Optical control of GPR40 signalling in pancreatic β-cells

Fatty acids activate GPR40 and K+ channels to modulate β-cell function.


Introduction
2][3][4] Free fatty acids most oen consist of a long, unbranched carbon chain attached to a carboxyl headgroup, which is largely deprotonated and thus negatively charged at physiological pH. 5 They are amphiphilic molecules with diverse structures that vary in the chain length and the level of unsaturation.A number of transmembrane signalling proteins, including G proteincoupled receptors (GPCRs) such as GPR40, 6 are stimulated by free fatty acids, 7 resulting in a rise in the intracellular Ca 2+ concentration ([Ca 2+ ] i ) in insulin-secreting pancreatic b-cells through activation of phospholipase C. [8][9][10] Given the role of GPR40 in glucose homeostasis, synthetic agonists for these receptors such as Gw-9508 (ref.11 and 12) and TAK-875 (ref.13 and 14) have received signicant attention as potential treatments for type 2 diabetes mellitus. 15,16However, a phase III clinical trial for TAK-875 was recently terminated due to offtarget effects and toxicity concerns. 17,18lucose-stimulated insulin secretion (GSIS) relies on transport of glucose into the b-cell, followed by its metabolism to ATP.The resulting increase in the ATP/ADP ratio leads to Fig. 1 Glucose-stimulated insulin secretion (GSIS) from pancreatic bcells.Upon uptake into the pancreatic b-cell, glucose is metabolized into ATP.The rising ATP/ADP ratio inhibits K ATP which causes membrane depolarization and the opening of Ca v .The resulting increased [Ca 2+ ] i triggers the fusion of secretory granules and the release of insulin.K v channels work to repolarize the cell, generating oscillations in [Ca 2+ ] i .GPR40 stimulation also leads to increased [Ca 2+ ] i , further potentiating GSIS.
closure of ATP-sensitive K + channels (K ATP ) and subsequent membrane depolarization.This causes the opening of voltage activated L-type Ca 2+ channels (Ca v ) and an increase in [Ca 2+ ] i , driving exocytosis of insulin secretory granules. 19Subsequent activation of delayed rectier voltage-activated K + (K v ) channels leads to repolarization of the membrane, reduced Ca 2+ entry through Ca v channels and termination of insulin secretion (Fig. 1). 20This is complemented by the action of other messengers, including those stemming from GPCRs (so-called "amplifying" signals).Notably, the amplifying effects of GPR40 activation on insulin secretion remain elusive due to conicting results in different experimental conditions, 12,21 which could be attributed to effects of FAs at different targets.For example, fatty acids are known to directly affect various K + channels that are involved in modulation of the [Ca 2+ ] i oscillation frequency, 1,22,23 demonstrating their complex pharmacology and vital role in b-cell signalling.Therefore, a tool that could enable precise control over GPR40 signalling may be useful to better understand the effects of fatty acids, as well as specic agonists, on band other cell functions.This could lead to the development of novel therapeutics by delineating the receptor conformations required for biased signalling. 18,246][27][28][29] We also showed that photoswitchable diacylglycerols [30][31][32] affect b-cell [Ca 2+ ] i and insulin secretion.These diacylglycerols were constructed from a photoswitchable fatty acid (FAAzo) chain, however the pharmacology of these FAAzos alone remains largely unexplored.Given the sensitivity of GPR40 to unsaturated, and sometimes aryl-containing free fatty acid-like molecules, we hypothesized that the FAAzos themselves could enable optical control of this GPCR.Herein, we describe a novel approach towards the optical control of fatty acid/GPR40 signalling in b-cells.

Results and discussion
Although GPR40 is activated by long-chain fatty acids such as arachidonic or linoleic acid, 10 various aryl-containing carboxylic acids such as Gw-9508 are known to produce a similar effect (Fig. 2a). 3 We recognized that the benzyl-aniline moiety of Gw-9508 could be easily substituted by a phenyl diazene, and would afford a photoswitchable ligand with little disturbance to the overall size and structure of the drug.Therefore, we synthesized the azologue 33 of Gw-9508, FAAzo-10, using the Mills reaction aer nitroso formation in two steps and 45% overall yield (Fig. 2b).Similar to the other members of the FAAzo family, 30 FAAzo-10 behaved as a regular azobenzene and could be isomerized between its thermally stable trans-form to the cis-form with UV-A light (Fig. 2c).The process could be reversed by irradiation with blue light, and photoswitching could be repeated over many cycles.
We then characterized the effects of FAAzo-10 on GPR40 in HeLa cells using confocal uorescence microscopy and the genetically encoded uorescent [Ca 2+ ] i reporter R-GECO. 34hen transiently transfected with GPR40, a portion of cells displayed spontaneous [Ca 2+ ] i oscillations without the addition of any external stimuli (Fig. 3a and S1a †).Gw-9508 induced a GPR40-dependent increase in the rate and intensity of [Ca 2+ ] i oscillations, that was not affected by UV-A-irradiation (Fig. 3b and S1b †).In cells without GPR40, no response was observed (Fig. S1c and d †).Complementary to this result, the application of trans-FAAzo-10 (200 nM) stimulated a signicant increase in [Ca 2+ ] i in HeLa cells expressing GPR40 (Fig. 3c and d).On isomerization to cis with 375 nm irradiation, a sharp decrease in the [Ca 2+ ] i was observed.The effect was reversed and [Ca 2+ ] i increased on termination of the irradiation.In cells lacking GPR40, FAAzo-10 did not affect [Ca 2+ ] i (Fig. 3e and S1e †).We also evaluated the effect of FAAzo-4, which possesses a similar structure to FAAzo-10, but was not active at this low concentration (Fig. 3f).Histamine 35 (HIS, 10 mM) was used as a positive control and triggered a large increase in [Ca 2+ ] i , independent of GPR40 expression (Fig. 3 and S1 †).
To investigate the downstream effects of GPR40 activation, we expressed the uorescent diacylglycerol reporter C1-GFP, which translocates to the plasma membrane in response to increased diacylglycerol levels following PLC activation. 36Gw-9508 (200 nM) triggered C1-GFP translocation towards the plasma membrane, indicating activation of the GPCR (Fig. S1f and g †).On application of trans-FAAzo-10 (20 mM), we observed a similar effect on C1-GFP translocation.This could be reversed following isomerization to cis-FAAzo-10 with 375 nm irradiation, and translocation could be repeated over many cycles (Fig. 3g).These results demonstrate that oscillations in GPR40 activity and its downstream effectors (i.e.PLC, [Ca 2+ ] i and diacylglycerols) can be modulated with good temporal control.
Surprisingly, the effects induced by the FAAzos in HeLa cells did not diminish over time (Fig. 3), unlike those induced by the photoswitchable diacylglycerol PhoDAG-1, which decreased in magnitude over multiple UV-A pulses of the same length. 31To control for differences in cell loading, we applied the coumarinyl-ester of AA (cg-AA) to the HeLa cells. 6This uorescent fatty acid-derivative localizes predominantly at the inner cellular membranes. 31By monitoring the quenching of coumarin uorescence by the azobenzene of FAAzos, we demonstrated that this observed variance in activity was not due to variable FAAzo uptake by cells.Application of both FAAzos caused a rapid and large (>60%) decrease in coumarin uorescence (Fig. 3h), especially when compared to the quenching effect of PhoDAG-1 (<20%), which is known to remain trapped on the outer plasma membrane. 31A cellular lipid analysis by thin layer chromatography (TLC) conrmed only minor FAAzo metabolism in cells incubated with FAAzo-4 and FAAzo-10 (100 mM) for up to 1 h (Fig. S2 †).Together, these results demonstrate that the FAAzos are quickly taken up into cells, and only minimally metabolized over the timeframe of a typical imaging experiment.
A major advantage FAAzo-10 when compared to conventional agonists is the ability to modulate GPR40 activity with increased spatial precision.By illuminating only cells of interest, we were able to selectively control GPR40 activity without affecting signalling in neighbouring unilluminated cells (Fig. 4).This allows GPR40 activity to be controlled in a spatially dened manner in large patches of cells or complex tissues.
To evaluate the effects of FAAzo-10 on K + channels, we used whole-cell electrophysiology in dissociated mouse b-cells, which  express both K v and K ATP channels. 37,38K v channel conductance is a major determinant of the [Ca 2+ ] i oscillation frequency. 20ike AA 31 and Gw-9508 (Fig. 5a), trans-FAAzo-10 reduced K v channel conductance in the dark or under blue irradiation (Fig. 5b).On isomerization to cis-FAAzo-10, K v channel activity was restored to a level comparable with the vehicle controls (Fig. 5a).FAAzo-10 could be switched ON and OFF repeatedly, effectively allowing us to quickly mimic the wash-in and washout of Gw-9508 using only a UV-A/blue irradiation (Fig. 5c).Furthermore, we could also ne-tune the effect of FAAzo-10 with greater precision by scanning through different irradiation wavelengths.The K v conductance could be precisely controlled by gradually increasing the blocking effect of FAAzo-10 when scanning from UV-A to blue wavelengths.This was demonstrated by applying voltage ramps under 350-450 nm irradiation (Fig. 5d and e).
Gw-9508 has also been shown to potentiate K ATP channels in mouse b-cells. 11We measured the whole-cell K ATP current without extracellular glucose.IV-curves were measured between À110 and À50 mV to exclude any effect of the K v channels.Aer dialysis of the cytoplasm with intracellular buffer to reduce the ATP/ADP ratio, the K ATP current increased to a steady state (Fig. 6a and S3a †).In line with previous reports, Gw-9508 increased the K ATP conductance further (Fig. 6a and c).
Interestingly, trans-FAAzo-10 behaved differently, and reduced the K ATP conductance, while isomerization to cis-FAAzo-10 reversed the effect (Fig. 6b).Similar to the effects observed on K v channels, FAAzo-10 activity at K ATP could be netuned by altering the irradiation wavelength (Fig. 6d).Under blue irradiation, the K ATP current was reduced, while the blockade was reversed towards UV-A wavelengths.In control experiments, application of the sulfonylurea tolbutamide reduced the K ATP current signicantly (Fig. 6e and S3a †), and neither UV-A nor blue irradiation alone affected the K ATP conductance (Fig. 6e and S3 †).
Finally, we evaluated our photoswitchable ligands for their effects on intact pancreatic islets using confocal uorescence imaging.We employed the uorescent small-molecule [Ca 2+ ] i indicator Fluo-8 to monitor [Ca 2+ ] i oscillations stimulated by a high glucose concentration (11 mM).Similar to the application of Gw-9508 (Fig. 7a and b), application of trans-FAAzo-10 (20 mM) caused a marked increase in the [Ca 2+ ] i oscillation frequency (Fig. 7c and d).In line with the effects that would be expected from our results on GPR40, K v , and K ATP , isomerization to cis-FAAzo-10 with 365 nm irradiation reversed this effect entirely (Fig. 7c-f).Lower concentrations of FAAzo-10 (2.5 mM) did not affect oscillation frequency in either conguration (Fig. 7f).To exclude imaging artifacts, in particular uorescence quenching, the cells were treated with a methyl ester FAAzoderivative, FAAzo-5(OMe), which possesses an azobenzene photoswitch with similar spectral characteristics to FAAzo-10. 30AAzo-5(OMe) produced a small increase in the [Ca 2+ ] i oscillation frequency in either conguration (Fig. S4a-c †), as methyl esterication of the acid group abolished cis-activity.Although FAAzo-10 effectively increased [Ca 2+ ] i oscillations, we did not observe a signicant increase in insulin secretion in either trans or cis at both low (3 mM) and high (11 mM) glucose concentrations (Fig. 7g).Similarly, benchmark Gw-9508 did not stimulate GSIS at 3 mM or 11 mM glucose (Fig. 7g).An effect of BSA on Gw-9508 and/or FAAzo-10 potency was unlikely, since assays   with low (3 mM) glucose concentration but performed in the absence of the carrier were identical (data not shown).Experiments were also repeated at high (17 mM) glucose, but without BSA, showing a similar lack of stimulation with Gw-9508 or FAAzo-10 (Fig. S4d †).Neither FAAzo-10 nor Gw-9508 were able to suppress tolbutamide-stimulated insulin secretion, further supporting an effect on K ATP channel conductance (Fig. S4e †). 11UV-A irradiation alone did not affect oscillatory behavior or insulin secretion levels, as expected from previous studies 25,31 (Fig. S4f †).

Conclusions
In summary, we have demonstrated that FAAzo-10 is a potent photoswitchable agonist of GPR40, and reversibly inactivates K + channels in dissociated mouse b-cells.Although our previous studies using the FAAzos conjugated to different headgroups afforded cis-active compounds, 30,31 we found the opposite in this case.FAAzo-10 was more active in the trans-form at all targets, and can reversibly stimulate [Ca 2+ ] i oscillations in pancreatic bcells using light.Interestingly, stimulation of [Ca 2+ ] i oscillations with FAAzo-10 did not translate to increased insulin secretion in primary mouse islets, in line with the effects of the benchmark drug, Gw-9508.This suggests that oscillations by themselves are potentially not a sufficient signal for effective granule fusion, and that an additional factor was not triggered under these conditions.Of note, previous studies using Gw-9508 have afforded either stimulatory, inhibitory or no effect on insulin secretion, 11,21,39 with two conicting reports in mouse islets. 12,40s previously alluded to using the PhoDAGs, 31 the variation of the effects induced by Gw-9508 application may stem from different protein expression levels or membrane area between immortalized and primary cells, or conversely off-target effects on GPR120, which shares some homology with GPR40.Similarly, differential effects caused by plasma membrane vs.intracellular fatty acid-signalling, as was observed using caged AA-derivatives, may contribute to this effect. 6By contrast, long chain fatty acids such as linoleic and palmitic acid have been consistently shown to potently stimulate insulin secretion, and this can be abrogated by GPR40 knockdown/silencing. 18 Our studies thus reinforce the notion that signals in addition to GPR40 activation may be required for fatty-acid-stimulated insulin release, highlighting the complexity of fatty acid signalling in the b-cell, and underscoring the importance of FAAzo-10 for studying the intricate relationship between [Ca 2+ ] i oscillations and insulin secretion.More broadly, FAAzo-10 opens up the possibility to precisely interrogate the contribution of GPR40 signalling in different body compartments (e.g.brain and liver) to glucose homeostasis.

Fig. 4
Fig. 4 Spatial control of GPR40 signalling with FAAzo-10.(a) Confocal images of HeLa cells expressing GPR40 and R-GECO before and after treatment with FAAzo-10 (200 nM) and illumination with 375 nm light.The green rectangle indicates the area of illumination.After addition of FAAzo-10, all transfected cells showed increased [Ca 2+ ] i .Following illumination, only cells within the green rectangle showed a sharp decrease in [Ca 2+ ] i levels, which recovered after termination of illumination.Scale bar ¼ 100 mm.(b) Normalized [Ca 2+ ] i in illuminated cells (within the green rectangular in (a)) in blue (n ¼ 52) and those in unilluminated cells (outside the green rectangular) in black (n ¼ 82).Time points 1-4 correspond to the respective time frames in (a).Error bars were calculated as AEs.e.m.

Fig. 5
Fig. 5 Optical control of b-cell K v channel activity.The whole-cell K v channel current in dissociated wt mouse b-cells was measured using patch clamp electrophysiology.(a) An IV-plot showed that Gw-9508 (50 mM) (n ¼ 8 cells from 2 animals) reduced the K v conductance when compared to a vehicle control (n ¼ 6 cells from 3 animals).(b) Under blue light, trans-FAAzo-10 (20 mM) reduced the whole-cell K v current.Isomerization to cis-FAAzo-10 with UV-A light reversed this effect (n ¼ 7 cells from 3 animals).(c) Similar to the wash-in and wash-out of Gw-9508, FAAzo-10 could be activated and inactivated over several cycles using irradiation.Shown are IV-steps from À70 to +80 mV from representative cells.(d, e) An action spectrum between 350-450 nm showed that K v activity could be fine-tuned by changing the irradiation wavelength.Displayed as (d) overlaid sequential voltage ramps (À70 to +80 mV) from a representative cell and (e) the normalized current (to I 350 nm ) under each wavelength (n ¼ 3 cells from 2 animals).Error bars were calculated as AEs.e.m.

Fig. 7
Fig. 7 FAAzo-10 enables optical control of [Ca 2+ ] i oscillations in pancreatic islets.[Ca 2+ ] i oscillations were stimulated by a high glucose concentration (11 mM, G11) and monitored in intact mouse islets using the fluorescent [Ca 2+ ] i indicator Fluo-8.(a, b) The application of Gw-9508 (50 mM) caused an increase in the [Ca 2+ ] i oscillation frequency.Displayed as (a) a representative trace from a single islet and (b) the oscillation frequency averaged over multiple islets (n ¼ 6 recordings).(c, d) The application of trans-FAAzo-10 (20 mM) also caused a marked increase in the oscillation frequency.Isomerization to cis-FAAzo-10 with 365 nm irradiation reversed this effect.Results are displayed as (c) a representative trace from a single islet and (d) the average oscillation frequency from multiple islets (n ¼ 5 recordings).(e, f) FAAzo-10 enabled optical control of b-cell [Ca 2+ ] i oscillations at 20 mM, but not at 2.5 mM (n ¼ 4-5 recordings) (representative images cropped to show a single islet; scale bar ¼ 25 mm).(g) FAAzo-10 (20 mM) did not afford a consistent effect on GSIS (3 mM glucose, G3).Gw-9508 (20 mM) also did not affect GSIS (n ¼ 3-8 assays using islets from at least 3 animals) (* denotes significance between G3 and G11).Grey lines are raw traces (to show frequency effects), black lines are smoothed traces (to show amplitude effects).*P < 0.05 and **P < 0.01, ANOVA, with repeated measures as necessary.Error bars were calculated as AEs.e.m.

Fig. 6
Fig. 6 Optical control of b-cell K ATP channels.The whole-cell K ATP current from dissociated mouse b-cells was measured between À110 to À50 mV.(a-c) After dialysis of the cytoplasm with the pipette solution, the K ATP current developed to a steady state (black ¼ before, n ¼ 21; green ¼ after, n ¼ 20 cells from 2 animals).Application of Gw-9508 (20 mM, red, n ¼ 9 cells from 2 animals) increased K ATP conductance.In contrast, the application of trans-FAAzo-10 (20 mM, blue) decreased the K ATP current, while isomerization to cis-FAAzo-10 (gray) reversed this effect (n ¼ 7 cells from 2 animals).Data is displayed as (a, b) the full IV relationship between À110 to À50 mV and (c) the % K ATP current (at À110 mV) for multiple cells, normalized to the K ATP open (green) state.(d) In the presence of FAAzo-10, an action spectrum between 350-450 nm revealed that K ATP was inhibited the most under blue irradiation.Irradiation with UV-A light prevented FAAzo-10 from blocking the K ATP current.Displayed as the normalized current (to I 350 nm ) under each wavelength (n ¼ 3 cells from one animal).(e) UV-A or blue irradiation alone did not affect the K ATP current, and tolbutamide (40 mM) significantly reduced the magnitude of the K ATP current (DI from À110 to À50 mV, n ¼ 3 cells from one animal).ns ¼ P > 0.05, *P < 0.05, **P < 0.01.Error bars were calculated as AEs.e.m.