A red-shifted photochromic sulfonylurea for the remote control of pancreatic beta cell function

Abstract

Azobenzene photoresponsive elements can be installed on sulfonylureas, yielding optical control over pancreatic beta cell function and insulin release. An obstacle to such photopharmacological approaches remains the use of ultraviolet-blue illumination. Herein, we synthesize and test a novel yellow light-activated sulfonylurea based on a heterocyclic azobenzene bearing a push-pull system.

Type 2 diabetes (T2D) is a modern pandemic currently affecting ~ 10% of the global population. This disease is characterized by diminished insulin secretion from pancreatic beta cells, which together with peripheral resistance to the secreted hormone, leads to defective glucose homeostasis.¹ The resulting elevated glucose concentration drives a variety of complications including heart disease, cancer, retinal degeneration, and nerve and vascular problems.²

While current medical treatments work well, they are associated with complications largely due to off-target or persistent actions.³ Moreover, they are unable to recreate pulsatile insulin release, a more effective signal for glucoregulation.⁴ Thus, T2D is ideally suited to photopharmacology, which harnesses the temporal precision of light to spatiotemporally deliver drug activity.⁵ We have recently shown that a sulfonylurea possessing an azobenzene photoresponsive element (a.k.a. AzoSulfonylurea) can be used to optically control beta cell function and insulin release via its effects on ATP-sensitive potassium (K_ATP) channels and Exchange Protein directly Activated by cAMP 2A (Epac2A) signalling.⁶

However, a significant barrier to the use of such ‘azo-drugs’ for T2D treatment is their ultraviolet-blue absorption spectra, increasing phototoxicity and limiting tissue penetration due to photon scattering.⁷ By contrast, visible/near infrared wavelengths demonstrate better penetrance in the body. ⁸

Spurred on by recent studies of ortho- or para-substituted azobenzenes,^9–11 we therefore devised a novel approach for the synthesis of wavelength-tuned photopharmaceuticals with red-shifted photochromism. An AzoSulfonylurea based on glimepiride was achieved by installing a heterocyclic aromatic unit, rather than sterically bulky electron-donating halogen or amine moieties (Scheme 1).

Scheme 1.

(a) Structures of glimepiride (Glim) and the original blue light-responsive AzoSulfonylurea JB253 for comparison. (b) The logic of a red-shifted AzoSulfonylurea. Following illumination with green-yellow light (1), the AzoSulfonylurea binds Epac2A, closing K_ATP channels (2) and opening voltage-dependent Ca²⁺ channels (Ca_v) (3). This allows optical control of Ca²⁺ influx (4) and insulin secretion (5). (c) Synthesis of the AzoSulfonylurea JB558 that can be switched from the trans- to the cis-isomer using green/yellow light.

Starting with the deacetylation of acetazolamide (1) in refluxing HCl, heterocycle 2 was obtained that could be further diazotized with in situ generated HNO₂. Trapping the resulting diazonium salt with N, N-diethylaniline generated sulfonamide azobenzene 3. Finally, reaction with cyclohexyl isocyanate yielded JB558 via acylation of the sulfonamide, giving unprecedented access to a sulfonylurea containing a heterocyclic azobenzene. While yields were reduced compared to the previously described JB253 (37% versus 97%),⁶ this was most likely due to the presence of a less reactive sulfonamide intermediate, as predicted by the lower pK_a value for 3 (7.36, Fig. S1) and JB558 (2.35, see SI).

JB558 possessed a red-shifted absorption spectra (λ_max = 526 nm) in DMSO (Fig. 1a), and could be repeatedly photoconverted to its cis-state with green-yellow light (λ = 520 nm) (Fig. 1b). Thermal back relaxation occurred rapidly in the dark and switching kinetics were within the millisecond range (τ_cis = 64.9 ± 1.5 ms; τ_trans = 410.8 ± 12.6 ms), without obvious decomposition (Fig. 1b). JB558 was stable in the presence of Escherichia coli azoreductase, an enzyme expected to limit oral bioavailability through diazene cleavage in the intestine (Fig. S2).

Figure 1.

(a) UV-Vis spectra of JB558 in DMSO following illumination with λ = 520 nm (green) or under dark-adapted conditions (black). (b) Robust photoswitching between cis- and trans-JB558 induced with λ = 520 nm and dark, respectively.

To determine the binding affinity of JB558 to the K_ATP channel subunit SUR1, as well as Epac2A, [3H]-glibenclamide displacement and FRET assays were performed. While trans-JB558 bound SUR1 with ~10,000-fold less affinity than glimepiride (IC₅₀ (trans-JB558) = 37.3 µM; IC₅₀ (Glim) = 1.8 nM) (Fig. 2a), it was able to strongly and light-dependently activate an Epac2A-camps biosensor containing the sulfonylurea binding domain¹² (Fig. 2b-d).

Figure 2.

(a) trans-JB558 and glimepiride (Glim) displace [3H]-glibenclamide from SUR1 (n = 3 repeats). (b) Glimepiride decreases FRET (shown here as an increase in R/R_min) in HEK293T cells expressing full length Epac2-camps (n = 32 cells). (c) As for (b) but cis-JB558 (λ = 561 nm) (n = 41 cells). (d) As for (c) but trans-JB558 (dark) (n = 37 cells). Values represent mean ± s.e.m.

Electrophysiological recordings of K⁺ currents in HEK293T-SUR1-Kir6.2 cells revealed partial K_ATP channel blockade by trans-JB558, presumably due to the momentary stationary state favouring some continued cis-isomerisation (Fig. S3).

We next assessed the photoswitching properties of JB558 in native beta cells where sulfonylurea-mediated K_ATP channel-Epac2A signalling is intimately linked to voltage-dependent Ca²⁺ channel (VDCC) activity and insulin exocytosis.^13–16 As expected, JB558 was able to evoke large increases in intracellular Ca²⁺ concentrations in ~60% of beta cells following exposure to yellow (λ = 561 ± 5 nm)- (Fig. 3a and b), but not violet (λ = 405 ± 5 nm)-light (Fig. 3c). These effects were potentiated using a high concentration of glimepiride (Fig. 3d), and abrogated using diazoxide (Fig. 3e) to force open the K_ATP channel pore. Repeated switching of cytosolic Ca²⁺ concentrations could be achieved in the same islet following a brief period of dark exposure to induce trans-JB558 accumulation (Fig. 3f).

Figure 3.

(a) JB558 increases intracellular Ca²⁺ concentrations in 59% of beta cells residing within rodent islets of Langerhans following illumination with λ = 561 nm (scale bar = 75 µm) (n = 8 islets). (b) Photoswitching is rapid following exposure to λ = 561 nm. (c) As for (b), but showing the absence of photoswitching with λ = 405 nm. (d) A high concentration (100 µM) of glimepiride (Glim) augments JB558-stimulated Ca²⁺ rises. (e) Diazoxide (Dz) reverses cis-JB558-induced Ca²⁺ fluxes. (f) Reversible manipulation of Ca²⁺ transients can be achieved in the same islet following thermal back relaxation of JB558 in the dark (5 min between stimulation 1 and 2). Traces represent n = 6-10 recordings from 3 animals. Islets were maintained in 5 mM D-glucose throughout.

Similar to the results observed in rodent tissue, cis-JB588 was able to confer light-sensitivity on Ca²⁺-spiking activity in human pancreatic islets (Fig. 4a-c), and this effect could be reversed following 5 min relaxation in the dark (Fig. 4d).

Figure 4.

(a) JB558 increases intracellular Ca²⁺ concentrations in human beta cells in response to illumination with 561 nm to induce cis-formation (scale bar = 50 µm). (b) Photoswitching is rapid following exposure to 561 nm and can be potentiated with glimepiride (Glim). (c) cis-JB588 activates 54% of beta cells (n = 4 islets). (d) Reversible manipulation of Ca²⁺ rises following thermal back relaxation of JB558 in the dark (5 min between stimulation 1 and 2). Traces represent n = 3-9 recordings from a single donor. Islets were maintained in 5 mM D-glucose throughout.

To link photocontrol of Ca²⁺ levels with insulin secretion, batches of rodent islets were incubated with JB558 and exposed to either dark (no illumination) or light (λ = 560 ± 10 nm). JB558–treated islets kept under dark conditions were no different to controls (5 mM glucose-alone) (Fig. 5a), suggesting that the observed stationary state K_ATP channel block was insufficient to elicit exocytosis. By contrast, irradiation dramatically stimulated insulin release (Fig. 5a). Finally, cytotoxicity assays demonstrated that JB558 did not adversely affect cell viability, as assessed using the vital stain calcein and the necrosis indicator propidium iodide (Fig. 5b and c).

Figure 5.

(a) JB558-treated islets respond to illumination with λ = 560 nm by increasing insulin secretion (**P<0.01 and NS, non-significant versus Con; one-way ANOVA). (b) Incubation with JB558 for 1 hr did not adversely alter cell viability versus dimethyl sulfoxide (DMSO), as assessed by the ratio of calcein (live):propidium iodide (dead) fluorescence (positive control; Triton X-100) (NS, non-significant versus Con; one-way ANOVA). (c) Representative images of islets stained with calcein and propidium iodide. In all cases, n = 36 islets per treatment group from 6 animals. Values represent mean ± s.e.m.

The data presented here outline a synthetic route for the production of AzoSulfonylureas with red-shifted photochromism. Consistent with its sulfonylurea backbone, JB558 was able to bind SUR1 and activate Epac2A. Formation of cis-JB558 occurred with green-yellow light (λ = 520–561 nm), and thermal back relaxation in the dark yielded trans-JB558. While photoconversion between cis- and trans- forms was rapid in solution, it was slower in the tissue setting, taking minutes for reisomerisation. An effect of Fluo-2 excitation on the isomer equilibrium cannot be excluded, although illumination at λ = 491 nm per se was unable to evoke Ca²⁺ rises in both Fluo-2- and Fura2 (λ = 340 nm/380 nm)-loaded islets (Fig. S4).

A more plausible explanation is the wind-up of Epac2A-mediated beta cell signalling cascades¹⁷, which may outlast JB558 inactivation. Such tissue effects may be desirable for the development of photopharmaceuticals, since pulsed illumination would reduce phototoxicity, while sustaining compound activity to match long-lasting (dozens of minutes) insulin peaks.⁴ Indeed, JB558 displayed almost 3-fold more potency than its blue-light activated predecessor JB253,⁶ most likely due to slower back-relaxation during the light pulses used in the secretion assays.

Neither were we able to detect photoswitching of K⁺ currents in HEK293T cells overexpressing K_ATP channels, free from orthogonal wavelengths (Fig. S5). This was likely because HEK293T cells do not express sufficient Epac2A to allow JB558 to properly toggle K_ATP activity,^{13, 15, 18} and/or the inability to deliver sufficient illumination using the non-coherent source on our patch-clamp setup (ε_{520 nm} (JB558) = 1.14 x 10⁵ mol^-1 cm^-1; see SI).

Nonetheless, we clearly show that JB558 light-dependently binds Epac2A, allowing optical control of cell function and insulin secretion with λ = 560 nm in the most physiologically-relevant testbed, viz the islets of Langerhans. Thus, JB558 represents a blueprint for red-shifted AzoSulfonylureas based upon heterocyclic azobenzenes. Further studies are now warranted to improve isomerisation kinetics in tissue to improve the use of JB558 as a research tool for rapid K_ATP channel manipulation. Importantly, similar synthetic approaches may also be applicable to other clinically-relevant azobenzene-possessing compounds where steric bulk may not be well-tolerated e.g. neuromodulators,¹⁹ neurotransmitters,^{20, 21} enzymes²² and antibiotics. ²³

Supplementary Material

Electronic Supplementary Information (ESI) available: [details of any supplementary information available should be included here]. See DOI: 10.1039/c000000x/