Chronic Palmitate Exposure Inhibits Insulin Secretion by Dissociation of Ca2+ Channels from Secretory Granules

1Type 2 diabetes mellitus is characterized by insufficient insulin secretion in response to elevations of plasma glucose. This decreased insulin response shows a strong correlation with obesity and elevated plasma levels of free fatty acids (FFAs). Protracted exposure to FFAs results in marked suppression of glucose-induced insulin secretion both in vivo and in vitro. Although the latter effect has been extensively characterized, the cellular mechanisms remain enigmatic. The adverse effects of FFAs on β cell function, often referred to as “lipotoxicity”, have been postulated to culminate in β cell death. Indeed, experiments on human cells in vitro demonstrated marked apoptosis after just 24 hr exposure to elevated FFAs and glucose. Apoptosis of β cells has been attributed to both oxidative stress and ER stress as well as other pathways. However, other studies have failed to find evidence of any significant apoptosis following long-term exposure to FFAs. These discrepancies have been attributed to differences in the levels of unbound FFAs or islet viability.

2In healthy β cells, elevation of glucose produces (via its metabolism) an increased cytoplasmic ATP/ADP ratio and closure of ATP-regulated K+ channels (KATP channels). This causes membrane depolarization and the opening of voltage-dependent Ca2+ channels. The resultant increase in [Ca2+]i triggers the exocytosis of insulin-containing granules. This secretory response consists of a rapid but transient first phase followed a slowly developing second phase. Exocytosis during brief (action potential-like) depolarizations primarily involves granules belonging to a readily releasable pool (RRP) in close proximity to the voltage-gated Ca2+ channels. In a previous study, we reported that inhibition of glucose-stimulated insulin secretion by long-term exposure to the FFAs oleate and palmitate occurred without any signs of increased β cell death, reduced insulin synthesis, impaired glucose metabolism, KATP channel regulation, or Ca2+ signaling. It was therefore proposed that the perturbation occurs at the level of exocytosis. Here we have examined this possibility in greater detail by measurements of lipid composition, membrane fluidity, imaging of submembrane [Ca2+]i, and exocytosis.

3The effects of long-term palmitate treatment on the amounts of mRNA encoding key proteins involved in granule priming and exocytosis were evaluated by quantitative PCR (qPCR; see Table S1 available online). No major changes in gene transcription were observed that would account for the strong inhibition of insulin secretion observed.

4Changes in the lipid composition have been shown to affect the energy requirements of membrane fusion. We therefore compared the phospholipid (PL) and triglyceride (TG) composition of palmitate-treated and control islets. The palmitate content of the phospholipid fraction increased by ∼3–4 mol%, and stearate content correspondingly decreased in islet exposed to palmitate (see definition in the Experimental Procedures) (Figure 1A). These changes are much larger than those observed in erythrocytes from animals fed a high-fat diet. The overall ratio of saturated to unsaturated fatty acids in the phospholipid fraction remained unaltered when comparing control and palmitate-treated groups (0.88 ± 0.03 and 0.81 ± 0.03; p < 0.13; n = 4). Exposure of islets to palmitate resulted in a >3-fold increase in overall TG levels (Figure 1B, inset). Whereas the palmitate content of the triglyceride fraction increased by ∼50%, the stearate content was reduced by one third. These findings are in good agreement with a previous study that measured total changes in fatty acid composition in MIN6 clonal cells.

5We also made use of uniformly labeled (U13C) palmitate to track the incorporation of exogenous palmitate into phospholipids and triglycerides. In islets exposed to exogenous 13C-labeled palmitate for 72 hr, a significant amount of 13C-labeled palmitate was converted into phospholipids (Figure 1C), as well as some expected deposition as TGs (Figure 1D). Thus, palmitate is not simply deposited inside the β cell but is actively processed (by both desaturation and elongation) and is converted into phospholipids, the major component of the plasma membrane. No incorporation of palmitate into phospholipid was observed during a 2 hr incubation (data not shown).

6Impaired insulin exocytosis may be secondary to changes in membrane fluidity resulting from the altered phospholipid composition. We therefore examined membrane fluidity by analyzing lipid mobility by fluorescence recovery after photobleaching (FRAP). Pancreatic β cells isolated from islets incubated for 72 hr in the presence or absence of palmitate were labeled with the 14-carbon lipids fluorescently tagged with individual BODIPY fluorophores. No palmitate-induced changes in membrane fluidity were detected by this method (Figure S1).

7Exocytosis of insulin granules depends on localized increases in [Ca2+]i close to the voltage-gated Ca2+-channels. We examined the effects of palmitate exposure on β cell Ca2+ channel distribution by measurements of near-plasma membrane [Ca2+]i transients elicited by 50 ms depolarizations from −70 to 0 mV using the low-affinity Ca2+ indicator Oregon Green 6F (Figure 2A). This strategy was chosen as the Ca2+ channel density in mouse β cells is too low to allow immunocytochemical studies of Ca2+ channel distribution. Line scans obtained along the white lines are shown for one control cell and one palmitate-treated cell (Figure 2B). Whereas membrane depolarization produced discrete regions with high [Ca2+]i in control cells, the increase in [Ca2+] was more diffuse in palmitate-treated cells and did not attain as high concentrations. To quantify this, we calculated the coefficient of variation (CV) of the Ca2+ signal in the entire cell footprint; a greater spatial heterogeneity will produce a higher CV. This analysis revealed that depolarization-evoked [Ca2+]i (labeled “0”) is significantly less compartmentalized in palmitate-treated β cells than in control cells (Figure 2C). However, resting [Ca2+]i prior to depolarization (labeled “−70”) was not different between control and palmitate-treated cells. The latter finding argues that the observed effect on depolarization-evoked Ca2+ influx cannot be attributed to differences in plasma membrane adherence to the coverslip or dye infusion. We ascertained that the amount of Ca2+-sensitive dye infused into the cell did not differ between the two conditions (data not shown; p = 0.35). The redistribution of [Ca2+]i was only observed in response to long-term palmitate exposure and in β cells that were first cultured under control conditions for 72 hr and then exposed to palmitate for 2 hr, the measured CV remained the same as in control cells both at −70 and 0 mV.

8The amplitude of the whole-cell Ca2+ current is not affected by exposure to palmitate. Likewise, there is no change in the relative contribution of L-type Ca2+ channels; the L-type Ca2+ channel blocker isradipine (10 μM) reduced the whole-cell Ca2+ current by 55% ± 13% (n = 4) and 57% ± 10% (n = 4) in control and palmitate-treated cells, respectively. Thus, the observed changes in Ca2+ entry are likely to reflect redistribution of Ca2+ channels rather than altered subtype expression/magnitude of the Ca2+ current. This conclusion is in agreement with the qPCR data (Table S1).

9In separate experiments, we analyzed the distribution of syntaxin-1a in control and palmitate-treated cells; no difference was observed (CV = 0.17 ± 0.01 and 0.16 ± 0.01 in control and palmitate-treated cells, respectively; Figure S2). Palmitate was likewise without effect on the membrane distribution of SNAP-25 (data not shown).

10What are the effects of Ca2+ channel redistribution on exocytosis? Figure 3A shows increases in membrane capacitance (ΔCm) measurements from control and palmitate-treated cells when exocytosis was elicited by progressively longer (25–500 ms) depolarizations from −70 to 0 mV. The inset compares the responses in control cells (black) and palmitate-treated cells (red) during 50 and 250 ms depolarizations. Whereas a substantial (>50%–80%) decrease was observed in palmitate-treated cells during short (<100 ms) depolarizations, no significant inhibition was detected during depolarizations ≥250 ms.

11We next optically measured exocytosis of individual granules in parallel with measurements of submembrane [Ca2+]i and ΔCm in β cells expressing a fluorescently tagged variant of the granule protein IAPP (IAPP-mCherry; Figure S3). Exocytosis was elicited using a pulse protocol simulating glucose-induced β cell electrical activity that consisted of ten 40 ms depolarizations from −70 to −10 mV. The total ΔCm evoked by the train was reduced by 55% (Figure 3B), from 118 ± 39 fF (n = 7) under control conditions to 48 ± 18 fF (n = 8; p < 0.05) after palmitate exposure. This inhibition of exocytosis occurred without any changes in overall [Ca2+]i (compare red and black traces in the lower panel of Figure 3B (p > 0.5). Inhibition of exocytosis was also detected in the optical measurements. In control cells, 5% of the proximal (near-membrane) granules underwent exocytosis and released their fluorescent cargo during the train as compared to 1.5% in cells cultured in the presence of palmitate (Figure 3C). By contrast, no palmitate-induced inhibition of exocytosis was detectable when stimulation was elicited by depolarization of a supraphysiological duration (ten 500 ms depolarizations to zero mV; Figure 3D). Under these conditions, the total ΔCm averaged 425 ± 61 fF (n = 7) and 432 ± 82 fF (n = 8) in control cells and palmitate-treated cells, respectively. In the fluorescent imaging experiments, 20%–25% of the docked granules were released during the train in both control and palmitate-treated cells (Figure 3D). The suppression of exocytosis by action potential-like stimulation in palmitate-treated cells cannot be accounted for by a dearth of available granules, which averaged ∼0.7 granules/μm2 in both control and palmitate-treated cells (Figure 3E). The latter value is in excellent agreement with that reported by others in human β cells but different from that reported previously in rat β cells.

12Figure 3F compares the exit of the IAPP-mCherry cargo in both control (top) and palmitate-treated cells (bottom). In both cases, granule emptying is complete within four to five frames (<250 ms). Figure 3G depicts the average time courses of granule emptying in control (black) and palmitate-treated cells. The initial increase reflects the discharge of fluorescent peptide cargo into the space between the coverslip and the cell membrane, which then rapidly decays to background values as the fluorescent proteins diffuse away from the cell. Thus, FFA pretreatment does not, contrary to what was previously hypothesized, inhibit insulin secretion by interfering with the emptying of the granules.

13Granules within secretory cells belong to different functional pools. In addition to a RRP of granules situated close to the Ca2+ channels, β cells possess a slowly releasable pool (SRP) of granules that are molecularly similar to the RRP but geographically separated from sites of Ca2+ entry. To test the hypothesis that chronic palmitate exposure interferes with the coupling between the Ca2+ channels and granules undergoing rapid exocytosis, we simultaneously measured the rate of exocytosis and changes in submembrane [Ca2+]i during an 800 ms depolarization (see Movie S1). Circular regions with a diameter of 1 μm were centered on each granule that underwent exocytosis during this period (white circles, Figures 4A and 4B). These regions were then used to measure the local increases in [Ca2+]i for each exocytosing granule. The [Ca2+]i changes (insets) in these regions were compared to that of adjacent stationary granules (purple circles). In control cells, granules that underwent exocytosis had an ∼45% larger increase in local [Ca2+]i than those that were not released (p < 0.05, Figure 4C). No such difference existed for palmitate-treated cells (Figure 4D). We analyzed the consequence of this on the kinetics of exocytosis (Figure 4E). In control cells, 2% of the proximal granules discharged their peptide cargo during the first 100 ms of the depolarization, compared to <0.3% in palmitate-treated cells (an inhibition of >85%). At later times, however, exocytosis proceeded in parallel in control and palmitate-treated cells.

14The data of Figure 4E support the notion that control cells contain both RRP (close to Ca2+ channels) and SRP (further away from the Ca2+ channels) granules, whereas palmitate-treated cells contain only SRP granules. Exocytosis of these two pools of granules should be differentially influenced by Ca2+ buffering; exocytosis of granules situated far away from the Ca2+ channels should be inhibited by the slow Ca2+ buffer EGTA, whereas granules residing in close proximity to the Ca2+ channels should be less affected. Figure 5A compares [Ca2+]i increases elicited by a 50 (Figure 5Aa) and a 500 ms (Figure 5Ab) depolarization (0 mV) in a β cell dialyzed with 50 μM EGTA. Under these conditions, a longer depolarization expands [Ca2+]i microdomains. Thus, there is a progressive increase in submembrane [Ca2+]i (measured as ΔF/F) beyond that attained during the first 50 ms that reflects the intracellular diffusion of Ca2+. By contrast, the microdomains did not expand with pulse duration when the cells were infused with 10 mM EGTA (Figures 5Ca and 5Cb). Consequently, the increase in ΔF/F was maximal at 50 ms (Figure 5D).

15In cells containing 50 μM EGTA, exocytosis evoked by 500 ms depolarizations was unaffected by palmitate, whereas the response to the 50 ms depolarization was reduced by 68% after palmitate treatment (Figure 5E; same data as in Figure 3A). When intracellular Ca2+ diffusion was restricted by including 10 mM EGTA in the intracellular solution, the amplitude of the response elicited by 500 ms depolarizations was reduced and became comparable to that evoked by a 50 ms depolarization at low intracellular buffering. Under these conditions, exocytosis elicited by the 500 ms depolarization was inhibited by 70% in palmitate-treated cells (inset). We conclude that the time- and diffusion-dependent expansion of the [Ca2+]i domain explains why palmitate does not inhibit exocytosis elicited by longer depolarizations at low intracellular Ca2+ buffering.

16We tested this idea by comparing exocytosis evoked by brief voltage-clamp pulses with that elicited by a global increase in [Ca2+]i produced by photoliberation of “caged” Ca2+ from Ca2+—NP-EGTA in control and palmitate-treated cells (Figure 6A). Cells were first stimulated with five 40 ms depolarizations to 0 mV. In control cells, the first two to three pulses of the train evoked significant exocytosis, whereas the responses in palmitate-treated cells were almost abolished. Some recovery of exocytosis was observed in palmitate-treated cells during the fifth pulse, presumably reflecting accumulation of [Ca2+]i within the cell (cf. Figure 3B). In both control and palmitate-treated cells, photoliberation of caged Ca2+ elicited robust exocytotic responses. Figure 6B summarizes the average responses for each pulse of the train, the total increase evoked by the train (Σtrain) and the response to photoreleasing caged Ca2+ (flash). During the first three pulses, exocytosis was inhibited by >65%, whereas the responses elicited by uncaging of [Ca2+]i were almost identical. The increases in ΔCm elicited by photorelease could be reasonably fit by double exponentials with time constants of 23 ± 5 ms and 497 ± 213 ms (n = 10) in control cells and 26 ± 4 ms and 920 ± 290 ms (n = 11) in palmitate-treated cells. The corresponding increases in ΔCm for the rapid and slow components of exocytosis averaged 85 ± 5 fF and 130 ± 38 fF in control cells and 91 ± 3 and 112 ± 39 fF in palmitate-treated cells. We conclude that pretreatment with palmitate inhibits insulin exocytosis by affecting the location of Ca2+ influx rather than the Ca2+ sensitivity of exocytosis.

17The capacitance measurements and the optical imaging of exocytosis in single β cells suggest that palmitate treatment selectively interferes with rapid exocytosis by disrupting the tight functional association between the secretory granules and the Ca2+ channels, whereas that elicited by longer stimuli is unaffected. If the suppression of glucose-evoked insulin secretion from intact pancreatic islets also results from this effect, then experimental maneuvers that expand the boundaries of [Ca2+]i microdomains should rescue insulin secretion. To test this idea, we used the K+ channel blocker tetraethylammonium (TEA), which markedly prolongs the β cell action potential. In intact islets cultured for 72 hr, the duration of the action potentials observed in the presence of 20 mM glucose alone and following addition of 20 mM TEA averaged 31 ± 1 (n = 3) and 326 ± 17 ms (n = 4), respectively (Figure 7A). In agreement with previous studies, exposure of mouse islets to palmitate for 72 hr reduced insulin secretion in response to 20 mM glucose by ∼60%. TEA (20 mM) stimulated insulin secretion in both control and palmitate-treated islets (Figure 7B). Importantly, TEA also corrected the secretion defect induced by palmitate; in the presence of the K+ channel blocker, insulin secretion in control and palmitate-treated islets was not different. The reduction of glucose-stimulated insulin secretion could not be attributed to a palmitate-induced decrease in insulin content, which averaged 107% ± 8% (n = 12) of its control value. Basal secretion in 1 mM glucose was marginally stimulated in palmitate-treated cells compared to control islets, and TEA had only a weak effect under these conditions. Long-term exposure to palmitate also reduced glucose-induced insulin secretion from human islets (Figure 7C) without affecting insulin content (104% ± 16% of control islets; n = 6). This inhibition was also counteracted by TEA.

18We finally investigated the effects of palmitate exposure on the kinetics of glucose-induced insulin secretion from mouse islets (Figure 7D). In control islets, glucose (20 mM) produced a 30-fold stimulation of insulin secretion that peaked after 5 min and later stabilized at a somewhat lower plateau (20-fold higher than basal). In islets that had been treated with palmitate, glucose-stimulated insulin secretion was inhibited by ∼60% throughout the stimulation period.

19Here we show that the inhibitory effect of long-term exposure to palmitate on insulin secretion is associated with alterations in the microdomains of Ca2+ entry in β cells. This leads to a selective inhibition of exocytosis triggered by brief (<100 ms) action potential-like depolarizations. Although the β cell contains a large number of release-competent granules, only those in close proximity to the voltage-gated Ca2+ channels are released during β cell electrical activity. We have previously proposed that Ca2+ channels are physically tethered to the insulin granules. Thanks to this arrangement, a subset of release-competent granules are exposed to the very high [Ca2+]i that occurs in the close vicinity of the Ca2+ channel. Our data suggest that palmitate causes disassembly of the tight complexes of Ca2+ channels and weakens the Ca2+ microdomains that form and trigger insulin exocytosis. Under control conditions, the close proximity of Ca2+ channels and secretory granules ensures that exocytosis is triggered by the localized increases in [Ca2+]i evoked by an action potential (Figure S4A). Following palmitate treatment, however, the increase in [Ca2+]i does not reach sufficiently high concentrations close to the secretory granule to produce exocytosis. This would explain the strong inhibition of exocytosis during action potential-like stimulation as well as the much greater inhibition of exocytosis caused by strong Ca2+ buffering (Figure 5E). This scenario also accounts for the failure of tolbutamide to correct the secretion defect. This is because tolbutamide triggers brief action potentials of the same type as those occurring in response to glucose.

20We emphasize that insulin secretion from palmitate-treated β cells is not inhibited as a consequence of reduced release competence (Figure 6C). The fact that exocytosis during 250 ms depolarizations is unaffected is also consistent with the idea that FFAs suppress glucose-induced insulin secretion by disruption of the tight coupling between Ca2+ channels and the secretory granules. During such long depolarizations, [Ca2+]i equilibrates within the cell and exocytosis will no longer be restricted to the immediate vicinity of the Ca2+ channels. This accounts for the ability of TEA, which leads to a marked prolongation of the β cell action potential, to restore insulin secretion in palmitate-treated islets to control values.

21It may seem surprising, given that rapid exocytosis is selectively affected in the capacitance measurements and optical imaging experiments, that first-phase and second-phase insulin secretion are similarly affected (Figure 7D). Accumulating evidence suggests that first-phase insulin secretion reflects exocytosis of granules residing in close proximity to the voltage-gated L-type Ca2+ channels (RRP) (; but see ). Once these granules from the RRP have undergone exocytosis, they must be replenished by physical mobilization of new granules to the Ca2+ channels and/or lateral movement of the Ca2+ channels to docked granules (Figure S4B). Importantly, both first- and second-phase insulin secretion are triggered by the Ca2+ influx during the brief action potentials, and they occur at a frequency too low to allow major summation of the individual [Ca2+]i transients (Figure 3B). First-phase insulin secretion is normally larger than the second phase because the rate at which RRP is replenished is lower than the maximum secretory capacity.

22We emphasize that our data do not exclude the possibility that other adverse effects (including increased cell death and mitochondrial uncoupling) ultimately develop during protracted hyperlipidemia and contribute to the impaired insulin release. It is pertinent, however, that the mechanism we describe here, that results in reduced Ca2+-dependent exocytosis, develops within 72 hr and occurs before the other functional abnormalities are detectable.

23What causes the Ca2+ channels to dissociate from the secretory granules? Our data indicate that this is not a consequence of altered membrane fluidity. Previous work suggests that the tethering of the secretory granules to the Ca2+ channels involves the interaction of the synprint peptide (the cytoplasmic domain connecting the second and third homologous domain of the L-type Ca2+ channel) with the exocytotic core complex consisting of syntaxin-1, synaptotagmin, and SNAP-25. In mouse insulinoma MIN6 cells, long-term exposure to palmitate alters the expression of several proteins involved in β cell exocytosis. Mice lacking syntaxin-1a exhibit a reduced number of docked granules and loss of first-phase insulin secretion but normal second-phase release, but our own data did not reveal any abnormalities in the expression or distribution of syntaxin-1 (Figure S3) or SNAP-25. There were likewise no major changes in the transcription of other exocytotic proteins/Ca2+ channel subunits that would explain the functional data (Table S1). Clearly, mRNA levels need not equate to protein levels. Chronic exposure to palmitate results in increased expression of the microRNA miR34a. MicroRNAs act by inhibition of translation and may consequently affect protein expression even in the absence of any changes in gene transcription. It should be noted, however, that long-term exposure to palmitate has no effect on rapid exocytosis elicited by photorelease of caged Ca2+ (Figure 6). This makes it less likely that the suppression of insulin secretion reflects reduced expression of key exocytotic proteins. Alternatively, chemical modification of these proteins, via changes in acylation or palmitoylation resulting from the altered phospoholipid composition of the plasma membrane (Figure 1), may interfere with the stability of the granule/Ca2+ channel complexes. This will be challenging to examine, as the pharmacological inhibitors of acylation have numerous side effects including inhibition of glucose metabolism.

24What is the physiological rationale for the ability of lipids to interfere with the association of secretory granules and Ca2+ channels in β cells? Importantly, short-term (2 hr) exposure to palmitate (which stimulates insulin secretion; ) did not affect Ca2+ channel clustering (Figure 2C). Thus, the effect is only observed during protracted exposure. Long-term elevation of plasma FFAs occurs physiologically during fasting, when the plasma FFA concentration may increase to ∼1.5 mM (review, ). In vivo recordings of β cell electrical activity have shown that the increased plasma FFA associated with fasting leads to stimulation of β cell action potential firing but suppression of insulin secretion. Thus, the normal correlation between β cell electrical activity and insulin secretion is lost when plasma FFA levels are chronically elevated. Our data suggest that this results from a lipid-induced dissociation of Ca2+ channels from the secretory granules. It is tempting to speculate that this represents an evolutionarily conserved catabolic mechanism that suppresses insulin secretion. If it were not for this mechanism, elevation of FFAs would stimulate insulin secretion at plasma glucose levels found during protracted fuel withdrawal: 3.5 mM in man and 4.5 mM in mice. However, this mechanism may backfire, at least in genetically predisposed individuals, when plasma lipids become chronically elevated as a consequence of a high-fat diet or impaired insulin secretion. Once this has occurred, it is easy to envisage a vicious cycle resulting in progressive deterioration of insulin secretion and elevation of plasma FFAs culminating in glucose intolerance and overt diabetes.

25All animal experiments were carried out on standard laboratory CD-1 mice. Human islets were obtained from the Oxford Islet Isolation Centre. Previously described methods were used for preparation of islets and single β cells, electrophysiology, and insulin release measurements. Palmitate was prepared in a solution bound to fatty acid-free bovine serum albumin (BSA). Palmitate was dissolved in 95% ethanol, and stoichiometric amounts of NaOH were added. The solution was dried using nitrogen gas; water was subsequently added and the solution heated to create a hot soap. The solution was stirred and BSA was added to a final concentration of 10% w/v, to create a 10x stock solution. The pH was set to 7.4 (NaOH). Islets were cultured in RPMI 1640 together with 0.5 mM palmitate in the presence of 1% BSA for 72 hr. The resulting free FFA concentration was estimated to be 26 nM. Control islets were cultured in the presence of 1% BSA alone.

26Islet perifusions were performed using an in-house-designed system with four temperature-controlled chambers. Fifty size-matched islets were transferred onto a 1 μm nylon mesh cut to the size of the chamber, which was filled and perfused continuously with extracellular medium containing (in mM) 120 NaCl, 4.7 KCl, 25 NaHCO3, 1.2 KH2PO4, 1.2 MgSO4, 10 HEPES (pH 7.4 with NaOH), 2.5 CaCl2, 0.5 mg/ml BSA, and D-glucose as indicated. The chamber temperature was kept at 37°C. Samples were collected in 96-deep-well format plates every 30 s and kept at −20°C pending insulin measurements.

27Expression of IAPP-mCherry was achieved by adenoviral infection using an in-house vector modified to express the mCherry fluorophore. Homologous recombination of the vector into the adenovirus was performed by Vector BioLabs (Philadelphia, USA).

28The standard extracellular medium for imaging and electrophysiological measurements consisted of (in mM) 138 NaCl, 5.6 KCl, 1.2 MgCl2, 2.6 CaCl2, 5 HEPES, and 5 D-glucose at pH 7.4 (adjusted with NaOH). The intracellular solution consisted of (in mM) 125 Cs-glutamate, 10 CsCl, 10 NaCl, 1 MgCl2, 0.05 EGTA, 3 Mg-ATP, 0.1 cAMP, 5 HEPES (pH 7.2 using CsOH), and 0.03 Oregon Green 6F. In Figures 2, 5D, and 5E, intracellular Ca2+ buffering was increased by adding 10 mM EGTA (with iso-osmolar reduction of 10 Cs-glutamate).

29Photorelease of “caged” Ca2+ was initiated using a JML-C2 flashlamp (Rapp Optoelektronik GmbH, Hamburg, Germany). The pipette solution consisted of (mM) 110 K-glutamate, 10 KCl, 20 NaCl, 1 MgCl2, 25 HEPES (pH 7.1 with KOH), 3 MgATP, 3 NP-EGTA (Synaptic Systems, Goettingen, Germany), and 2 CaCl2. The low-affinity Ca2+ indicator Oregon Green 6F was included in this medium at a concentration of 0.03 mM.

30All solutions were osmotically balanced to between 290 and 310 mOsm. Experiments were performed at +32°C (electrophysiology) or +37°C (hormone release).

31Images were obtained with a customized Olympus IX-81 total internal reflection fluorescence (TIRF) microscope equipped with a 150×/1.45 Apo lens (Olympus, UK). The 488 nm line of an argon ion laser (Melles Griot, Germany) was used to excite the Ca2+ dye Oregon Green 6F (Invitrogen, Carlsbad, USA), and a 561 nm line of a steady-state diode laser was used to excite IAPP-mCherry. Two excitation ports with separate focal pathways were used to make independent adjustments of each laser. The emission pathway for both fluorescent reporters was imaged simultaneously with an image splitter (Dual View, Optical Insights, Santa Fe, NM, USA) creating side-by-side images of both emission wavelengths on the chip of a charge-coupled device (CCD) camera (Cascade II 512B, Roper Scientific, Trenton, NJ, USA). All images were acquired using CellˆR software (Olympus, UK) and later analyzed with MetaMorph software version 7.53 (Universal Imaging, Downingtown, PA, USA). Images were acquired at 20–30 Hz with 10 ms excitation exposure. To correct for alignment between the two channels, live images were adjusted by hand to align the emission pathway to overlap in both channels by imaging 200 nm tetraspeck beads (Molecular Probes, Eugene, OR). This correction was done on a daily basis.

32Total phospholipids and triglycerides were separated by spotting lipid extracts onto silica gel 60 (Merck) thin-layer chromatography plates using hexane:diethyl ether:acetic acid (85:15:1 vol/vol/vol) as the solvent. Lipid bands were visualized under UV light after spraying with 0.1% ANS (8-anilino-1-naphthalene sulphonic acid). Identification was verified using commercial standards. Total phospholipid and triglyceride bands were transferred into glass tubes and methylated at 80°C with 1.5% (triglycerides, 2 hr) or 6% H2SO4 (phospholipids, 12 hr) in methanol. Total phospholipid and triglyceride fatty acids were eluted into hexane. Separation and quantification of the fatty acid methyl esters (FAMEs) from islet total phospholipids and triglyceride were achieved using gas chromatography (Agilent 6890 GC, Agilent Technologies, UK) fitted with a 30 m × 0.53 mm (film thickness 1 μm) capillary column (RTX-Wax). Individual fatty acid peaks were identified by reference with known FAMEs. Fatty acid compositions (μmol/100 μmol total fatty acids; mol%) in these peaks were then determined.

33[U-13C] enrichment was determined by GC-mass spectrometry (GC-MS) using a 5890 GC coupled to a 5973N MSD (Agilent Technologies, UK). The GC was equipped with a 30 m × 0.25 mm (film thickness 0.25 μm) capillary column (rTX-5) and ions with mass-to-charge ratios (m/z) of 270 (palmitate; M+0) and 286 (i.e., palmitate uniformly labeled with 13C; M+16) were determined by selected ion monitoring. Dwell time was 100 ms. Tracer-tracee ratios (TTRs) for [U-13C]palmitate (M+16)/(M+0) were multiplied by the corresponding PL or TG fatty acid concentrations to give islet tracer concentrations.

34Measurements of the rate of exocytosis (Figure 4E) as well as the magnitude of release (Figures 3C and 3D) were compared between cells by normalizing to the percentage of released granules ([number released]/[number visible]). Granules were identified by an automatic Metamorph routine that identified all “spots” between 400 and 900 nm in diameter with brightness >15% above the local background of surrounding pixels. For the total granule count, large clusters of bright granules were ignored: these accounted for ∼8% of all granules in both control and palmitate-treated cells.

35For measurements of Ca2+ influx (Figure 2), the standard deviation (σ) and mean fluorescence (μ) were measured for the entire cell footprint, which was determined from the basal fluorescence of the Oregon Green 6F dye infusion. The CV, measured as the ratio between σ and μ, was determined by measuring the Ca2+ signal within the footprint during membrane depolarization after subtracting the average resting Ca2+ signal. The resting signal for each depolarization was calculated by averaging the four frames recorded prior to depolarizing the cell. The average background image thus generated was then subtracted from that obtained during depolarizations. Increases in [Ca2+]i (measured as ΔF) around exocytosing granules were measured within 1 μm diameter circular regions centered on the location of all identified exocytotic events within a cell (identified by the optical increase and dispersion of IAPP-mCherry fluorescence that occurred in a separate, but aligned, optical channel). These localized increases were then compared to increases of [Ca2+]i within adjacent, but nonoverlapping (>1.5 μm but <2 μm) regions around granules that did not undergo exocytosis. To correct for irregularities in the evanescent field and amount of membrane within a measured region, all measurements of [Ca2+]i increases (ΔF) were normalized to basal fluorescence (F) levels prior to depolarization for each region.

36Data are presented as mean values ± SEM for the indicated number of experiments (n). Statistical significance measurements were performed by Student's t test, with Bonferroni corrections in Table S1, and in Figures 1 and 7 an ANOVA test with Dunnet C post-hoc adjustments.