Potent Prearranged Positive Allosteric Modulators of the Glucagon‐like Peptide‐1 Receptor

Abstract Drugs that allosterically modulate G protein‐coupled receptor (GPCR) activity display higher specificity and may improve disease treatment. However, the rational design of compounds that target the allosteric site is difficult, as conformations required for receptor activation are poorly understood. Guided by photopharmacology, a set of prearranged positive allosteric modulators (PAMs) with restricted degrees of freedom was designed and tested against the glucagon‐like peptide‐1 receptor (GLP‐1R), a GPCR involved in glucose homeostasis. Compounds incorporating a trans‐stilbene comprehensively outperformed those with a cis‐stilbene, as well as the benchmark BETP, as GLP‐1R PAMs. We also identified major effects of ligand conformation on GLP‐1R binding kinetics and signal bias. Thus, we describe a photopharmacology‐directed approach for rational drug design, and introduce a new class of stilbene‐containing PAM for the specific regulation of GPCR activity.

Drugs that allosterically modulate Gp rotein-coupled receptor (GPCR)a ctivity display higher specificity and may improved isease treatment. However, the rational design of compounds that target the allosteric site is difficult, as conformations required for receptor activation are poorly understood.G uided by photopharmacology,aset of prearranged positive allosteric modulators (PAMs) with restricted degrees of freedom was designed and tested against the glucagon-like peptide-1 receptor (GLP-1R), aG PCR involved in glucoseh omeostasis. Compounds incorporating a trans-stilbene comprehensively outperformed those with a cis-stilbene, as well as the benchmark BETP,a s GLP-1R PAMs.W ealso identified major effects of ligand conformation on GLP-1Rb inding kineticsa nd signal bias. Thus, we describe ap hotopharmacology-directed approach forr ational drug design, and introduce an ew class of stilbene-containing PAMf or the specific regulation of GPCR activity.
To produce PAMs with improved activity at the GLP-1R, the PhotoETP azobenzene diazene unit, which exhibits non-binary photostationary states, [5] was replaced with aC =Cf ragmentt o introduce as tilbene that displays single isomers. Ar ange of prearranged molecules were synthesized and tested for their ability to potentiate GLP-1R signaling responses. In all cases, compounds incorporating the trans-stilbene outperformed those with cis-stilbene, in addition to native BETP.A ss uch, we introduce ap hotopharmacology-based strategy to direct the rational designo fp rearranged PAMs, with broad applicability to the allosteric regulation of GPCRs involved in health and disease.
BETP adopts ac onformationally free benzyl ether when bound, [6b] suggestingt hat, upon activation of the GLP-1Rb y orthosteric ligands, it is able to rearrange its shapet of ully engaget he allosterics ite (Figure 1a). This motion is restricted by light in PhotoETP,w hich possesses ap hotoisomerizable azobenzene in place of the O-benzyl group (Figure 1b). We reasoned that replacement of the diazene bridge in PhotoETP by aC =Cm oiety to introduce as tilbene would allow production of prearranged PAMs where the cis-a nd trans-states are mimicked, but withoutc omplicationsa rising from photostationary states. Stilbenes have the added advantage of being more drug-like than azobenzenes and may demonstrate better metabolic stability in terms of double-bond cleavage in the intestine and the possible liberation of reactive anilines.F urthermore, BETP uses an ethyl sulfoxide as al eaving group that is replaced by covalent cysteine attachment on the receptor to give ethyl sulfenic acid as al eaving group, [6b] and other electrophilicm oieties, mainly aryl chlorides and aryl sulfones, have also been reportedt ou ndergo covalent labellingt oward the GLP-1R. [7] We, therefore, wanted to explore whether am ethyl sulfoxide would be tolerated (Figure 1c), as this moiety would: 1) be easier and faster to introduce synthetically from commercially-available substrates, giving cheaper access to PAMs that target the GLP-1R ( Figure 1d); and 2) exhibit methyl sulfenic acid as an unstable and, therefore, quickly-cleared leaving group. [8] To obtain BMTP,t hat is, the methyl analogueo ft he lead compound BETP,a safirst model compound, boronic ester 1 was coupled under Suzuki-Miyaura conditions with chloropyrimidine 2 to biaryl 3 in quantitative yield, [9] which was monooxidized with one equivalent of mCPBA to give BMTP in 94 % yield (Scheme 1a). SMTP and SETP weres ynthesized in ac omparable synthetic sequence. Commencing with benzyltriphenylphosphonium bromide (4)a nd 3-bromobenzaldehyde (5) that underwent an on-stereoselectiveW ittig reaction at room temperature in THF with LiHMDS as base, cis-a nd trans-bromo stilbene 6 could be separated through flash columnc hromatography (FCC) and isolated in 79 %o verall yield (Scheme 1b). Single crystals of trans-6 were obtained, which unambiguously provided the isomeric identity,a nd this was further confirmed by the coupling constant 3 J HH of the olefinic protons through 1 HNMR spectroscopy,b eing approximately 12 and 16 Hz for cis-a nd trans-isomers, respectively.N ext, Miyaura coupling with B 2 pin 2 using PdCl 2 (dppf) under standard conditions, employing DMSO as the solventa nd KOAc as the base, [10] gave access to cis-a nd trans-boronic ester 7 in good yields (78 and 77 %, respectively), which were subsequently subjected to Suzuki-Miyaura Pd cross-coupling with chloropyrimidine 2 to yield the cis-a nd trans-thioetherp recursor 8 in good 76 %a nd excellent 95 %y ields, respectively.P rogressing towards SMTP, thioether 6 was monooxidized with one equivalent mCPBA to give cis-a nd trans-SMTP in 91 and8 9%,r espectively.
To progress to the ethyl sulfoxide, thioether 6 was oxidized with two equivalents of mCPBA to give the corresponding sulfone 9 (yield for cis:9 5%; trans:5 8%), whichw as subjected to aromatic substitution with am ixture of ethylsulfide/sodium thioethanolate to ethyl thioether 10 (cis:8 9%; trans:8 0%)i n amicrowave reactor before final installation of the ethyl sulfoxide with one equivalent of mCPBA to obtain cis-a nd trans-SETP in 74 and 85 %y ields, respectively.L astly,w eo btained ac rystal structure of BETP by leaving aD MSO solution open to the atmosphere for 2weeks (Scheme 1c), providing atomic coordinates and showing that BETP is stable for this timeframe in solution at room temperature and under benchtop light.  [3] endowsGLP-1R with light sensitivity whenthe azobenzene is lockedi nits active trans-state underb lue light illumination (blue circle), whichi si nterchangeable with UV light to its inactive cis-state( gray circle). c) Novels tilbene congeners are constantly locked (orange circle)i n their respective state and do not exhibit photostationarystates.The functional group for displacementc an be am ethyl(Me)o re thyl (Et) sulfoxide (green circle),giving pure trans-a nd cis-isomers of the prearranged PAMs SMTP and SETP.d)Schematicr epresentation of GLP-1Ractivationw ith GLP-1(9-36)NH 2 in the presence of either trans-o rcis-SMTP.
PAMs stabilizea ctive receptor conformations and can behavea sa gonists, even in the absence of the orthosteric ligand.W hen BETP, BMTP,a nd related prearranged PAMs were appliedi np ure agonist mode, cAMP responses were greatest with trans-diastereomers (Figure 2d). There was, however,n o detectable increase in b-arrestin2 recruitment with any PAM ( Figure S2). We also investigated the ability of these compounds to liberateC a 2 + from internal stores. In the absence of GLP-1, BMTP induced strong cytosolic Ca 2 + rises, more so than BETP (Figure 2e,f). SETP and SMTP trans-isomers werea gain most effective. The effects of BETP and BMTP were likely to be GLP-1R mediated, as they were almost absent in CHO cells withoutG LP-1R overexpression ( Figure S3 a) and, like GLP-1(7-36)NH 2 -induced responses, could be reduced by inhibiting G-protein signaling intermediates including Epac2 (ESI09)a nd phospholipase C( U73122) (Figure S3 b-e). Surprisingly, BETP was the only PAMt hat enhanced the Ca 2 + response to GLP-1(7-36)NH 2 when pre-incubated, as previously described. [2a] This may reflect depletion of accessible intracellular Ca 2 + stores at ar ate dependent on the intrinsic activity of the PAM( Figure S4).
The GLP-1Ru ndergoes extensive internalization after agonist stimulation, and ongoing cAMP generation by internalized receptors may play ar ole in GLP-1Rs ignaling. [16] We hypothesized that the PAMs described here, which unlikep eptidel igands are membrane-permeating, might access and modulate the behavior of pre-internalized ligand-receptor complexes (Figure 3d). After complete GLP-1Ri nternalization with 100 nm GLP-1(7-36)NH 2 -FITC (Figure 3e)a nd washout of remaining extracellularl igand,as teady reduction in FRET was detected, indicative of ligand-receptor dissociation within endosomes. However,w hen BETP, BMTP,a nd prearrangedP AMs were applied to the cell immediately post-washout, marked reductions in dissociation from endosomal receptors were observed ( Figure 3f,g). Thiseffect was most marked with trans-SMTP.
In the present study,w ed escribe ap hotopharmacology-inspired strategy for the rational design of potent GLP-1R PAMs based upon the allostericp hotoswitch PhotoETP. [3] By restricting degrees of freedom using stilbenes,i tc ould be shown that compounds prearranged as their trans-isomer were more potent than other PAMs, including BETP. [6] Such findings likely reflect the requirement for bound PAMs to undergo fine conformational changes at the allosterics ite for full activation in response to orthosteric binding. [6b] Prearrangement may circumventt his by stabilizing molecule dynamics, allowing PAMs to adopt an allosteric 'ON' state when unbound. Moreover,r eplacemento ft he ethyl sulfoxide moiety with am ethyl sulfoxide maintained full activity,w hilst promoting cheaper and faster access to PAMs and prearranged PAMs from commercially available chemicals in high yields. Although the differences in signaling described here for the prearranged PAMs may seem small,i ts hould be noted that GLP-1R PAMs allow less active GLP-1 metabolites to signal at the orthosteric site, [2b] Figure 2. Prearranged PAMs potently enhanceGLP-1R signaling.a)Allosteric enhancemento fG LP-1(9-36)NH 2 ("9-36") cAMPr esponses (30 min incubation; n = 4) (4-parameter logistic fit shown). b) As for (a), but with exendin(9-39) (ex9-39) (n = 5). c) Allosteric enhancementof1 0mm GLP-1(9-36)NH 2 -induced b-arrestin2 (barr2) recruitment(30 min incubation; n = 5). d) cAMP responses to indicatedP AM or prearrangedP AM concentration in the absenceo forthosteric ligand (30 min incubation; n = 3). e) Cytosolic Ca 2 + responsesi nC alcium 6d ye-loaded cells, expressed relative to baseline fluorescent signal( 60 min incubation; n = 7). f) Area underc urve (AUC)d eterminedfrom (e). *P < 0.05, **P < 0.01,* **P < 0.001; one-or two-way randomized block ANOVA followedbye ither Sidak's or Tukey's post-hoc test. Values represent the mean + or AE SEM. Except where indicated, PAMs or prearranged PAMs wereapplied at 10 mm. Intriguingly,i ntracellular Ca 2 + fluxes induced by BMTP exceeded those of BETP,d espite equivalent cAMPr esponses. As such, the prearranged PAMs mayp rovide useful tools to tease apart the receptor conformationsr equired for biased signaling and, more widely, the impact of second messenger preference on relevant biological endpoints. Althougho ur studies implicated classical GLP-1R pathways, the precise mechanism(s) underlyingd ifferences in Ca 2 + responses remain unclear. The different leaving groups for BMTP versus BETP (methyl versus ethyl sulfenate) are potentialc andidates, as sulfenic acids are known signaling intermediaries in the context of cysteine modification. [17] In addition, further studies using recently described conformational FRET sensors, [18] or concentration responses to calculate alpha values, [12,19] are required to unambiguously demonstrate the differentiala ctions of the prearranged PAMs at the GLP-1R. It should be noted that the herein-described prearrangedP AMs, as well as BETP,e xist as racemates, and in the future it will be interesting to study the labeling kinetics of the separate enantiomers before chirality is lost through covalentmodification of GLP-1R.
Finally,w ep rovide the first demonstration that GLP-1R PAMs can markedly increase agonistr esidencet ime, proposed as at herapeutic strategyt od rive sustained responses in vitro and in vivo. [20] These molecules can also directly access pre-internalized GLP-1Rs to modulate ligand-receptorb inding within endosomes. They could, therefore, be used to prolong non-canonicalc AMP signaling from internalized GLP-1Rs, [16] delineate effects of membrane versus endosomal GLP-1Rs ignaling, or study the influence of occupancy on post-endocytic receptor trafficking. [21] In summary,weunveil anew class of positive allosteric modulator PAMs that are prearranged. These compounds perform bettert han their compacted stablemates (i.e. those incorporating a cis-stilbene), as well as benchmark PAM, BETP,a nd also display signal bias. Thus, prearranged PAMs provide at emplate for the production of newer and more potent allosteric modulators of the GLP-1R, with broad-applicability to GPCR research and drug discovery.