LG). Interestingly, the addition of stigmasterol with HGP decreased the amount of carboxy-H 2 DFFDA fluorescence, consistent with the idea that stigmasterol could decrease ROS production in β-cells exposed to glucolipotoxicity (Fig. 4c).

Stigmasterol alters expression of cholesterol homeostasis genes. Because different phytosterols
have cell-type specific effects on cholesterol regulatory genes 27 , we examined whether stigmasterol influences genes associated with cholesterol homeostasis in β-cells. Genes involved in cholesterol regulation are under the control of sterol regulatory element-binding protein 2 (SREBP2). The activation of SREBP2 is well characterized and is highly regulated by intracellular cholesterol levels. Low cholesterol levels are sensed in the ER by SCAP, which chaperones SREBP2 to the Golgi apparatus, where SREBP2 is cleaved by proteases, and is transported to the nucleus. The transcription of the SREBP2 gene Srebf2 is under a feed forward mechanism, where the transcription of Srebf2 is increased by SREBP2 activation 37 . Stigmasterol has been shown to interfere with SREBP processing 38 , which may elicit a compensatory response in Srebf2 expression in stigmasterol-treated cells. As shown in Fig. 5a, the addition of stigmasterol to either LG or HGP treated INS-1 cells increased transcription of Srebf2 (Stig vs Control). An SREBP2 target gene Ldlr 39 , responsible for cholesterol uptake from LDL, was similarly upregulated (Fig. 5b). HGP + Stig treated cells had decreased cholesterol efflux transporter ABCA1 when compared with HGP treatment at both the protein and transcriptional level (Fig. 5c-e). The exact mechanism that causes cholesterol genes to change in cells loaded with phytosterols is unknown. It has been shown that it is the accumulation of stigmasterol, rather than changes in the cholesterol level, that is responsible for altered expression of cholesterol homeostasis genes 38 . Stigmasterol improves β-cell function and reduces β-cell apoptosis. The direct effect of stigmasterol on β-cells is unclear. While Panda and colleagues reported a decrease in serum glucose accompanied by increased circulating insulin when stigmasterol was administered to mice 36 , others have shown no effect of stigmasterol feeding on glucose tolerance 40 . Glucolipotoxicity is known to induce β-cell death and reduce β-cell function 41 . We first examined the effect of stigmasterol on β-cell survival. External phosphatidylserine is a marker of early apoptosis, since phosphatidylserine is normally located on the inner leaflet of the plasma membrane, but is flipped to the outer membrane during apoptosis and necrosis 42 . Annexin V binds to phosphatidylserine; thus, increased annexin V staining is a measure of cell death. As shown in Fig. 6a, HGP treatment increased annexin V staining in INS-1 cells. The addition of stigmasterol partially blocked the increase in annexin V, indicating that stigmasterol prevents β-cell death due to glucolipotoxicity. Similar results were obtained with Western blotting of active caspase-3 ( Supplementary Fig. 4), a marker for apoptosis 43 .
In order to examine the direct effects of stigmasterol on β-cell function, static GSIS assays were performed. Addition of stigmasterol rescued GSIS in cells treated with HGP (Fig. 6b). Under this condition, stigmasterol also exerted a beneficial effect of increasing total insulin in HGP treated INS-1 cells (Fig. 6c). Together these data indicate that stigmasterol treatment can preserve β-cell function and survival in the presence of glucolipotoxicity. Stigmasterol rescues glucose-stimulated actin reorganization. Actin reorganization plays an essential role in glucose-stimulated insulin secretion. Pancreatic β-cells display a dense actin web below the plasma membrane that may inhibit granule docking. Glucose-stimulated actin reorganization serves to remove the F-actin barrier and permit access of granules to the plasma membrane. When this process is inhibited, GSIS is reduced 44,45 . In the next series of experiments, we tested the hypothesis that glucolipotoxicity inhibits GSIS through promoting actin polymerization induced by excess cholesterol and that this process could be prevented by stigmasterol. We previously showed that cholesterol overloading induces actin polymerization, leading to membrane protrusion, membrane ruffling, and cell expansion in cultured β-cells, suggesting a role of cholesterol in regulating cortical F-actin 46 . Here we extended our previous actin studies of fixed cells to live cell total internal reflection fluorescence microscopy (TIRFM) studies of real-time cortical actin dynamics in β-cells expressing actin-GFP. TIRFM was used to ensure that the focus was on cortical F-actin at the plasma membrane where insulin granule exocytosis took place. Within minutes after adding soluble cholesterol, formation of actin filaments was seen in both cells in Fig. 7a. Rapid actin polymerization occurred in the lower cell, evidenced especially by the thick actin ring around the rim (arrows). Enlarged images of the upper cell are presented in the bottom panels, which show that long actin filaments appeared from what were previously smaller fragments (arrowheads). This set of images provides evidence for a direct involvement of excess cholesterol in actin polymerization in cultured β-cells.
To further confirm that cholesterol regulates actin reorganization upon glucose stimulation, a process that is critical for GSIS, cells were first cholesterol depleted (by MβCD) or overloaded (by soluble cholesterol, "CHOL"), then stimulated with 20 mM glucose, fixed and stained using phalloidin, a probe specific for polymerized F-actin 47 (Fig. 7b). Compared with basal unstimulated cells, glucose-stimulated cells displayed slightly less F-actin staining, consistent with the idea that glucose stimulates actin reorganization to remove the F-actin barrier and thus to facilitate insulin release 44,45 . Cholesterol overloaded cells stimulated with glucose ("CHOL"), on the other hand, had a mesh-like layer of F-actin network and significantly increased amount of F-actin, compared with untreated cells stimulated with glucose. Quantification of F-actin content under the four conditions is presented in Fig. 7b. These data show that cholesterol is actively involved in the actin reorganization process during glucose stimulation.
Finally, we examined phalloidin staining in HGP-treated cells with or without stigmasterol. HGP treatment significantly increased F-actin in INS-1 cells, which was normalized by stigmasterol treatment (Fig. 7c). Cholesterol depletion by LPDS, serving as a positive control for cholesterol-mediated actin dynamics, also reduced polymerized actin. The results from Fig. 7 support a model that the crucial step of glucose-stimulated  actin depolymerization in GSIS is inhibited by glucolipotoxicity-induced cholesterol accumulation and that stigmasterol rescues GSIS at least partially by increasing actin dynamics in β-cells.

Discussion
The study presented here shows that stigmasterol protects pancreatic β-cells from glucolipotoxicity by preventing accumulation of free cholesterol and ROS, improving insulin secretion, increasing insulin content and decreasing markers for early apoptosis. We provide evidence that glucolipotoxicity stimulates ROS production, and that free radicals in turn promote cholesterol accumulation within β-cells. Stigmasterol treatment can normalize cellular cholesterol levels by reducing ROS production from glucolipotoxicity in β-cells. Indeed, stigmasterol can serve as a mild antioxidant or pro-oxidant to lipid peroxidation depending on the substrate 48 . Intraperitoneal injection of stigmasterol decreases lipid peroxidation and increases scavenger proteins in the livers of Ehrlich Ascites Carcinoma mice 49 . Stigmasterol did not increase Abca1 expression in INS-1 cells, indicating that stigmasterol may work differently in β-cells than in macrophages 27 .
It has long been recognized that Type 2 diabetes is associated with disordered lipid homeostasis, including increased VLDL and decreased HDL 50 . Thus, perturbations in circulating cholesterol may be a predisposing factor for type 2 diabetes. In agreement with these clinical findings, excess cholesterol and disrupted cholesterol export can promote β-cell dysfunction 12,51,52 . HMG-CoA reductase inhibitors (statins) are widely used to decrease serum cholesterol to prevent cardiovascular diseases. Several meta-analyses indicate, however, that statins may actually increase diabetes risk [16][17][18][19] . In vitro studies also suggest that statin treatment can negatively affect β-cells. Statins increase basal insulin secretion in MIN-6 cells, lowering the fold change in insulin secretion upon glucose stimulation 53 . Atorvastatin decreased insulin content in rat islets and INS-1 cells 54 . It should be noted that these in vitro studies were performed in systems with normal cholesterol levels. Cholesterol extraction using MβCD from LDLR −/− islets with elevated cholesterol reported an improvement in insulin secretion 55 . Therefore, depletion of β-cell cholesterol may be beneficial during times of cholesterol excess.
Studies on the link between statin usage and increased diabetes risk caution us against relying on statins to reduce cholesterol as a means to improve β-cell function under glucolipotoxicity. Few studies have focused on the effect of phytosterols on diabetes. Db/db mice represent a widely used animal model for obesity-induced diabetes. When they were fed Aloe Vera, they displayed decreased fasting blood glucose and HbA1c levels (a marker of blood glucose control) which were attributable to the phytosterol containing extracts 56 . Similarly, Zucker diabetic fatty rats exhibited increased glucose tolerance and decreased HbA1c levels when Aloe Vera derived phytosterols were administered orally 57 . Patients fed a phytosterol-enriched spread displayed a small, but significant, decrease in HbA1c levels 58 . One study looking at the effects of phytosterols on high fat fed mice reported that phytosterols did not improve glucose and insulin tolerance 40 , although β-cell function was not measured.
Cortical F-actin is dramatically increased in cholesterol-overloaded cells (Fig. 7), which could inhibit insulin granule exocytosis by several mechanisms. It may present a physical barrier, which impairs the ability of insulin granules to dock at the plasma membrane and reduces the size of the readily releasable pool of insulin granules. Indeed, an increase in the readily releasable pool of insulin granules has been suggested in cholesterol depleted cells 59 and cholesterol overloading reduces the number of insulin granules docked at the plasma membrane 60 . Dense F-actin could restrict syntaxin 4 accessibility, which is required for insulin exocytosis 61 . Disruption of F-actin increases syntaxin 4 accessibility to VAMP2 by promoting dissociation of syntaxin 4 from F-actin, leading to increased GSIS 62 . Because cholesterol overloading results in cortical F-actin polymerization, it is likely that cholesterol alteration would interfere with the interaction between syntaxin 4 and F-actin, thereby affecting the required function of syntaxin 4 in GSIS. One potential mechanism for the involvement of stigmasterol in rescuing GSIS from HGP-treated β-cells is reducing the detrimental effect of excess cholesterol on glucose-stimulated actin reorganization.
In conclusion, our study shows beneficial effects of stigmasterol treatment on diabetic β-cells. Further studies will be needed to determine if oral intake of stigmasterol prevents β-cell dysfunction in diabetes.

Materials.
Alexa546-phalloidin, lipidTOX red, actin-GFP and carboxy-H 2 DFFDA were from Thermo Fisher Scientific. AAPH was from Cayman Chemical. Rabbit anti-ABCA1 was from Novus Biologicals. Antibodies to active Caspase-3 and GAPDH were from Abcam. Filipin, mevinolin, N-acetyl-l-cysteine (NAC), MβCD, water-soluble cholesterol (CHOL) and all other chemicals were from Sigma. KRBH buffer (128. 8  Cholesterol manipulation. Cholesterol depletion using methyl-β-cyclodextrin (MβCD) was done by incubating cells with 5 mM MβCD at 37 °C for 1 h; to cholesterol overload, cells were incubated with 5 mM soluble cholesterol (CHOL, 1 g contains approximately 40 mg cholesterol) at 37 °C for 1 h. LPDS treatment was done by culturing cells in normal growth medium with 10% FBS replaced by lipoprotein deficient serum (LPDS medium). Mevinolin treatment was done by including 1 μM mevinolin in LPDS medium. HDL treatment was done by including 50 μg/ml HDL in LPDS growth medium.
Glucose stimulated insulin secretion. Static glucose stimulated insulin secretion was measured as previously described 65 . Briefly, INS-1 cells were incubated in glucose-free RPMI-1640 culture medium for 2 h. Cells were starved for 1 h in glucose-free KRBH buffer. INS-1 cells were incubated in 2 mM glucose in KRBH for 1 h (basal samples), followed by 1 h in 20 mM glucose in KRBH (glucose-stimulated samples). Incubations were performed at 37 °C in an air incubator. Total insulin samples were taken after 1% Triton X-100 extraction. To measure the basal and stimulated insulin secretion separately, secreted insulin was normalized to total protein. Insulin samples were measured by insulin ELISA kits (Alpco Diagnostics).
Fluorescence Microscopy. Cells were grown in imaging dishes similar to MatTek dishes. Wide-field fluorescence microscopy utilized a Leica DMIRB microscope (Leica Mikroscopie und Systeme GmbH, 35578 Wetzlar, Germany) equipped with a Princeton Instruments (Princeton, NJ, USA) cooled charge coupled device (CCD) using MetaMorph Imaging System software (Molecular Devices). Images were acquired using an oil-immersion objective (×40). For filipin and lipidtox staining, INS-1 cells were fixed in 1% paraformaldehyde for 20 minutes then incubated in filipin (50 μg/ml) or lipidtox (1:1000 dilution) for 45 min. Filipin and lipidtox images were captured using UV and rhodamine filters, respectively. Details of filipin staining and imaging are described previously 66 . For carboxy-H 2 DFFDA fluorescence, INS-1 cells were incubated in 20 μM carboxy-H 2 DFFDA for 1 h. Cells were washed and counterstained with Hoechst 33342 to label the nuclei for cell counting. For phalloidin staining, cells were fixed with 4% paraformaldehyde for 20 min, permeabilized with 0.1% Triton X-100 for 5 min, and stained with Alexa 546-phalloidin. All staining was done at room temperature.
TIRFM. Transient transfections were performed with Lipofectamine 2000 (Thermo Fisher Scientific) according to the manufacturer's protocol and cells were cultured for 48 h prior to microscopy. Actin-GFP-expressing MIN6 cells grown in MatTek dishes were kept in an Air-therm (WPI) temperature-regulated environmental box at 37 °C throughout the imaging experiments. Cells were imaged for 2 min to establish basal baseline. A 10 × concentrated stock of soluble cholesterol (CHOL) was then added to the edge of the MatTek dish on the microscope stage. TIRFM was performed using an Olympus objective-type IX-70 inverted microscope fitted with a 60×/1.45 NA TIRFM lens (Olympus), controlled by Andor iQ software (Andor Technologies), and detected with a back-illuminated Andor iXon 897 EMCCD camera (512 × 512, 14 bit; Andor Technologies). The depth of the evanescent field was calculated to be 98 nm 67 .
Western blotting. Total cell lysates were prepared, and protein concentration was determined by BCA (Thermo Scientific). Samples were immunoblotted against active caspase 3 (1:1000) and GAPDH (1:2000) at 4 °C overnight. Secondary antibodies conjugated with HRP were incubated for 1 h at room temperature. The membranes were then developed by ECL (Thermo Scientific), according to the manufacturer's instructions.
Image quantification and statistics. All images were analyzed using MetaMorph software (Molecular Devices). Images were background subtracted and either quantified as individual cells or whole fields normalized to cell counts. Unless otherwise indicated, each data point represents the average of three independent experiments. Data are presented as the mean ± SEM. Statistical significance was analyzed using unpaired Student's t tests for comparison between two groups and one-way repeated measures ANOVA with Post-hoc comparisons using the Tukey HSD test for data from more than two groups, with p-values less than 0.05 deemed significant. Data availability. All data generated or analyzed during this study are included in this published article (and its Supplementary Information files).