ATP

adenosine triphosphate

CB2R

cannabinoid 2 receptor

GLP1R

glucagon‐like peptide‐1 receptor

GPCR

G protein‐coupled receptor

GSIS

glucose‐stimulated insulin secretion

IP₃R

inositol triphosphate receptor

K_ATP

ATP‐sensitive K⁺‐channel

N

number of cells

PLC

phospholipase C

ROI's

regions of interest

T

trials, number of biological replicates

T2DM

type II diabetes mellitus

Ca²⁺ regulates intracellular signaling and the secretion of hormones and neurotransmitters from excitable cells. In pancreatic β‐cells, Ca²⁺ controls the release of insulin, a key hormone that regulates glucose homeostasis [1]. After a meal, blood glucose is elevated and is transported into the β‐cell where it is metabolized to ATP. This closes ATP‐sensitive K⁺ channels (K_ATP) and depolarizes the β‐cell, triggering Ca²⁺ influx through voltage‐dependent Ca²⁺ channels and subsequent insulin release. In addition, β‐cell Ca²⁺ is regulated by myriad receptor pathways, including ion channels and G protein‐coupled receptors (GPCRs). These Ca²⁺ fluctuations are oscillatory in nature, and their intensity and frequency correlate with insulin release [2]. Dysfunction in β‐cell Ca²⁺ signaling and glucose‐stimulated insulin secretion (GSIS) leads to type 2 diabetes mellitus (T2DM), a disease whose prevalence is rising throughout the world at epidemic rates. As such, novel approaches to modulate β‐cell Ca²⁺ levels and insulin release are urgently needed to address this epidemic.

Small‐molecule therapeutics that treat diabetes by promoting insulin release include sulfonylureas, which block K_ATP channels [3], and approaches that modulate the β‐cell's intracellular Ca²⁺ concentration ([Ca²⁺]_i) through alternative receptor pathways, including GPCR targets, have gained significant therapeutic interest [4]. For example, the GPR40 agonist TAK‐875 improved glycemic control in patients but was pulled from clinical trials due to toxicity concerns [5]. More recently, glucagon‐like peptide‐1 receptor (GLP1R) agonists have demonstrated improved safety for treating metabolic disease and are increasingly being prescribed for T2DM [6]. Importantly, β‐cells express nearly 300 GPCRs [7], and these represent unexplored therapeutic targets that could eventually be leveraged as new therapeutic targets for T2DM.

One such GPCR family is the cannabinoid receptors (CBRs), which are activated by endocannabinoid lipids and phytocannabinoids found in marijuana. Cannabinoid CB1 and CB2 receptors (CB1R and CB2R) are expressed in the human pancreas and affect β‐cell excitability [8]. Additionally, noncanonical ‘orphan’ CBRs—such as GPR55 [9, 10]—affect insulin release from β‐cells in vitro and in rodent models [11]. Unfortunately, cannabinoid‐based therapeutics have had difficulty translating to the clinic, likely because we have a limited understanding of how CBRs affect β‐cell function. This is due to the hydrophobic nature of cannabinoid ligands, which are insoluble in physiological solutions and makes their application to cells slow and irreversible [12]. Cannabinoids also have complex pharmacology, and act through multiple GPCR and ion channel targets [13, 14]. To deconvolute the precise roles of CBR signaling in modulating β‐cell excitability and leverage this important signaling system for therapeutic development, we require tools that enable greater spatiotemporal resolution over cannabinoid signaling pathways.

Photopharmacology is an emerging approach to address these challenges by enabling optical control of biological processes with light‐sensitive small molecules, including photocaged and photoswitchable probes [15]. Photocaged ligands have their activity blocked by a photo‐labile protecting group, where a flash of light irreversibly releases the active compound. Photocages for Ca²⁺ and other ligands, including endocannabinoids [10, 16, 17], have been applied to optically control β‐cell [Ca²⁺]_i [18]. Alternatively, photoswitchable probes are ligands that contain a photo‐isomerizable motif, most commonly an azobenzene, that switches between the trans‐ and cis‐conformations using two different wavelengths of light [19]. The change in molecular structure through cis/trans isomerization tunes the potency and affinity of the ligand, enabling reversible optical control of receptor activation with superior kinetics compared to standard pharmacology. Numerous photoswitches—including sulfonylureas or endogenous lipids—have been shown to enable dynamic manipulation of [Ca²⁺]_i in cultured β‐cells and pancreatic islets [20, 21, 22, 23, 24]. Although our group and others have developed an array of photoswitchable cannabinoids to modulate both CB1R [25, 26, 27] and CB2R [28, 29], to date, they have not been applied to β‐cells.

Recently, our group developed a photoswitchable ligand based on the CB2R‐selective synthetic cannabinoid, HU308 [30]. We incorporated an azobenzene into HU308's alkyl chain to afford a photoswitchable ligand, azo‐HU308‐3 (now coined azo‐HU308) (Fig. 1A). Using a Ca²⁺ imaging assay in AtT‐20 cells overexpressing CB2R [AtT‐20(CB2) cells], we demonstrated that azo‐HU308 enables reversible, optical control of [Ca²⁺]_i when photoswitched to cis, and that the mechanism proceeded by CB2R and phospholipase C, and subsequent Ca²⁺ release from intracellular stores. As a follow‐up to this study, here we present the application of azo‐HU308 to optically control [Ca²⁺]_i in cultured pancreatic β‐cells. We show that azo‐HU308 robustly elevates intracellular Ca²⁺ in β‐cells when isomerized to cis with UV‐A light and permits multiple rounds of stimulation. Using a pharmacological screen to determine the mechanism of action, we observed that the Ca²⁺ influx was not CB2R‐mediated, but rather involves TRPC ion channels and Ca²⁺ influx across the plasma membrane. This study presents a novel tool to optically modulate β‐cell Ca²⁺ dynamics and sheds light on a new mechanism by which synthetic cannabinoids may modulate Ca²⁺ influx through non‐GPCR targets.

Results

Azo‐HU308 enables reversible optical control of Ca²⁺ influx in INS‐1 pancreatic β‐cells

Motivated by studies that showed that CBR signaling modulates Ca²⁺ in β‐cells [17, 34] and our finding that azo‐HU308 reversibly controls Ca²⁺ levels via CB2R and Ca²⁺ release from intracellular stores [31], we asked whether azo‐HU308 could optically modulate [Ca²⁺]_i in cultured β‐cells. We chose the rat insulinoma INS‐1 832/13 (INS‐1) cells because they exhibit glucose‐stimulated Ca²⁺ oscillations that couple to insulin release [32], and little is known about CBR signaling in INS‐1 cells. INS‐1 cells were transfected with the genetically encoded Ca²⁺ sensor R‐GECO [35], which allowed us to monitor [Ca²⁺]_i levels in real time by fluorescence microscopy (Fig. 1B, top). We observed robust Ca²⁺ oscillations when the cells were exposed to elevated glucose (20 mm) solution (Fig. 1C). Application of trans‐azo‐HU308 did not affect [Ca²⁺]_i (Fig. 1C and 1D, red); but strikingly, when we irradiated the cells with 375 nm UV‐A light (~ 95 μW laser irradiation) to isomerize azo‐HU308 to cis, we observed a robust [Ca²⁺]_i burst across the cells (Fig. 1B bottom, C, D, red, E, Fig. S1A and Table S1). The Ca²⁺ increase was intense but short‐lived and decayed rapidly before the photostimulation period was finished. The effect of azo‐HU308 was dose‐sensitive, and the [Ca²⁺]_i‐stimulating effect continued to increase up to 20 μm (Fig. S1B,C and Table S2). In control experiments, the parent molecule HU308 [30] caused a similar Ca²⁺ response immediately upon addition. The Ca²⁺ response induced on HU308 application quickly died off, and there was no further Ca²⁺ effect induced by 375 nm irradiation, confirming that the parent compound HU308 is not light‐sensitive (Fig. 1E, Fig. S2A–D and Table S1). In vehicle control experiments, where DMSO (0.1%) was added in the absence of azo‐HU308, there was no change to [Ca²⁺]_i upon addition or irradiation (Fig. 1D, black). This confirms that the effect was not directly caused by the irradiation or an artifact of compound addition.

To test the reversibility of azo‐HU308's effects, 375‐nm photostimulation was repeated over six cycles. We observed a Ca²⁺ increase on each irradiation; however, the magnitude decayed on each subsequent stimulation, which indicates that receptor desensitization was likely occurring (Fig. 1F). To highlight the outstanding spatial precision of azo‐HU308, we applied targeted illumination of three different square regions of interest on the plate (Fig. 1G). We observed [Ca²⁺]_i increases only in the cells that were irradiated, while cells that were outside the irradiation area did not respond (Fig. 1H). Combined, these results demonstrate that azo‐HU308 is a photoswitchable ligand that robustly stimulates Ca²⁺ in INS‐1 β‐cells when activated with UV light and isomerized to cis.

Azo‐HU308's effect on β‐cell Ca²⁺ is not mediated by CB2R

After verifying the robust effect of cis‐azo‐HU308 stimulating INS‐1 cell [Ca²⁺]_i, we wanted to determine its mechanism of action. Our previous study showed that azo‐HU308 and HU308 elevate [Ca²⁺]_i in AtT‐20(CB2) cells through CB2R and phospholipase C (PLC)‐mediated Ca²⁺‐release from intracellular stores [31]. Therefore, we co‐applied CB2R antagonists alongside azo‐HU308 to verify the contributions of CB2R signaling. To our surprise, azo‐HU308 still caused a large Ca²⁺ response in INS‐1 cells upon irradiation, even when co‐applied with three different CB2R antagonists—AM630, SR144528, and JWH133 (Fig. 2A,B and Table S3) [36, 37, 38]. Although HU308 is touted as a selective ligand for CB2R [30], it is possible that our probe has off‐target effects at other CBRs. INS‐1 cells express multiple CBRs including CB1R and GPR55 [10, 34]. However, co‐application of the CB1R antagonist rimonabant [39], or the GPR55 antagonist CID16020046 [40] failed to block azo‐HU308's photoswitching effect (Fig. 2A, Fig. S3A,B and Table S3). These results suggest that the effects of azo‐HU308 in β‐cells are not mediated by CB2R or through off‐target CBR pathways.

Next, we asked whether PLC and Ca²⁺‐release from internal stores were being stimulated by cis‐azo‐HU308, as found in our previous study [31]. Surprisingly, both the PLC inhibitor U73122 and the inositol 1,4,5‐trisphosphate receptor (IP₃R) inhibitor Xestospongin C failed to block the photoswitching effect (Fig. 2A,C), indicating that an entirely different Ca²⁺ signaling mechanism was involved [41, 42]. We tested several inhibitors of G protein signaling. The G_q/11 G protein inhibitor YM254890 produced only a modest inhibition of the photoswitching effect, but the response was still largely intact (Fig. 2A,D, yellow) [43]. We applied the G_s protein inhibitor NF449 (Fig. 2A,D, cyan) and the G_i/o protein blocker pertussis toxin (PTX, Fig. 2A,D, magenta) [44, 45]. Both drugs caused only a small reduction of cis‐azo‐HU308's effect on [Ca²⁺]_i. Combined, these results suggest that the mechanism of action for azo‐HU308's effect on [Ca²⁺]_i release in INS‐1 cells was independent of CB2R, other CBRs, or even GPCR activation entirely. These results suggest an entirely unique mechanism compared to our previous study in CB2R‐overexpressing AtT‐20 cells.

Azo‐HU308's photoswitching effect is modulated by ion channel targets

Besides GPCRs, pancreatic β‐cell [Ca²⁺]_i is affected by numerous transmembrane ion channels, including transient receptor potential (TRP) channels [46]. Based on the rapid and transient nature of azo‐HU308's effect, we hypothesized that these may be involved as an unintended receptor target. To this end, we tested the effect of 2‐aminoethoxydiphenyl borate (2‐APB), which is a nonselective TRP channel inhibitor and IP₃ receptor modulator [47], and observed a dose‐dependent inhibition of azo‐HU308 photoswitching (Fig. 3A,B and Table S4). At low concentrations of 2‐APB (1 μm, red), we observed a potentiation of the effect of trans‐azo‐HU308, as a larger [Ca²⁺]_i increase was observed upon probe addition, while the cis‐effect remained constant. At a mid‐range concentration (50 μm, blue), the photoswitching effect was significantly reduced, and at the highest concentration (100 μm, black), the effect of trans‐ and cis‐azo‐HU308 was completely abolished. We then conducted our Ca²⁺ imaging experiments in Ca²⁺‐free extracellular solution (with added ethylene glycol‐bis(β‐aminoethyl ether)‐N,N,N′,N′‐tetraacetic acid (EGTA)), where we observed that azo‐HU308's photoswitching effect on Ca²⁺ was completely abolished (Fig. 3A,C and Table S4). Combined, these results indicate that the Ca²⁺ response of cis‐azo‐HU308 is mediated by ion channels expressed on the INS‐1 cell surface, and that the rise in [Ca²⁺]_i requires extracellular Ca²⁺ influx.

We then set out to determine the specific ion channels involved. INS‐1 cells express a variety of TRP channels, which are known to respond to lipid agonists [46, 48, 49, 50]. Transient receptor potential vanilloid 1 (TRPV1) is expressed in INS‐1 cells [51] and has been shown to be activated by lipids. However, when INS‐1 cells were pretreated with the TRPV1 antagonist capsazepine [52], azo‐HU308 photoswitching effects remained intact (Fig. 3A and Fig. S4A), ruling out TRPV1's role. Next, we set our sights into TRPC channels. Human and murine pancreatic islets and β‐cells, including INS‐1 cells, express multiple TRPC channels, which are important for store‐operated Ca²⁺ homeostasis and can regulate insulin secretion [53, 54]. Gratifyingly, when we pretreated with the broadband TRPC inhibitor SKF96365, there was a reduction in amplitude to the photoswitching effect of azo‐HU308 (Fig. 3A,D, green) [55]. We then applied more selective antagonists. Interestingly, co‐application of the TRPC4 inhibitor ML204 potentiated the photoswitching response (Fig. 3A, D, orange). Additionally, the TRPC6 inhibitor SAR7334 failed to block the photoswitching response (Fig. 3A and Fig. S4B) [56, 57].

TRPC channels mediate azo‐HU308's photoswitching effect on Ca²⁺ in INS‐1 cells

A recent study has shown TRPC channels' expression and regulation of insulin secretion from human and murine islets [58], and our broadband TRPC inhibitor SKF96365 reduced the amplitude of azo‐HU308 photoswitching effects in INS‐1 cells. Therefore, we focused on TRPC channels using blockers that have greater selectivity. Intriguingly, treatment with YM58483—which blocks TRPC3 and TRPC5 channels [59]—completely abolished the [Ca²⁺]_i response caused by cis‐azo‐HU308 (Fig. 4A,B, red and Table S5). Next, we treated INS‐1 cells with the selective TRPC3‐specific inhibitor Pyr10 [60]. In this case, we saw a ~ 50% reduction in the Ca²⁺ response caused by photoswitching (Fig. 4A,B, blue and Table S5), demonstrating that TRPC3 may be involved but is not solely responsible for the effect. With both YM58483 and Pyr10 effectively reducing the azo‐HU308‐mediated Ca²⁺ response during irradiation, these results suggest that the proposed target of our photoswitch is likely through TRPC channels expressed in INS‐1 cells.