Allosteric Optical Control of a Class B G‐Protein‐Coupled Receptor

Abstract Allosteric regulation promises to open up new therapeutic avenues by increasing drug specificity at G‐protein‐coupled receptors (GPCRs). However, drug discovery efforts are at present hampered by an inability to precisely control the allosteric site. Herein, we describe the design, synthesis, and testing of PhotoETP, a light‐activated positive allosteric modulator of the glucagon‐like peptide‐1 receptor (GLP‐1R), a class B GPCR involved in the maintenance of glucose homeostasis in humans. PhotoETP potentiates Ca2+, cAMP, and insulin responses to glucagon‐like peptide‐1 and its metabolites following illumination of cells with blue light. PhotoETP thus provides a blueprint for the production of small‐molecule class B GPCR allosteric photoswitches, and may represent a useful tool for understanding positive cooperativity at the GLP‐1R.

We set out to confer photoswitching on the GLP-1R allosteric site by subjecting BETP to our "azologization" strategy [11] (Figure 1B;s ee also Figure S1 in the Supporting Information). By coupling commercially available chloropyrimidine 1 and boronic acid 2 under Suzuki-Miyaura conditions,b isaryl thioether 3 was obtained in ay ield of 95 %. After oxidizing the sulfur atom with mCPBAt oi ts sulfone counterpart 4 in ay ield of 90 %, it was exchanged in an aromatic substitution with ethyl sulfide to give ethyl thioether 5 in ayield of 55 %. Subsequent oxidation with one equivalent of mCPBAg ave access to sulfoxide 6 (96 %), which was deprotected with TFAbefore undergoing Mills reaction with nitrosobenzene to produce PhotoETP in ayield of 54 %over two steps.A ttempts to crystallize PhotoETP from MeOH yielded compound 7,t hus providing further evidence for the electrophilicity of PhotoETP ( Figure 1C). Crystals suitable for X-ray crystallography were obtained for both 3 and 7 ( Figure 1D). TheUV/Vis spectrum of PhotoETP under illumination at l = 440 nm (trans,b lue) and l = 330 nm (cis,g ray) demonstrated the presence of wavelength-dependent switching (Figure 2A), as expected for a meta-azobenzene system ( Figure 2B). Photoswitching could be repeated over several cycles with reasonably fast kinetics (t cis = 204.2 AE 7.3 s; t trans = 54.5 AE 2.9 s) and without obvious fatigue ( Figure 2C), and the cis and trans isomers were separated by LC-MS analysis ( Figure 2D). Together, these features of PhotoETP provide abasis for the allosteric photocontrol of GLP-1R activity.
We next examined whether PhotoETP was able to yield optical control over GLP-1(9-36)NH 2 -induced cAMP generation. Similarly to BETP, trans-PhotoETP potentiated cAMP rises in response to GLP-1(9-36)NH 2 [EC 50 (trans-Photo-ETP) = 163.3 nm;E C 50 (BETP) = 99.5 nm].I mportantly,t his activity could be switched off using UV illumination to induce cis-isomer formation ( Figure 3A;u nable to calculate EC 50 ). Thee xtent of photoswitching is similar to that recently reported for an allosteric modulator of the metabotropic glutamate receptor mGluR5, aclass CGPCR. [12] Theeffect of PhotoETP on cell viability was determined in islets by using necrosis and apoptosis assays in the dark. At the concentration used throughout the present study,PhotoETP did not induce significant necrosis ( Figure 3B)o ra poptosis (Figure 3C), as measured using propidium iodide incorporation and terminal deoxynucleotidyl transferase dUTP nick end labelling (TUNEL), respectively.Bycontrast, treatment with staurosporine or thapsigargin positive controls resulted in large increases in necrotic and apoptotic indices (Figure 3B,C). Furthermore,l evels of cleaved caspase-3, an enzyme involved in the proteolytic cleavage of critical intracellular effectors including poly (ADP-ribose) polymerase,w ere unaffected by incubation with PhotoETP ( Figure S2).
Next, photoswitching of intracellular Ca 2+ dynamics was assessed by using PhotoETP directly in beta cells residing within intact islets of Langerhans.W hereas cis-PhotoETP (l = 350 nm) showed little effect, the trans isomer (l = 440 nm) potentiated GLP-1(7-36)NH 2 -induced increases in Ca 2+ levels ( Figure 3D-G), as previously described for cAMP. Thel atter could be abolished by using either low glucose ( Figure S3A,B) or the specific GLP-1R antagonist exendin 9-39 ( Figure S3C,D). In all cases,results in islets were replicated in MIN6 beta cells subjected to high-throughput Ca 2+ screens ( Figure 3H-J). When using batch-incubated islets, trans-PhotoETP potently amplified GLP-1(9-36)NH 2 -induced insu-  PhotoETP was almost 20 % cis-enriched under benchtop conditions ( 1 HNMR spectrum;s ee the Supporting Information), whereas the UV/Vis spectra revealed am ore pronounced p-p*band in the dark (see Figure 2B). This finding can be explained by consulting am odel of the glucagon receptor mutant F345C bound to BETP,w here at wisted conformation of the benzylether is found to be the lowest energy state. [7] Such ac onformation may also be adopted by cis-PhotoETP owing to its higher affinity for covalent binding. However,i nc ontrast to BETP,w hich can reorganize its molecular shape in response to orthosteric ligand binding, PhotoETP would remain trapped in its cis state until illumination to induce trans-isomerization. Although the exact isomer ratio at the receptor is difficult to determine empirically,s uch properties may nonetheless afford fine control over photoswitching, with dark conditions,4 40 nm illumination, and 350 nm illumination leading to graded Ca 2+ responses ( Figure S4). Further studies using rigid E-a nd Zstilbene bioisosteres of PhotoETP will be required to better delineate the mechanisms involved.
Thed ata presented herein outline as traightforward synthetic strategy for the production of ablue-light-activated positive allosteric modulator, which enables photocontrol of GLP-1R activity through af eedforward loop encompassing the orthosteric site ( Figure 1A). Although similar "alloswitches" have been described for ionotropic and metabotropic mGluRs, [12,13] this is the first demonstration of their use in at herapeutically relevant class BG PCR. Using ac ombination of Ca 2+ ,c AMP,a nd insulin assays in CHO-GLP-1R and MIN6 cells,aswell as islets of Langerhans,wewere able to show that PhotoETP allows photoswitching of responses to GLP-1(7-36)NH 2 and its less active breakdown product, GLP-1(9-36)NH 2 ,w ith similar potency to native BETP. Notably,P hotoETP displays unusual behavior in cells, where it shows an enriched cis content when interacting with its target in the dark. Indeed, the more active trans isomer has to be photochemically induced by irradiation with blue light. As aresult of these properties,PhotoETP,together with the recently described signal-biased GLP-1R photoswitch LirAzo, [14] may enable the precise dissection of allosteric-orthosteric cooperativity,m olecular movement, and binding.
Both BETP and PhotoETP were more effective at potentiating GLP-1(9-36)NH 2 -induced compared to GLP-1(7-36)NH 2 -induced insulin secretion. This suggests that cAMP rather than Ca 2+ is the primary driver of the "incretin effect", and is consistent with previous results obtained using LirAzo. [14] Intriguingly,optical control of insulin release could only be observed in GLP-1(9-36)NH 2 -and PhotoETP-treated islets,w here blue light provoked at wo-fold higher response than UV illumination. Although the exact reasons for this remain unknown, it may reflect an inability to detect relatively small isomer-induced differences in intracellular Ca 2+ versus cAMP concentration at the level of secretion in islets.
BETP was susceptible to UV-A-induced but not whitelight-induced reactions,t hus making it ap oor control for photoswitching purposes (see Figure S5-S8). In contrast, PhotoETP was remarkably robust. This protective effect  stems from the azobenzene unit, which preferentially harvests UV-A photons with its p-p*band to undergo isomerization. In other words,b yi nstalling an azobenzene moiety onto BETP,s ide reactions can be quenched and the resulting molecule stabilized. Nevertheless,t he UV-A-induced rearrangement of BETP to its sulfenic ester counterpart via aMeisenheimer complex, and the accompanying transformation, is in itself an interesting finding ( Figure S5,S6). Although related rearrangements of sulfoxides have been reported, [15] sulfenic esters have not been isolated as products owing to the low (UV-C) wavelengths used in these experiments.S uch rearrangements are relevant for drug activity,a sb est exemplified by acid-activation of the irreversible proton-pump inhibitor omeprazole (Prilosec). [16] In summary,w es howcase PhotoETP,alight-activated modulator for allosteric optical control of GLP-1R function, and highlight the requirement to run parallel control experiments with benchmark drugs in photopharmacology.P hoto-ETP,oroptimized derivatives thereof,may be useful in drugdiscovery programs aimed at unraveling the complexity of allosterism and class BG PCR signaling.