Endogenous Fatty Acids Are Essential Signaling Factors of Pancreatic β-Cells and Insulin Secretion

In vertebrates, insulin secretion is a synchronized multicellular process. Sufficiently high blood insulin levels are ensured by the coordinated stimulation of β-cells in pancreatic islets. Coupling of the secretory activity of pancreatic islets results in oscillatory elevations of insulin levels in the bloodstream (1). Studies on patients with diabetes indicate that under oscillatory infusion of insulin, less insulin is required to sustain normoglycemia compared with a constant insulin supply (2). Disturbed synchronized oscillatory insulin secretion was found in hyperglycemic and hyperinsulinemic ob/ob mice and was suggested to contribute to insulin resistance and to the development of type 2 diabetes mellitus (3). The coordination of islets in the pancreas, in addition to gap junctions (4), relies on indirect intercellular communication (i.e., on soluble messengers that reach β-cells by diffusion in the intercellular space of islets and across pancreatic lobes) (5). Individual β-cell activity as well as cell-to-cell coordination in pancreatic islets are predominantly fine-tuned by the close coordination of a multitude of diffusible factors that oscillate in the extracellular space. This includes a number of neurotransmitters and hormones, such as γ-aminobutyric acid (6) and neuropeptide Y (7). In particular, β-cells are equipped with many G-protein–coupled receptors, which are involved in intracellular Ca²⁺ signaling and significantly contribute to orchestrating oscillations of the intracellular free Ca²⁺ concentration ([Ca²⁺]_i oscillations) and insulin secretion. Examples are purinergic receptors that are triggered by the auto- and paracrine release of ATP from β-cells (8). In 2003, three independent groups reported that GPR40, a G-protein–coupled receptor that is abundantly expressed in the brain and in the pancreas, acts as a receptor for medium- and long-chain fatty acids (FAs) (9–11). GPR40 was found to mediate the majority of the effects that FAs cause on insulin secretion, promoting increased [Ca²⁺]_i (12,13). Deletion of GPR40 expression in pancreatic β-cells led to resistance toward this FA-induced response in [Ca²⁺]_i oscillations and insulin secretion (12,14). GPR40 overexpression, on the other hand, was described to augment glucose-stimulated insulin secretion (GSIS) (15), pointing to a major role of GPR40 in FA-stimulated [Ca²⁺]_i signaling and GSIS. Recently, we have shown that photorelease of arachidonic acid leads to GPR40 stimulation and subsequent activation of trimeric G-proteins together with [Ca²⁺]_i signaling in MIN6 cells (16). For decades, elevated levels of FAs in the bloodstream have been described to significantly augment GSIS in isolated islets (17), animals (18), and nonobese human subjects (19). Studies in rat islets indicate a hypersensitivity of β-cells to glucose after prolonged exposure to FAs (20). In turn, clinical studies correlated artificial decreases of plasma FA levels with the reduction of insulin secretion, indicating an essential role of circulating FAs for efficient GSIS (21,22). Even though FAs have been discussed as insulin secretagogues for decades (as reviewed in Nolan et al. [23]), the role of endogenous FAs in single β-cell activity, intercellular and interislet signaling, and in the synchronization of β-cells has not been investigated so far. In the current study we therefore aimed 1) to determine the role of endogenous FAs as essential endogenous signaling factors for β-cell activity and insulin secretion and 2) to examine whether β-cell activity can be modulated by the removal or replenishment of endogenous FAs. Our approach to subordinate β-cell activity and insulin secretion to the presence of endogenous signaling factors, in particular FAs, contributes to a more detailed and complete understanding of the fundamental regulation of β-cell activity and insulin secretion.

BSA (FA free [FAF]: catalog #A7030, lot #SLBK9735V; conventional: catalog #A7906, lot #SLBR0420V), ¹³C-oleic, ¹³C-palmitic, and ¹³C-decanoic acid were obtained from Sigma-Aldrich (St. Louis, MO). Thermo Fisher Scientific (Carlsbad, CA) provided gradient-grade Gibco FBS (catalog #16000–044) and DMEM. Solvents were obtained in the highest analytical grade from Biosolve (Valkenswaard, the Netherlands) and Sigma-Aldrich. All other commercially available chemicals and enzymes were purchased from Sigma-Aldrich and Cayman Chemical (Ann Arbor, MI). See Supplementary Data for unit definitions of applied recombinant enzymes.

Insulin-secreting MIN6 cells (24) (passage 25–36) were cultured at 37°C and 8% CO₂ in a humidified atmosphere. MIN6 cells were grown in DMEM containing 4.5 g/L glucose, supplemented with 15% FBS and 70 μmol/L β-mercaptoethanol (PAN Biotech, Aidenbach, Germany). MIN6 cells were plated onto eight-well LabTek microscope dishes (catalog #155411; Thermo Fisher Scientific, Rockford, IL), 40-mm coverslips (Menzel Gläser, Braunschweig, Germany) for imaging, or ∅ 60 mm/∅ 35 mm dishes (Nunc Delta Surface, catalog #150288/catalog #153066; Thermo Fisher Scientific, Roskilde, Denmark) to form pseudoislets for in vitro assays.

Mouse islets were isolated from C57BL/6 mice (female Ctrl:CD1(ICR), outbred (catalog #CD1S1FE07W; Charles River Laboratories) by collagenase digestion, as previously described (25). Briefly, mice were euthanized by cervical dislocation, and a collagenase solution (1 mg/mL) was injected into the bile duct, followed by digestion at 37°C for 10 min and separation of islets using a Histopaque (Sigma-Aldrich) gradient (1.083 and 1.077 g/mL). Islets were cultured in RPMI medium (supplemented with 10% FCS, 100 units/mL penicillin, and 100 mg/mL streptomycin) for 24 h. Upon trypsin digestion (5 min at 37°C), islets were dissociated into single β-cells, which were seeded on poly-l-lysine–coated coverslips. Animals were housed in the European Molecular Biology Laboratory animal facilities under veterinarian supervision and the guidelines of the European Commission (revised directive 2010/63/EU, and American Veterinary Medical Association guidelines 2007).

ELISA for the specific quantification of insulin was performed in 96-well format (Mercodia AB, Uppsala, Sweden). Insulin levels were normalized to the protein content of MIN6 cells, using a BCA assay (Pierce BCA Protein assay kit; Thermo Fisher Scientific). Experiments for the determination of supernatant (SN) insulin content were performed in quadruplicate per condition.

Imaging was performed on a FluoView1200 (catalog #IX83; Olympus) confocal laser-scanning microscope at 37°C (incubator box made by the European Molecular Biology Laboratory), using Olympus 60× Plan-Apo (numerical aperture 1.4, oil) or 20× UPLSAPO (numerical aperture 0.75, air) objectives and FluoView software, version 4.2. The images were acquired with a Hamamatsu C9100-50 EM-CCD camera. The green channel was imaged using a 488-nm laser line (120 mW/cm², 2.5%) and a 525/50 emission mirror. The red channel was imaged using a 559-nm laser (120 mW/cm², 2.0%) and a 643/50 emission filter. Images were acquired in 4-s intervals (frame time 3.9 s). A pulsed 375-nm laser (10 MHz) was applied for uncaging experiments in the entire field of view for eight frames (3.2 s/frame). The dual scanner setup allowed for simultaneous laser stimulation and confocal imaging. This permitted the acquisition of cellular responses that occur during or immediately after laser stimulation. To test changes in [Ca²⁺]_i at the single-cell level, MIN6 cells, grown to 70% confluence in pseudoislets, were incubated with an acetoxymethyl ester of the Ca²⁺ indicator Fluo-4 (Life Technologies, Eugene, OR), 5 μmol/L in DMEM (1 g/L glucose), for 20 min at 37°C. Cell experiments were performed in HEPES buffer (in mmol/L: 115 NaCl, 1.2 CaCl₂, 1.2 MgCl₂, 1.2 K₂HPO₄, and 20 HEPES, pH 7.4).

For [Ca²⁺]_i imaging experiments, HEPES buffer (1.5 mL) was preincubated for 30 min on 2 × 10⁶ MIN6 cells, grown to ∼80% confluence in ∅ 60 mm dishes (Thermo Fisher Scientific). One hundred fifty microliters of this SN was transferred to LabTek microscope dishes with prewashed and Fluo-4–stained MIN6 cells residing in 150 µL of buffer (1:1 dilution). In dish experiments for the determination of insulin levels, 1.5 mL of SN was transferred to 2 × 10⁶ target cells in ∅ 60 mm dishes (Thermo Fisher Scientific), residing in 1.5 mL of buffer.

For perfusion experiments, MIN6 cells were seeded on 40-mm glass coverslips and mounted into a Bioptechs FCS2 chamber (chamber volume 0.53 cm³, ∅ 30 mm), connected to a calibrated FCS2 microperfusion pump using Teflon-coated tubing. MIN6 cells were perfused at a constant flow rate of 0.15 mL/min for 30 min. For static incubations, MIN6 cells, prepared in equal densities in ∅ 30 mm dishes were incubated in buffer for 30 min. Samples were tested for their insulin content (ELISA; Mercodia AB) after respective dilution to equalize differences in volumes. Experiments were performed in the presence of 11 mmol/L glucose.

Fluorescence intensities were extracted from individual cells as a function of time using Fiji (26) and were calculated relative to the maximum detected fluorescence intensity after subtraction of background (F/F₀). Representative cells within the field of view were averaged to generate [Ca²⁺]_i traces or to determine the number of detected high-intensity [Ca²⁺]_i events within every 60-s interval. The height of each [Ca²⁺]_i event was determined relative to the highest detected peak in each trace as a criterion for grouping [Ca²⁺]_i transients into high-intensity events (≥60% of highest peak) and lowintensity events (<60% of highest peak). For statistical analysis, four to six independent experiments were performed per condition. At least 30 individual MIN6 cells or 60 mouse primary β-cells showing representative behavior were picked per condition, and their responses were averaged.

A solution of Nile red 2-O-butyric acid (NRBA) and FAF-BSA (10 μmol/L) in a 1:1 ratio was preincubated at 37°C for 20 min. NRBA preferably binds hydrophobic protein domains along with a turn-on of fluorescence emission and a strong shift to shorter wavelengths (27). Buffer (11 mmol/L glucose) was incubated on 2 × 10⁶ MIN6 cells, allowing for the accumulation of FAs. The FA-containing buffer was concentrated to 10% of its original volume using a concentrator (catalog #7310032; Labconco, Fort Scott, KS) and added to NRBA-loaded FAF-BSA. Assays were performed on a QuantaMaster QM4/2000SE Photon Technology International spectrofluorometer in 150-µL quartz cuvettes. After excitation at λ = 540 nm, full emission scans were recorded (λ = 550–750 nm) with 2-nm step intervals and 0.2-s integration time.

A total of 1 × 10⁶ MIN6 cells were seeded to form pseudoislets in 60-mm dishes within 5 days. MIN6 cells were prestimulated with buffer (11 mmol/L glucose) for 30 min at 37°C. A total of 1.6 mL of buffer (supplemented with BSA, 11 mmol/L glucose) was incubated on MIN6 cells for 20 min at 37°C. A total of 1.5 mL of the SN was concentrated to 120 µL at 4°C using a concentrator (catalog #7310032; Labconco). MIN6 cells were scraped in 120 µL of buffer.

FAs were extracted using an MeOH/CHCl₃ extraction procedure at 4°C, according to Clark et al. (28). Seven hundred fifty microliters of mix A (4.48 mL of MeOH, 2.42 mL of CHCl₃, and 0.236 mL of HCl [1 N]) were added to biological samples, which were vortexed for 30 s. Thirty microliters of MeOH, containing the internal standards ¹³C-palmitic acid (100 ng/sample), ¹³C-oleic acid (OA) (50 ng/sample), and ¹³C-decanoic acid (25 ng/sample) were added per sample. Seven hundred twenty-five microliters of CHCl₃ and 170 μL of 2 mol/L HCl were added, followed by vortexing for 30 s. Samples were centrifuged at 8,000g (10 min, 4°C) to create a two-phase system with an upper aqueous layer and a protein band at the interface. The lower organic phase was collected in a fresh tube. Seven hundred eight microliters of mix B (7.2 mL of CHCl₃, 3.6 mL of MeOH, and 2.7 mL of 0.01 mol/L HCl) were added to the lower phase. Samples were mixed and centrifuged (8,000g, 10 min, 4°C). The upper phase was removed, and the samples were dried at 4°C using a concentrator (catalog #7310032; Labconco). All applied materials were pretreated with methanol to eliminate possible palmitate contaminations, according to Yao et al. (29).

Liquid chromatography-tandem mass spectrometry (MS) analysis was performed on a Vanquish UHPLC System (Thermo Fisher Scientific, Waltham, MA), coupled to a Q-Exactive Plus high-resolution mass spectrometer (Thermo Fisher Scientific) in electrospray ionization–negative mode. The separation of FAs was carried out on an Agilent Poroshell column (3 × 50 mm; 2.7 μm) at 0.26 mL/min and 40°C. The mobile phase consisted of solvents A (acetonitrile/water 6:4) and B (isopropyl alcohol/acetonitrile 9:1), buffered with 10 mmol/L ammonium acetate. The Vanquish UHPLC System was set up in gradient mode (Supplementary Table 1). FAs were detected with a full high-resolution mass spectrometer scan at the mass resolving power R = 70,000 in a mass range of 100–1,200 m/z. Data-dependent tandem MS scans were obtained along with full scans using higher-energy collisional dissociation with normalized collision energies of 10, 20, and 30 units at the mass resolving power R = 17,500. See Supplementary Data for details.

It is well established that static incubation of MIN6 cells in the presence of glucose (11 mmol/L) is characterized by periodic transients of cytosolic [Ca²⁺]_i ([Ca²⁺]_i oscillations) (Fig. 1Ai and Bi). In accordance with literature data, MIN6 cells respond to elevated glucose levels by [Ca²⁺]_i oscillations and increased insulin secretion, similar to mouse primary β-cells (Supplementary Fig. 1) (30). Because [Ca²⁺]_i oscillations correlate with insulin secretion (Supplementary Fig. 1) (31–33), they may serve as a sensitive readout of β-cell responses. [Ca²⁺]_i oscillations can be followed in real time, using Ca²⁺-sensitive cell-permeant fluorescent indicators, such as Fluo-4 (34). We observed that extensive perfusion (1.5 mL/min for 1 min) of MIN6 cells with buffer containing 11 mmol/L glucose paused [Ca²⁺]_i oscillations, which gradually recovered only after the termination of perfusion (Fig. 1Ai, Bi, and D). Similar effects were observed upon washing MIN6 cells and mouse primary β-cells in the microscopy dish before imaging (Fig. 1Aii, Bii, Av, and Bv). In line with this observation, constant perfusion (0.15 mL/min) of MIN6 cells significantly reduced insulin secretion, compared with static incubation (Fig. 1C), suggesting that stringent washing removed components from the extracellular space that are necessary for GSIS. In an attempt to verify whether endogenous signaling factors, secreted by β-cells, are modulating GSIS, we preincubated a population of 2 × 10⁶ MIN6 cells with buffer for 30 min (i.e., SN) and transferred this SN to a dish with freshly washed MIN6 cells. Such a buffer exchange led to an immediate and full recovery of [Ca²⁺]_i oscillations in all MIN6 cells (Fig. 1Aii, Bii, and E) and enhanced insulin secretion (Supplementary Fig. 4), and was observed for all tested SNs at various glucose concentrations (Supplementary Fig. 5). SNs generated from HeLa cells under identical conditions restarted [Ca²⁺]_i oscillations only in 3 of 100 MIN6 cells, but did not change insulin secretion (Supplementary Fig. 4).

Guided by our results from the photolysis of arachidonic acid in MIN6 cells (16), we decided to examine the effect of FA depletion on [Ca²⁺]_i oscillations and insulin secretion. Depletion of FAs in cell medium was achieved by adding FAF-BSA to MIN6 cells cultured with 11 mmol/L glucose (see the MS analysis below for quantification, Fig. 4). The addition of 1% and 0.1% FAF-BSA to the medium led to immediate termination of [Ca²⁺]_i oscillations in MIN6 cells and in mouse primary β-cells (Fig. 1Aiii, Biii, Aiv, Biv, G, and H and Supplementary Figs. 6, 7, and 8, Supplementary Movie 1), whereas the addition of 0.01% FAF-BSA had a weak to no effect (Supplementary Fig. 9). In the presence of 0.1% FAF-BSA, the levels of secreted insulin decreased by >80% (Fig. 2H). The exchange of FAF-BSA to BSA-free buffer led to slow (∼20–30 min) recovery of [Ca²⁺]_i oscillations in the case of 1% FAF-BSA and close to immediate recovery in the case of 0.1% and 0.01% FAF-BSA (Fig. 2Ai, Bi, and C and Supplementary Figs. 6, 7, and 9). The effects of BSA on [Ca²⁺]_i oscillations were readily reproduced using human serum albumin (see Supplementary Fig. 10) and were independent of the glucose levels that were applied (Supplementary Fig. 11). The addition of conventional BSA (i.e., BSA saturated by natural FA cargo) (Fig. 4C, mass spectrometric analysis) to MIN6 cells and mouse primary β-cells in 0.1% (and 1%) final concentrations did not show pronounced effects on [Ca²⁺]_i oscillations (Fig. 2Aii, Bii, Aiii, Biii, D, and E and Supplementary Figs. 6 and 7). Furthermore, FAF-BSA, presaturated with the BSA binders l-tryptophan (l-Trp) (35) or ibuprofen (Ibu) (36) exhibited much weaker effects on [Ca²⁺]_i oscillations in MIN6 cells and mouse primary β-cells than FAF-BSA alone (Fig. 2Aiv–vi, Biv–vi, F, and G and Supplementary Figs. 6 and 7), indicating that BSA with a low capacity to extract FAs does not significantly affect [Ca²⁺]_i oscillations.

The addition of FAF-BSA, preloaded by incubation on 2 × 10⁶ MIN6 cells, to prewashed MIN6 cells immediately resumed [Ca²⁺]_i oscillations (Fig. 1Av, Bv, and F), indicating that BSA was saturated with cargo that was readily off-loaded again. The recovery of [Ca²⁺]_i oscillations directly correlated with the preloading time of FAF-BSA on MIN6 cells and therefore presumably with the total amount of transferred FAs. Consistently, increased responses were observed for higher concentrations (5%) of BSA preloaded on MIN6 cells for the same preloading time (Supplementary Fig. 12). The addition of Ibu or l-Trp (400 μmol/L) to MIN6 cells in the presence of BSA preloaded on MIN6 cells also augmented [Ca²⁺]_i oscillations. This points to a stimulation of MIN6 cells caused by FAs that were released from BSA because of out-competition by Ibu or l-Trp (Supplementary Fig. 13). In order to verify whether FAs are actually the major cargo in BSA transfer experiments, the GPR40 inhibitor GW1100 (10 μmol/L) (37) was added together with FAF-BSA (1%, preloaded on MIN6 cells) to prewashed MIN6 cells. This led to one prominent [Ca²⁺]_i spike followed by the suppression of continuous [Ca²⁺]_i oscillations (Fig. 3Ai, Bi, and C). This suggests that that GPR40 activation is crucial for inducing [Ca²⁺]_i oscillations in glucose-stimulated MIN6 cells, in accordance with the findings of Schnell et al. (12). In line with this, GW1100-treated MIN6 cells showed a significant decrease (10 μmol/L) or termination (25 μmol/L) of [Ca²⁺]_i oscillations, possibly because of competition with endogenous FAs for GPR40 (Fig. 3Aiii, Biii, E, and F and Supplementary Fig. 14).

We turned to MS to demonstrate the enrichment of FAs upon incubation of FAF-BSA on MIN6 cells. We used 0.1% FAF-BSA, because this concentration was shown to immediately terminate [Ca²⁺]_i oscillations in MIN6 cells. All materials were cleaned with methanol to avoid palmitic acid contamination, according to Yao et al. (29). SNs were subjected to chloroform/methanol extraction, followed by MS analysis. Palmitic acid and stearic acid (SA) were detected as the most prominent FAs in MIN6 cell SNs (Fig. 4A), which supports the hypothesis that FAs serve as endogenous signaling factors. Incubation of MIN6 cells with 0.1% FAF-BSA increased the detected levels of oleic, elaidic, and SA 38-, 57-, and 29-fold (Fig. 4A). The occurrence of FAs in FAF-BSA preloaded on MIN6 cells was similar to FAs obtained from MIN6 cell extracts. In cell extracts, stearic, palmitic, arachidonic, oleic, and elaidic acid were the most abundant (Fig. 4B). These data support the hypothesis that FAF-BSA treatment is an effective approach for capturing FAs from MIN6 cells. As expected from [Ca²⁺]_i imaging, conventional BSA was found to be loaded with considerable amounts of FAs even before incubation on MIN6 cells, explaining its reduced capacity to capture FAs from MIN6 cell SNs (Fig. 4C).

FA levels in the SNs of MIN6 cells, as revealed by MS, were confirmed by a fluorescence displacement assay, based on the environmentally sensitive fluorophore NRBA (38). Fluorescence emission from BSA-bound NRBA dropped in the presence of MIN6 cell SNs, depending on the preincubation time of the SNs on MIN6 cells (Fig. 5A). This detected incubation time dependence underlines the important role of FAs as endogenous signaling factors for MIN6 cells.

The previously described results indicated that lowering endogenous FA levels was the major cause of the observed termination of [Ca²⁺]_i oscillations and the depletion of insulin secretion in FAF-BSA–treated MIN6 cells and mouse primary β-cells. In order to verify whether replenishing extracellular FA pools would restart [Ca²⁺]_i oscillations, we used sulfo-caged (Scg) FAs bearing a photolabile negatively charged caging group to allow quantitative delivery and localization of Scg-FA exclusively at the plasma membrane (PM) (Fig. 5D). Based on our findings from MS (Fig. 4) and our previous results (16), we applied Scg-SA and Scg-OA. Because of the fluorescence of the coumarin cage, BSA-mediated depletion of Scg-SA from MIN6 cells could be directly monitored by fluorescence microscopy. The addition of FAF-BSA to Scg-SA–loaded MIN6 cells significantly diminished the fluorescent signal at the PM within seconds (Fig. 5C and D). Furthermore, the fluorescence signal inversely correlated with the applied FAF-BSA concentration, indicating that the Scg-FA was efficiently extracted by FAF-BSA from PMs (Fig. 5B), leading to the termination of [Ca²⁺]_i oscillations. In line with our previous work, subsequent illumination of Scg-FA–loaded MIN6 cells with ultraviolet (UV) light permitted the release of FA with good spatiotemporal control (16). When applied to MIN6 cells and mouse primary β-cells, pretreated with FAF-BSA (0.1%), UV-mediated liberation of SA from Scg-SA (200 μmol/L) (Fig. 6Ai, Bi, Aii, Bii, C, and D) resulted in immediate resumption of [Ca²⁺]_i oscillations. Comparable effects were observed for Scg-OA (Supplementary Fig. 17). Less controlled replenishment of FAs by the simple addition of OA (200 μmol/L) to FAF-BSA (0.1%)–treated MIN6 cells and mouse primary β-cells also recovered [Ca²⁺]_i oscillations (Fig. 6Aiii, Biii, Aiv, Biv, E, and F). These observations indicate that endogenous FAs not only are essential, but provide a sufficient signal to induce [Ca²⁺]_i oscillations in glucose-stimulated β-cells.

As an alternative way to artificially elevate extracellular FA levels after FAF-BSA treatment, we used recombinant lipases. Phospholipase A₂ (PLA₂) is known to liberate FAs from membrane lipids by the hydrolysis of zwitterionic glycerophospholipids (39). Upon the addition of recombinant PLA₂ (10 units), [Ca²⁺]_i oscillations in MIN6 cells and mouse primary β-cells increased. The induced oscillations were terminated by the addition of 1% FAF-BSA (Fig. 7Ai–iii, Bi–iii, D, and E). Similar effects were observed for recombinant lipoprotein lipase (500 units), an enzyme that hydrolyzes triglycerides to FAs and glycerol (40) (Fig. 5Aiv, Biv, and F). Removal of PLA₂ from MIN6 cells via buffer exchange diminished induced [Ca²⁺]_i oscillations (Fig. 7Aii). The presence of recombinant PLA₂ on MIN6 cells increased insulin levels more than sevenfold (Fig. 7C). The inhibition of endogenous PLA₂ by methyl arachidonyl fluorophosphate (MAF) (10 μmol/L) or bromoenol lactone (BEL) (10 μmol/L) decreased and eventually terminated [Ca²⁺]_i oscillations in MIN6 cells and mouse primary β-cells (Fig. 7Av–vii, Bv–vii, G, and H) and insulin secretion from MIN6 cells (Fig. 7C). Our results demonstrate that elevated FA levels stimulate [Ca²⁺]_i oscillations and insulin secretion, independent of the origin of FAs. This effect was counteracted by the buffering capacity of albumins.

Previous studies in the field have focused on exogenous FAs as major secretagogues for nutrient-modulated release of adequate amounts of insulin (41,42). To the best of our knowledge, the essential role of endogenous FAs in intercellular pancreatic signaling for insulin secretion has not been demonstrated so far. To address this topic, we chose the insulinoma cell line MIN6 as a highly relevant model system of pancreatic β-cells. As has been previously shown, MIN6 cells and mouse primary β-cells exhibit very similar signaling patterns for the regulation of insulin secretion (30,43). Key results obtained from MIN6 cells were validated in mouse primary β-cells. Monitoring intracellular [Ca²⁺]_i oscillations allowed the sensitive assessment of individual β-cell responses on a single-cell level, while ELISA tests served for the quantification of bulk changes in insulin secretion in response to FAF-BSA treatment and variation of FA levels.

The objectives of this study were to examine the essential roles of endogenous FA signaling in basal β-cell activity and insulin secretion. To resemble physiological conditions, we applied serum albumin in FA-stripped and FA-loaded forms to MIN6 cells and mouse primary β-cells. Albumins are well known to bind FAs (44). We found that stringent washing of MIN6 cells and mouse primary β-cells or the application of serum albumins immediately terminated [Ca²⁺]_i oscillations and diminished insulin secretion. Hence, FAs seemed likely to belong to the secreted components that activate β-cells and are readily captured by FA-stripped serum albumin from PMs. This suggested that there are adequate FAs in PMs, which was supported by MS analysis. Compared with conventional BSA, the FA content of FAF-BSA was found to be ∼30 times lower, which indicated that FAF-BSA was significantly depleted in FAs and had a considerable capacity for the withdrawal of FAs from PMs. We observed that FAF-BSA preferentially binds saturated and monounsaturated long-chain FAs (C16–C20/22), such as OA, elaidic acid, SA, and palmitic acid. This is in line with reports in the literature, according to which FAF-BSA preferentially binds medium- and long-chain FAs (44,45). Indeed, the incubation of FAF-BSA on MIN6 cells led to a threefold increase in the levels of BSA-bound SA. In contrast, the levels of BSA-bound OA and elaidic acid increased by a factor of ∼18. Decreased [Ca²⁺]_i oscillations in MIN6 cells incubated with FAF-BSA swiftly recovered after uncaging of Scg-SA and Scg-OA or by the addition of OA. This indicates that [Ca²⁺]_i do not depend on a particular long-chain FA species, but can be induced by different long-chain FAs. Despite the fact that fluorescent Scg-FAs are not identical to natural FAs, we showed that they are readily extracted from the PM by the addition of FAF-BSA in different concentrations. In line with this result, [Ca²⁺]_i oscillations were stimulated by serum albumin–mediated off-loading of FAs, underlining their essential regulating role for β-cell activity and their physiological relevance under different nutritional conditions. Although the addition of FA-stripped serum albumin immediately terminated [Ca²⁺]_i oscillations, the effect was less dramatic in the presence of competing ligands, such as l-Trp or Ibu. FAs were shown to bind to albumin pockets I–III. Interestingly, site II has also been described as a discrete binding site for certain drugs and Trp (35,46,47). The occupation of albumin sites by l-Trp, Ibu, or preliminary FA loading (for conventional BSA, as shown by MS) might have reduced the FA loading capacity, leading to the weaker effects on [Ca²⁺]_i oscillations. Out-competition or cooperative effects at albumin binding sites provide an explanation for the observed stimulation of [Ca²⁺]_i oscillations in competition experiments, when Trp or Ibu was added on top of preloaded serum albumin on MIN6 cells.

Our combined results show that the presence of FAs is a prerequisite for β-cell stimulation by glucose and that insulin secretion relies entirely on the presence and maintenance of a basal FA level. The very small and confined volumes of the extracellular space of β-cells, as shown in electron microscopy data (48), might guarantee effective endogenous signaling, since only very low amounts of extracellular signaling factors might be necessary.

To demonstrate that the regulation of solely endogenous FA levels is sufficient to enhance [Ca²⁺]_i oscillations in glucose-stimulated β-cells, we artificially elevated FA levels by adding the recombinant lipases PLA₂ or lipoprotein lipase, which immediately translated into increased [Ca²⁺]_i oscillations. Further support came from the inhibition of endogenous PLA₂ using MAF and BEL, respectively, which decreased [Ca²⁺]_i oscillations and insulin levels. As lipase inhibition reduced [Ca²⁺]_i oscillations, endogenous PLA₂ activity seemed to be essential for driving cell responses to glucose. The results further suggest that both external and PM-localized FA pools are essential for β-cell activity, independent of the FA source.

To determine the target of FAs in β-cells, we used the GPR40 antagonist GW1100. It showed an immediate effect on the recovery of [Ca²⁺]_i oscillations both when spiked into BSA that was preloaded on MIN6 cells and when added to MIN6 cells. FA-mediated stimulation of GPR40 has been described to stimulate the Gαq-phospholipase C signaling pathway, resulting in an increase of intracellular inositol triphosphate levels and Ca²⁺ release from endoplasmic reticulum stores. This, in the presence of glucose, leads to the stimulation of L-type calcium channels and insulin release (13). FAs were also described to reduce the voltage-gated K⁺ current in β-cells through GPR40, along with protein kinase A activation, leading to increased [Ca²⁺]_i and insulin secretion (49). Withdrawal of FAs by the application of FAF-BSA is suspected to interrupt GPR40-mediated activation of L-type calcium channels and GPR40-mediated reduction of K⁺ currents and therefore to diminish overall [Ca²⁺]_i levels in β-cells, in accordance with the findings of Schnell et al. (12). However, the rapid termination, the fast recovery of β-cell activity after addition and replacement of FAF-BSA, and the results from our washing experiments point to additional targets other than GPR40, such as ion channels. Indeed, our recent work on photoswitchable FAs demonstrated a direct effect of FAs on the activity of voltage-activated and K_ATP channels, and subsequently on glucose-stimulated [Ca²⁺]_i oscillations (50). The effect of FAs on ion channels has also been reviewed (49,51).

In conclusion, our combined results imply that intact endogenous signaling is not only important for the synchronization of β-cell populations within pancreatic islets for pulsatile insulin release, but also for the activity at the single β-cell level. Our findings have implications for the general understanding of the regulation of β-cell activity and insulin secretion. In future studies, the impact of drugs that have an affinity for serum albumin on insulin secretion as well as the nutritional state of the organism should be considered. Displacement of anti-inflammatory drugs by FAs and vice versa from albumin binding sites, particularly compounds with low free fractions in plasma, may also be of pharmacokinetic significance. Nutrition- or drug-induced changes of FA levels in blood might have direct effects on insulin secretion.

Acknowledgments. The authors thank Rainer Müller (European Molecular Biology Laboratory [EMBL]) and Felix Hoevelmann (EMBL) for providing the Scg-SA and Scg-OA and Madeleine Schultz (EMBL) for critical reading of the manuscript. The authors also thank Alexander Aulehla (EMBL) for providing mice for islet isolation.

Funding. This research was supported by Deutsche Forschungsgemeinschaft Transregios 83 and 186 (C.S.). D.A.Y. received an Institute of Organic Chemistry and Biochemistry installation grant.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. S.H. designed the research; performed the experiments; analyzed the data; and wrote, reviewed, and edited the manuscript. K.K. performed the experiments, analyzed the data, and reviewed and edited the manuscript. P.P. analyzed the data and reviewed and edited the manuscript. D.A.Y. and C.S. designed the research and wrote, reviewed, and edited the manuscript. C.S. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.