Allosteric “beta-blocker” isolated from a DNA-encoded small molecule library

Significance

The present study reports the discovery of a small-molecule negative allosteric modulator for the β₂-adrenergic receptor (β₂AR) via in vitro affinity-based iterative selection of highly diverse DNA-encoded small-molecule libraries. Characterization of the compound demonstrates its selectivity for the β₂AR and that it negatively modulates a wide range of receptor functions. More importantly, our findings establish a generally applicable, proof-of-concept strategy for screening DNA-encoded small-molecule libraries against purified G-protein–coupled receptors (GPCRs), which holds great potential for discovering therapeutic molecules.

Abstract

The β₂-adrenergic receptor (β₂AR) has been a model system for understanding regulatory mechanisms of G-protein–coupled receptor (GPCR) actions and plays a significant role in cardiovascular and pulmonary diseases. Because all known β-adrenergic receptor drugs target the orthosteric binding site of the receptor, we set out to isolate allosteric ligands for this receptor by panning DNA-encoded small-molecule libraries comprising 190 million distinct compounds against purified human β₂AR. Here, we report the discovery of a small-molecule negative allosteric modulator (antagonist), compound 15 [([4-((2S)-3-(((S)-3-(3-bromophenyl)-1-(methylamino)-1-oxopropan-2-yl)amino)-2-(2-cyclohexyl-2-phenylacetamido)-3-oxopropyl)benzamide], exhibiting a unique chemotype and low micromolar affinity for the β₂AR. Binding of 15 to the receptor cooperatively enhances orthosteric inverse agonist binding while negatively modulating binding of orthosteric agonists. Studies with a specific antibody that binds to an intracellular region of the β₂AR suggest that 15 binds in proximity to the G-protein binding site on the cytosolic surface of the β₂AR. In cell-signaling studies, 15 inhibits cAMP production through the β₂AR, but not that mediated by other Gs-coupled receptors. Compound 15 also similarly inhibits β-arrestin recruitment to the activated β₂AR. This study presents an allosteric small-molecule ligand for the β₂AR and introduces a broadly applicable method for screening DNA-encoded small-molecule libraries against purified GPCR targets. Importantly, such an approach could facilitate the discovery of GPCR drugs with tailored allosteric effects.

G-protein–coupled receptors (GPCRs), also known as seven transmembrane receptors, represent the largest family of cellular receptors and the most common therapeutic drug targets. Accordingly, GPCRs are the subject of intensive research, both in academia and the pharmaceutical industry, aimed at elucidating their structures, detailed mechanisms of action, and discovery of novel ligands with therapeutic potential (1–3). To date, the overwhelming majority of GPCR drugs target the orthosteric site on the receptors. This is the binding site of endogenous ligands, which generally faces the extracellular surface of the receptor (4, 5). However, in recent years, functionally active allosteric ligands, which bind outside the orthosteric site, have also been discovered. Allosteric ligands that augment or reduce the binding affinity and/or functional responses of orthosteric ligands are referred to as positive or negative allosteric modulators (PAMs or NAMs), respectively (5). Such allosteric ligands hold great therapeutic promise due to their enhanced selectivity among receptor subtypes compared with orthosteric drugs targeting the same subtype. The first approved allosteric drugs for GPCRs target chemokine CCR5 (6) and calcium-sensing receptors (7) to treat HIV infections and hyperparathyroidism, respectively, with many more modulators in preclinical development. Allosteric modulators also have great utility as tool compounds in biophysical studies as they are able to lock receptors into specific conformations by virtue of their cooperative interactions with orthosteric ligands (4, 5).

In the past, screening for GPCR ligands, either allosteric or orthosteric, has been cumbersome and labor-intensive. Such screens have generally been based on the functional ability of compounds to either stimulate or block receptor-mediated activities in whole-cell–based settings (8). A more rapid and efficient approach is to use interaction-based methods for initial screening wherein large libraries of molecules are panned against the target receptor. However, until recently, such libraries have consisted only of macromolecules such as phage-displayed antibodies (9, 10) or RNA aptamers (11). Another approach, which has not yet been widely applied, is screening of DNA-encoded small-molecule libraries (DELs). In this approach, remarkably large combinatorial libraries consisting of up to billions of small-molecule compounds are displayed on DNA fragments that serve as barcodes for their subsequent identification (12, 13). In the past few years, application of the DEL screening technology for soluble proteins has produced inhibitors against cancer and immune disorders and against therapeutic targets. These are protein kinases such as Src, MK2, Akt3, Pim1, Aurora A kinase, p38 mitogen-activated protein (MAP) kinase, and antiapoptotic protein Bcl-xL (14, 15). Expanding such technology to GPCRs has the potential to yield both orthosteric and allosteric ligands. However, its application to GPCR screening has been challenging, largely because of difficulties associated with preparing appropriate receptor targets as well as the membrane-bound nature of the receptors, which can lead to nonspecific interactions. To date, identification of a compound that inhibits the NK₃ tachykinin receptor by screening against the receptor expressed in whole cells represents the only successful application of DEL technology against GPCRs (16).

The β₂-adrenergic receptor (β₂AR) has served as the model system for molecular studies of ligand-binding GPCRs for over 40 y (1) and plays a significant role in cardiovascular and pulmonary diseases. So-called “β-blockers,” which are orthosteric antagonists of the receptor, are mainstays of cardiovascular medicines used to treat a wide variety of illnesses (17–19). On the other hand, β-agonists have proven very effective against asthma (20). However, all known β-adrenergic ligands act orthosterically; thus, it is possible that allosteric modulators would possess enhanced therapeutic efficacy, selectivity, or even unique therapeutic properties such as signaling bias. Such ligands would also facilitate the isolation and characterization of specific receptor conformations for biophysical studies. Accordingly, here we set out to isolate allosteric ligands for the β₂AR using DELs. We report isolation of a small-molecule negative allosteric modulator (antagonist) for the β₂AR and provide a detailed characterization of its pharmacological properties and interaction with the receptor.

Results

Isolation of Compound 15 from DELs.

To identify unique chemotypes that bind at structurally relevant sites on the surface of β₂AR, we screened DELs against purified, unliganded β₂AR maintained in the detergent n-dodecyl-β-d-maltoside (DDM) (Fig. 1A and Fig. S1). A similar beads-only selection was performed in parallel as a control. In total, we screened three libraries together containing approximately a total of 190 million unique compounds synthesized using a DNA-tagged, split-and-pool combinatorial chemical synthesis approach (Chemetics; Nuevolution). In each library, ∼5 × 10¹³ molecules in total were used as input, and 1–7 × 10⁷ molecules remained after screening. Relative quantification of the recovered compounds was achieved by a combination of PCR amplification and next-generation sequencing of eluted DNA barcodes, followed by computational decoding approaches. To refine the output and eliminate potential nonspecific binders, compounds that displayed less than a 260- to 470-fold increase in frequency from baseline as well as those that were observed in bead-only control selections were filtered from the dataset, leaving a total of 394 potential β₂AR binders for further analysis (Table S1). These compounds were then clustered based on their structural similarity, and 16 putative hits were selected as representatives for these clusters. DNA-tagged versions of these 16 hits were resynthesized and screened individually to evaluate their influence on the binding affinity of orthosteric agonists in radioligand binding assays with membranes obtained from β₂AR-overexpressing cells. One compound [4-((2S)-3-(((S)-3-(3-bromophenyl)-1-(methylamino)-1-oxopropan-2-yl)amino)-2-(2-cyclohexyl-2-phenylacetamido)-3-oxopropyl)benzamide], designated as compound “15” (Fig. 1B), markedly decreased orthosteric agonist binding to the receptor and was thus selected for further characterization.

Fig. 1.

Screening of the DEL and the chemical structure of 15. (A) Schematic illustration of the screening of a DNA-encoded compound library. DNA-encoded library molecules, synthesized using a DNA-tagged, split-and-pool combinatorial chemical synthesis approach, were mixed with a target (purified β₂AR) immobilized on a matrix. Target binding (active) molecules were collected through affinity-based selection, and the encoding DNA tags were sequenced to identify the binding molecules. (B) Compound 15 is composed of three building blocks: methybenzamide (green), bromo-benzyl (red), and cyclohexylmethyl-benzene (blue). The amide backbone is shown in black.

Fig. S1.

Schematic illustration of the DEL selection procedure for isolation of β₂AR binders. DNA-encoded small molecules in a library were incubated with either 25 μg of detergent (DDM)-solubilized, Flag-tagged β₂AR immobilized on M1 anti-Flag beads, or a free-beads control for 1 h at room temperature (RT), while vigorously agitated. After the beads were repeatedly washed (five times), target-binding molecules were eluted by incubation at 72 °C for 30 min, as described in SI Materials and Methods, Affinity Selection. Denatured proteins were removed from the eluted sample, which was subsequently desalted. After the molecule number in the eluted and cleaned sample was determined by Q-PCR as described in SI Materials and Methods, the sample was subjected to the next round of selection for five times and subsequently one round of negative selection by incubating with free beads. The flow-through sample from the negative selection was subjected to the next-generation DNA sequencing to identify the binding molecules by their DNA tags.

Table S1.

Screening summary

Library ID	Library size, in millions	Valid sequence reads*	Minimum count/enrichment†	Total hits	Resynthesized compounds
DL5c_260	96	1,011,059	≥5/475-fold	39	4
DL7a_623	85	982,396	≥5/433-fold	175	8
DL11a_629	9	866,825	≥25/260-fold	180	4

^*Each full-length sequence read was matched against a sequence database representing all possible library encoding combinations to translate each DNA molecule into its corresponding set of chemical identifiers. Only those reads for which all encoding elements could be conclusively identified were considered valid (excluding reads that represented PCR-derived duplicates).

^†“Minimum Count/Enrichment” is an arbitrary cutoff used to increase the ratio of signal to noise in the postselection dataset by filtering out compounds that were observed infrequently and generally shared no structural similarities. “Count” represents the number of valid sequence reads observed for a given compound. “Enrichment” represents the calculated change in frequency from baseline for each compound, where baseline was assumed to equal 1/“library size.”

Characterization of Compound 15 for Its Binding to the β₂AR.

To further characterize the pharmacological properties of 15, we synthesized it in its DNA-free form (Fig. 1B). We first performed competition binding experiments with the radioiodinated antagonist cyanopindolol (¹²⁵I-CYP) to evaluate the influence of 15 on the binding ability of orthosteric ligands to the β₂AR reconstituted into high-density lipoprotein (HDL) (also known as nanodisc) particles (21). We found that, although 15 had little to no influence on binding of ¹²⁵I-CYP, which is a neutral orthosteric antagonist of the β₂AR, it robustly decreased the binding affinity of the agonist isoproterenol for the receptor in a dose-dependent fashion (Fig. 2A). Compound 15 caused the isoproterenol competition curve to shift to the right by close to one log (ninefold) at the maximal concentration tested. The half-maximal concentration of 15 for this shift was ∼1.9 μM (Fig. 2B). On the other hand, when the inverse agonist [³H]-ICI-118,551 was used as an orthosteric ligand tracer, 15 significantly increased (18 ± 1% max) its binding to the β₂AR in a dose-dependent manner (Fig. 2C) with an EC₅₀ value of 0.48 μM, which is close to that obtained in Fig. 2B (1.9 μM). This reflects positive cooperativity between 15 and the orthosteric antagonist [³H]-ICI-118,551 for binding to the receptor. To validate the direct binding of 15 to the β₂AR, we performed isothermal titration calorimetry (ITC). By this technique, we found that the equilibrium dissociation constant (K_d) of 15 for the receptor is 1.7 ± 0.8 μM, and the stoichiometry of the interaction is 1, suggesting that 15 binds to one site on the β₂AR (Fig. 2D). The K_d value obtained by ITC is in good agreement with the half-maximal concentration of the shift obtained in the radioligand competition binding experiment (Fig. 2B) as well as the EC₅₀ value obtained from the 15 titration curve for [³H]-ICI-118,551 binding (Fig. 2C).

Fig. 2.

Characterization of 15 for its binding to the β₂AR. (A) Dose–response curves of isoproterenol (ISO) competition binding to the β₂AR reconstituted in nanodiscs with ¹²⁵I-CYP were obtained in the presence of various concentrations of 15 as indicated. Values were expressed as percentages of the maximal ¹²⁵I-CYP binding level obtained from a one-site competition binding-log IC₅₀ curve fit of the vehicle [0.9% dimethyl sulfoxide (DMSO)] control data. Points on curves represent mean ± SEM obtained from at least three independent experiments done in duplicate. (B) The half-maximal concentration of 15 in the changes of isoproterenol competition binding was obtained from a dose–response curve replotted with the data set at 0.1 μM isoproterenol with various doses of 15 in A. (C) Dose-dependent increases in inverse agonist ³H-ICI-118,551 (ICI) binding to the β₂AR in nanodiscs. Points on the curve represent normalized values as percentages of the ³H-ICI-118,551 binding amount in the absence of 15 and mean ± SEM obtained from at least four independent experiments done in duplicate. (D) Characteristics of 15 for its physical interaction with the β₂AR were determined by the isothermal titration calorimetry (ITC) analysis with the detergent-solubilized, purified receptor. The thermogram (Top) and binding isotherm with the best titration curve fit (Bottom) shown are representatives of three independent experiments. Values represent mean ± SEM. (E) Extent of nanobody-60 (Nb60) binding to the β₂AR determined by ELISA in the presence of different ligands including 15. Values were expressed as ratios of the level of Nb60 binding in the vehicle (0.5% DMSO) control sample and represent mean ± SEM obtained from three independent experiments done in duplicate. BI, BI-167107. (F) The levels of ³H-Fen binding to the β₂AR upon treatment with the vehicle control (0.5% DMSO) or 15 at 50 μM in the absence or presence of transducers, either trimeric G_αβγ protein or β-arrestin1 (β-arr1) together with Fab30. Values were expressed as fold changes of the level of ³H-Fen binding in the vehicle (DMSO) control sample without the transducer and represent mean ± SEM obtained from three independent experiments done in duplicate. (G) Compound 15 dose-dependent decreases in the level of ³H-Fen high-affinity binding to the β₂AR promoted by either G_αβγ protein or β-arr1 together with Fab30. Values were expressed as percentages of the maximal ³H-Fen binding level promoted by each transducer in the vehicle control (0.5% DMSO) and represent mean ± SEM obtained from at least three independent experiments done in duplicate. All of the statistical analyses in the figure were performed, as described in Materials and Methods.

To further confirm that 15 favors an inactive conformational state of the receptor, we assessed the extent of binding of an inactive conformation-specific β₂AR single domain antibody [nanobody-60 (Nb60)] that binds to an intracellular region of the receptor (10, 22). Contrary to the positive cooperativity predicted to occur between a negative allosteric modulator and Nb60 (22), we found a decrease in Nb60 binding to the receptor in the presence of 15 (Fig. 2E). This suggests that 15 competes with Nb60 for binding to the receptor, and thus that their binding sites at least partially overlap, indicating that 15 binds to an intracellular region of the β₂AR.

It has been well appreciated that allosteric transducers such as heterotrimeric G proteins or β-arrestins promote high-affinity agonist binding (10, 23, 24). We also observed that 15 inhibited the high-affinity binding of a radiolabeled β₂AR-agonist, [³H](R,R′)-4-methoxyfenoterol (³H-Fen) (25) that is promoted by either heterotrimeric Gs protein or β-arrestin1 (Fig. 2F). This finding also suggests that 15 negatively modulates activation of the β₂AR by agonists. Furthermore, the dose-dependent effects of 15 in ³H-Fen binding assays conducted in the presence of either G protein or β-arrestin were similar to each other (Fig. 2G). We used Fab30, which stabilizes an active conformation of β-arrestin1 (26, 27), to enhance the weak high-affinity ³H-Fen binding signal induced by β-arrestin to levels comparable to those observed with G protein. The EC₅₀ values obtained from the 15 dose–response curves for the two transducer-promoted signals were comparable (2.8 vs. 2.2 μM, respectively, for G protein and β-arrestin; Fig. 2G). This suggests that 15 displays no strong “bias” between its inhibitory activities in the two transducer-induced high-affinity ³H-Fen binding signals. The EC₅₀ values of 15 obtained here (Fig. 2G) are also in good agreement with its K_d measured by ITC (Fig. 2D). Taken all together, the results in Fig. 2 show that 15 behaves as a negative allosteric modulator for the β₂AR, and suggest that it binds at the cytoplasmic surface of the receptor.

Functional Modulation of β₂AR Activity by Compound 15.

Next, we investigated the effects of 15 on β₂AR function in cells by measuring G-protein–mediated cAMP production (28, 29) and β-arrestin recruitment to the receptor (29, 30). Due to the high signal amplification of the G-protein activation assay compared with β-arrestin recruitment, to achieve comparable signaling outputs between the two, we used endogenously expressed β₂AR in the reporter cells to measure cAMP production, but we used stably overexpressed β₂V₂R for monitoring β-arrestin recruitment. The β₂V₂R, a chimeric receptor with a V₂R tail at the C terminus, displays stronger and more stable agonist-promoted β-arrestin binding than the native β₂AR while retaining the pharmacological properties of the native β₂AR (31). Compound 15 decreased the isoproterenol-stimulated responses in both assays (Fig. 3 A and B). We observed rightward shifts of the EC₅₀ values, as well as decreases in the maximal level of the stimulated responses in the presence of increasing concentrations of 15, indicating that 15 inhibits β₂AR agonist-induced functional responses. Additionally, the extent of rightward shift of isoproterenol potency promoted by 15 was similar for both G-protein activation and β-arrestin recruitment to activated receptor. On the other hand, the decreases in the maximal response induced by 15 were more robust in β-arrestin recruitment than in cAMP production (Fig. 3 A and B). This likely is attributed to the differences in the sensitivity between the assays, as described before (29), rather than to any biased activity of 15. Such unbiased inhibition of β₂AR activity by 15 is consistent with the finding that 15 similarly inhibits the high-affinity binding of ³H-Fen promoted by either transducer (G protein or β-arrestin) (Fig. 2G). These results confirm that 15 modulates β₂AR by antagonizing its agonist-induced activity.

Fig. 3.

The effect of 15 on β₂AR-mediated functional activities. After pretreatment with 15 for 20 min at various concentrations as indicated, the β₂AR-mediated activities in cells were measured upon stimulation with isoproterenol (ISO) in a dose-dependent manner: (A) cAMP production by the endogenously expressed β₂AR and (B) β-arrestin recruitment to the exogenously expressed β₂V₂R. Values were expressed as percentages of the maximal level of the isoproterenol-induced activity in the vehicle (0.5% DMSO) control. Points on curves represent mean ± SEM obtained from four independent experiments done in duplicate. All of the statistical analyses in the figure were performed as described in Materials and Methods.

We further validated that the inhibitory effect of 15 in functional assays was a result of specific inhibition at the receptor level and not due to nonspecific effects by testing 15 in other cell-based assays that have distinct readouts from the luminescence-based assays shown in Fig. 3. We monitored the effect of 15 on cAMP production upon stimulation of the overexpressed β₂AR using a fluorescence resonance energy transfer (FRET)-based biosensor, ICUE2 (32). In this experiment, 15 caused a rightward shift in the EC₅₀ for isoproterenol-stimulated cAMP production (Fig. S2A), consistent with that shown in Fig. 3A, albeit without changing the maximal response. This may be attributed to the increased amplification of the signal due to overexpression of the β₂AR in this assay compared with the signal by the receptor expressed at endogenous levels in Fig. 3A. Similarly, we also confirmed the inhibitory effect of 15 on agonist-stimulated β-arrestin recruitment to the β₂AR by assessing FRET signals between β₂AR-YFP and CFP-β-arrestin2 (Fig. S2B). We further confirmed the inhibitory effect of 15 on β₂AR activation using an in vitro GTPase activity assay with the β₂AR reconstituted into HDL particles together with purified Gs protein (Fig. S2C). The 15-induced shift in the EC₅₀ for isoproterenol obtained in this in vitro assay is comparable to that observed in the cell-based assays, suggesting that 15 must readily penetrate cell membranes to bind to an intracellular region of the β₂AR, which is suggested by the result in Fig. 2E.

Fig. S2.

Alternative functional assays confirming the inhibitory activity of 15. After pretreatment with 15 at different concentrations as indicated for 20 min, several dose-dependent isoproterenol (ISO)-induced activities were monitored. (A) The level of cAMP was monitored in cells stably expressing the FRET-based bio-sensor (ICUE2) and the β₂AR. (B) The extent of β-arrestin recruitment was determined by FRET signals between stably expressed CFP-β-arrestin2 and β₂AR-YFP in HEK-293 cells. (C) The Gs-protein activation was measured by GTP hydrolysis in vitro, with the β₂AR reconstituted in HDL particles and purified Gs proteins. Values were expressed as percentages of the maximal level of the isoproterenol-induced activity in the vehicle (0.5% DMSO) control. Points on curves represent mean ± SEM obtained from at least three independent experiments done in duplicate. Statistical analyses were performed as described in SI Materials and Methods.

Selective Inhibition of β₂AR-Mediated Activities by Compound 15.

To confirm the selectivity of 15 for the β₂AR, we performed functional assays to evaluate whether 15 inhibited activation of other members of the GPCR family closely or distantly related to the β₂AR. First, we compared the extent of 15 blockade of agonist-stimulated activities of β₂AR with two other endogenously expressed Gs-coupled receptors in HEK-293 cells, the prostaglandin E2 (PGE2) and vasoactive intestinal peptide (VIP) receptors. Notably, 15 had no effect on cAMP production following stimulation of these receptors (Fig. S3 A and B). Additionally, we looked at the specificity of 15 for the β₂AR by assessing its inhibitory effect on β-arrestin recruitment to other receptors besides the chimeric β₂AR (β₂V₂R). Here, we used the parental cell line stably expressing β-arrestin2 alone, and transiently expressed receptors indicated in Fig. 4. First, we confirmed that the extent of 15-mediated inhibition of agonist-stimulated β-arrestin recruitment to the wild-type β₂AR (Fig. 4A) is comparable to that obtained in the transiently expressed chimeric receptor, β₂V₂R (Fig. 4B). On the other hand, following stimulation of the β₁AR, a receptor closely related to the β₂AR, 15 substantially inhibited the maximal response as well as the basal activity in a concentration-dependent manner, whereas no significant changes were observed in the EC₅₀ value (Fig. 4C). We also observed significant, but much reduced, inhibitory effects of 15 on β-arrestin recruitment to the vasopressin V₂ receptor (V₂R), which is also a Gs-coupled receptor (Fig. 4D). In contrast to this, 15 only minimally inhibited agonist-induced β-arrestin recruitment to the VIPR, another Gs-coupled receptor. We observed only minimal decreases in the maximal response to stimulation with VIP without any change in the EC₅₀ value induced by 15 (Fig. 4E), consistent with the result obtained in cAMP accumulation (Fig. S3B). Furthermore, no significant inhibition by 15 was detected in β-arrestin recruitment to the Gq-coupled angiotensin II type 1 receptor (AT₁R) (Fig. 4F). To further assess the extent of the 15 inhibitory activities among different receptors, we quantified the level of 15-mediated decreases in the maximal response as well as shifts of the EC₅₀ value exhibited as fold shifts (Table S2). These results demonstrate that the inhibitory effect of 15 on agonist-stimulated responses is greatest for the β₂AR and is substantially diminished in even closely related receptors such as β₁AR.

Fig. S3.

No inhibitory activity of 15 in cAMP production mediated by other receptors. After pretreatment with 15 at different concentrations for 20 min, cAMP production was monitored upon stimulation of endogenously expressed (A) prostaglandin E2 (PGE₂) and (B) vasoactive intestinal peptide (VIP) receptors. Values were expressed as percentages of the maximal level of the activity induced by the agonist of each receptor in the vehicle (0.5% DMSO) control. Points on curves represent mean ± SEM obtained from at least three independent experiments done in duplicate. Statistical analyses were performed as described in SI Materials and Methods.

Fig. 4.

Specificity of 15 inhibition for β₂AR-mediated activity. Various receptors were transiently expressed to monitor β-arrestin recruitment, including (A) β₂AR, (B) β₂V₂R, (C) β₁AR, (D) V₂R, (E) VIPR, and (F) AT₁R. After pretreatment with 15 at different concentrations as indicated for 20 min, the extent of agonist-induced β-arrestin recruitment to these receptors was determined in a dose-dependent manner. Values were expressed as percentages of the maximal level of the activity induced by the agonist of each receptor in the vehicle (0.5% DMSO) control. Points on graphs represent mean ± SEM obtained from at least three independent experiments done in duplicate. AngII, angiotensin II; AVP, arginine vasopressin; ISO, isoproterenol; VIP, vasoactive intestinal peptide.

Table S2.

Compound 15-mediated allosteric inhibition of β-arrestin recruitment and endocytosis by different receptors

Receptors	β-Arrestin recruitment		β-Arrestin endocytosis
	Inhibition Emax, %	EC50 shift, fold	Inhibition Emax, %	EC50 shift, fold
β2AR	32.6 ± 4.8***	8.0 ± 1.4***
β2V2R	30.6 ± 4.7***	7.1 ± 1.4***	30.1 ± 4.3***	10.5 ± 1.3***
β1AR	30.6 ± 8.6***	1.2 ± 1.8***
V2R	13.9 ± 4.0***	3.4 ± 1.3***	11.0 ± 3.8***	3.9 ± 1.2***
VIPR	18.4 ± 3.6***	1.1 ± 1.2***
AT1R	10.9 ± 10.2***	0.8 ± 1.9***	10.5 ± 3.4***	0.8 ± 1.2

The values of 15-induced maximal “Inhibition E_max, %” and “EC₅₀ shift, fold” were obtained from the data presented in Fig. 4 and Fig. S4. Each value represents mean ± SEM. Statistical analyses were performed as described in SI Materials and Methods. ***P < 0.001 compared with the control value obtained in the presence of the vehicle (DMSO).

To obtain further insights into the specificity of 15 for the β₂AR, we examined its inhibitory activity on agonist-induced β-arrestin internalization. Unlike “class B” receptors, including the V₂R and the AT₁R whose tight interactions with β-arrestin allow for their cointernalization, “class A” receptors such as the β₂AR have weaker β-arrestin interactions and are not cointernalized with β-arrestin (33). Therefore, we examined the effect of 15 on this functional activity with the transiently expressed β₂V₂R (Fig. S4A), V₂R (Fig. S4B), and AT₁R (Fig. S4C). The extent of β-arrestin internalization was monitored by measuring the amount of β-arrestin targeted to endosomes (34). Results obtained in this assay are consistent with the inhibitory effects of 15 on β-arrestin recruitment to activated receptors as shown in Fig. 4, which are summarized in Table S2, further confirming the specificity of the modulating activity of 15 for the β₂AR.

Fig. S4.

The inhibitory activity of 15 in agonist-induced β-arrestin endocytosis. After pretreatment with 15 at different concentrations for 20 min, the level of β-arrestin endocytosis (Endo), upon stimulation of (A) β₂V₂R, (B) V₂R, and (C) AT₁R with respective agonists in a dose-dependent manner, was monitored. Values were expressed as percentages of the maximal level of the activity induced by the agonist of each receptor in the vehicle (0.5% DMSO) control. Points on curves represent mean ± SEM obtained from at least three independent experiments done in duplicate. AngII, angiotensin II; AVP, arginine vasopressin; ISO, isoproterenol.

We also investigated whether the NAM activity of 15 at the β₂AR was dependent on the presence of a specific agonist at the orthosteric site (i.e., probe dependence). We performed this by monitoring Gs-mediated cAMP production and β-arrestin recruitment to the receptor in the presence of orthosteric probes ranging from full to weak partial agonists, including epinephrine, fenoterol, and clenbuterol (Fig. S5) as done with isoproterenol (Fig. 3). We also compared competition binding of these agonists to the β₂AR with ¹²⁵I-CYP in the presence of different concentrations of 15 (Fig. 2A and Fig. S5), which allowed us to assess the probe dependence of 15 among the agonists in the absence of transducer coupling. Table S3 shows the summary of quantified values in each assay, including the extent of 15-mediated decreases in the maximal response and shifts of the EC₅₀ value exhibited as fold shifts. Overall, 15 appears to display no significant probe dependence among the tested agonists. We observed that the extent of the EC₅₀ value shift by 15, which is consistent among the tested assays, follows the efficacy of the tested agonists. On the other hand, the magnitude of 15 inhibition of the maximal response is negatively correlated with the efficacy of these agonists.

Fig. S5.

Compound 15-mediated inhibition in agonist-induced signals upon stimulation of the β₂AR with a range of agonists. The level of 15-induced inhibition of β₂AR-mediated signals upon stimulation with (A and B) epinephrine (Epi); (D and E) fenoterol (Fen); and (G and H) clenbuterol (Clen) was determined by monitoring cAMP production (A, D, and G) and β-arrestin recruitment (B, E, and H) as described for Fig. 3. Values were expressed as percentages of the maximal level of the isoproterenol (ISO)-induced activity in the vehicle (0.5% DMSO) control. (C, F, and I) The extent of 15-induced curve shifts in ¹²⁵I-CYP competition binding with Epi (C), Fen (F), and Clen (I) was also determined as essentially described for Fig. 2A. Points on curves represent mean ± SEM obtained from at least three independent experiments done in duplicate.

Table S3.

Compound 15-mediated allosteric inhibition of β₂AR signals stimulated with different agonists

Agonists	G-protein cAMP		β-Arrestin recruitment		125I-CYP competition binding
	Inhibition Emax, %	EC50 shift, fold	Inhibition Emax, %	EC50 shift, fold	EC50 shift, fold
Isoproterenol	31.5 ± 2.5***	7.1 ± 1.2***	61.7 ± 2.0***	7.4 ± 1.2***	9.1 ± 1.1***
Epinephrine	36.5 ± 5.1***	7.4 ± 1.3***	71.8 ± 2.4***	5.9 ± 1.2***	5.8 ± 1.1***
Fenoterol	55.7 ± 3.4***	2.6 ± 1.3***	81.2 ± 3.2***	2.1 ± 1.4***	1.9 ± 1.1***
Clenbuterol	78.1 ± 6.9***	No shift	88.4 ± 21.5***	No shift	1.1 ± 1.1***

The values of 15-induced maximal “Inhibition E_max, %” and “EC₅₀ shift, fold” were obtained from the data presented in Fig. 2A (for competition binding with isoproterenol), Fig. 3 (for functional data with isoproterenol), and Fig. S5 (for all of the data with the other agonists). Each value represents mean ± SEM. Statistical analyses were performed as described in SI Materials and Methods. ***P < 0.001 compared with the value from the vehicle (DMSO)-treated control sample.

Structure–Activity Relationships of Compound 15 Analogs at the β₂AR.

To discern the structure–activity relationship (SAR) pattern for the allosteric modulation of 15 at the β₂AR, we designed and synthesized a series of 15 derivatives (Table S4). We assessed the ability of these derivatives to modulate β₂AR functions in two different types of experimental settings. These were cell-based activity assays, including G-protein–mediated cAMP production and β-arrestin recruitment to the activated β₂AR, as well as high-affinity binding of the agonist ³H-Fen to the receptor induced by transducers, Gs or β-arrestin. To assist our SAR analyses, 15 was divided into three structural subunits, the methylbenzamide (region I), bromo-benzyl (region II), and cyclohexylmethyl-benzene (region III) regions, into each of which we introduced modifications. We found that the formamide group in region I (methylbenzamide) is an important determinant of functional properties of 15. Removal of this group on the phenyl ring (A1) led to a dramatic decrease in the inhibitory activity of 15 down to about 20% or less of its original activity. The same, but less severe, trend was observed when the position of this formamide group was changed from its original para-position to a meta-position (A2), which resulted in a ∼60% reduction of its original activity. In the case of region II (bromo-benzyl), removal of the electronegative atom bromine (A3) also caused variable but substantial attenuation of the inhibitory activity of 15 down to about 10–55%. Two other modifications in this region, replacement of bromine with fluorine, an atom of comparable electronegativity but smaller radius (A4), and introduction of additional bromine at the meta-position of the phenyl ring (A5) modestly decreased the functional effects of 15. Next, we evaluated the activity of derivatives with modifications on the aromatic ring in region III (cyclohexylmethyl-benzene). Interestingly, addition of a hydroxyl group to this ring at the para-position (A6) led to dramatic loss of inhibitory activity, while replacing the hydroxyl group at this position with a slightly hydrophobic methoxy group (A7) partially restored the inhibitory activity. This strongly suggests that the hydrophobic nature of this region is another important determinant for efficient interaction of 15 with the presumably hydrophobic portion of the putative binding site of 15 on the β₂AR.

Table S4.

Structure–activity relationships of 15 analogs

For each analog, only the modified region is illustrated. No illustration represents no change in the indicated region. Values are expressed as percentages of the 15 blockade level in each assay and represent mean ± SEM obtained from at least three independent experiments. Statistical analyses were performed as described in SI Materials and Methods. **P <0.01; ***P <0.001 compared with the value obtained with 15, which is normalized to 100%.

Discussion

We report here the discovery and characterization of a small molecule, compound 15, as an allosteric β-blocker. Compound 15 was derived from an in vitro affinity-based screening of DELs against the purified human β₂AR. Compound 15 shares no structural or chemical similarities with known β₂AR orthosteric ligands, and it does not compete with radiolabeled β₂AR ligands for binding at the orthosteric site. On the other hand, it binds allosterically to the β₂AR with low micromolar affinity. The compound negatively modulates the binding of agonists to the β₂AR while it clearly displays positive cooperativity with an orthosteric inverse agonist. In addition, in cell-based functional assays, 15 displays robust inhibition of β₂AR agonist-promoted, Gs-mediated cAMP generation as well as β-arrestin recruitment to the receptor. Together, these characteristics demonstrate that 15 allosterically binds to and stabilizes an inactive conformation of the β₂AR, which are the classic hallmarks of a negative allosteric modulator (5, 35).

A large number of orthosteric ligands for the β₂AR have been developed, whereas before this study, no allosteric small-molecule β₂AR ligand had been identified. The affinity-based screening strategy is an ideal way to identify allosteric ligands for a receptor, and DEL screening is an innovative strategy to perform affinity-based selections against targets that are isolated or expressed on whole cells (12). Although this technique enables an unprecedented increase in the size of libraries that can be screened compared with conventional activity-based screening formats, its use has mostly been limited to soluble protein targets (12, 13). Due to the inherent difficulty in isolating functional membrane proteins, this technique has been only rarely used to obtain ligands for GPCRs. To date, there has been only one report describing the discovery of a ligand for a GPCR from a DEL (16). There, the recombinant NK₃ tachykinin receptor expressed on HEK-293 cells was used as a target in a whole-cell selection format to identify an inhibitor for the receptor (16). Here, we have demonstrated that the DEL screening strategy can be successfully applied to the isolation of small-molecule ligands using a purified GPCR. Although our study was focused on isolating and characterizing β₂AR allosteric ligands, this strategy could be used as well to isolate ligands that target orthosteric sites of GPCRs. Despite the power of this approach, predicting the functional outcomes of the isolated compounds remains empirical. However, in our in vitro purified receptor target system, it is highly feasible to bias the selections through differential display of the receptor in unique conformations (e.g., agonist vs. antagonist vs. no ligand in the orthosteric site) or in complex with signaling partners such as G proteins or β-arrestins. This should provide allosteric modulators with distinct properties (e.g., NAMs, PAMs, or even biased molecules for coupling to transducers, leading to signaling bias).

To date, pharmacological studies of GPCR allosteric modulation have been restricted to a few receptor families including muscarinic acetylcholine, adenosine, chemokine, and metabotropic glutamate receptors (5, 36). Compared with orthosteric ligands, drugs targeting allosteric sites often display greater receptor subtype selectivity and therefore potentially reduce adverse side effects (4, 5). This is presumably due to decreased evolutionary pressure at allosteric sites than at the orthosteric site of GPCRs (35), leading to their greater divergence within a family. Moreover, multiple allosteric sites can exist on a given receptor (37). In addition, allosteric GPCR modulators may have greater potential than orthosteric ligands to engender biased signaling through selective modulation of specific signaling pathways, for example, G-protein versus β-arrestin pathways (4).

Our SAR studies provide insights into key regions of 15 that must engage in contacts with the allosteric binding site on the β₂AR to allow its functional modulation. We found some alterations, including complete deletions of the formamide group in region I and bromine in region II, lead to dramatic decreases in the functional activities of the parent compound, 15. We also observed a positive association between increased polarity of region III and loss of functional activity. This suggests that this region of the molecule might interface deep within the β₂AR allosteric site to establish contacts with core hydrophobic residues.

As with other GPCRs, several putative allosteric sites on the β₂AR have been recently proposed based on crystal structures (37). Some are located at the intracellular face of the receptor; these are relatively large and can accommodate a wide range of compound sizes. Interestingly, most of the currently reported non–small-molecule β₂AR allosteric modulators, such as nanobodies (10, 22, 38) and RNA aptamers (11), bind to intracellular cavities that overlap with the G-protein binding site (39). This appears to be true as well for 15 because it competes for β₂AR binding with a nanobody (Nb60) that favors an inactive conformation and that binds to this intracellular region of the β₂AR (22). Although our findings suggest that 15 binds to the intracellular region of the β₂AR, further SAR and structural studies at atomic-level resolution will be required to precisely define the site and mechanism of action, by which 15 acts as a NAM.

In summary, our study reports the discovery via in vitro affinity selection of a DEL against purified receptors and functional characterization of a β₂AR-selective negative allosteric modulator. Our findings suggest that targeting GPCR allosteric sites with such combinatorial small-molecule libraries provides a powerful and efficient approach for developing highly selective ligands that can modulate a wide range of receptor’s functional activities. Furthermore, our findings establish a proof-of-concept strategy using the DEL screening technique, which can be broadly applied to discover small molecules for other GPCRs.

Materials and Methods

Complete details and descriptions of the materials used; cell culture and transfections; expression and purification of the β₂AR; purified β₂AR-based DEL selection, quantitative PCR, and next-generation sequencing analysis; reconstitution of the β₂AR into HDL particles; radioligand binding; ITC; ELISA; measurement of cAMP accumulation and in vitro GTPase activity; β-arrestin recruitment and β-arrestin endocytosis assays; data analyses; and synthesis and characterization of compounds are provided in SI Materials and Methods.

SI Materials and Methods

Materials.

All of the orthosteric β₂AR ligands used here except BI-167107 and angiotensin II were purchased from Sigma-Aldrich. BI-167107 was synthesized as described previously (40). [Arg8]-Vasopressin (AVP) was purchased from GenScript. Prostaglandin E₂ (PGE₂) and vasoactive intestinal peptide (VIP) were obtained from Tocris. Compound 15 and its analogs were synthesized as described below. All compounds were sourced at 95% or greater purity. All of the synthesized compounds were further tested for purity by LC/MS at Changzhou University, China, and were found to be pure as judged by peak height and identity. DNA-encoded libraries screened here were provided by Lexicon Pharmaceuticals. [³H](R,R′)-4-methoxyfenoterol was provided by Irving Wainer (Laboratory of Clinical Investigation, National Institute on Aging Intramural Research Program, Bethesda, MD). Heterotrimeric Gs protein (39), β-arrestin1, and Fab 30 (26, 27) were purified as described before.

Expression and Purification of the β₂AR.

Full-length human β₂AR containing an N-terminal FLAG epitope tag, C-terminal 6×His-tag, and a N187E glycosylation mutation was expressed in Sf9 insect cells using the BestBac Baculovirus Expression System as previously described (41). Briefly, cells were infected with β₂AR baculovirus at a density of 3 × 10⁶ cells per mL. Next, cells were harvested 67 h later and solubilized using a buffer containing 1% n-dodecyl-β-d-maltoside (DDM) (Anatrace), 20 mM Hepes, pH 7.4, 150 mM NaCl, and protease inhibitors. Functional β₂AR was obtained using an affinity-chromatographic procedure involving a first M1 anti-FLAG antibody affinity column, followed by alprenolol-ligand column and a second M1 anti-FLAG antibody affinity column (11, 41). The receptor was further purified by size exclusion chromatography using a Superdex 200 (16/600 prep grade) column. The monomeric receptor peak was pooled and concentrated to 1–2 mg/mL, flash frozen supplemented with 20% (vol/vol) glycerol, 0.01% cholesteryl hemisuccinate, and protease inhibitors, and stored at −80 °C.

Cell Culture and Transfection.

HEK-293 and U2OS cells were cultured at 37 °C in a humidified environment (5% CO₂) using standard minimum Eagle’s growth media supplemented with 10% (vol/vol) FBS and penicillin/streptomycin. HEK-293 cell lines stably expressing the Glosensor (Promega) cAMP reporter (42), the ICUE2 cAMP reporter together with the β₂AR (32), or β-arrestin2-mYFP together with β₂AR-mCFP (43) were maintained as described before. U2OS cell lines stably expressing enzyme acceptor-tagged β-arrestin2 alone or together with an endosome-localized ProLink tag were provided by DiscoverX and maintained as described (24). U2OS cells stably expressing the ProLink-tagged β₂V₂R were created by stably transfecting the neomycin (G418)-resistant plasmid expressing ProLink-tagged β₂V₂R (DiscoverX), and subsequent selection with 300 μg/mL of G418 (Sigma-Aldrich). The clonal line with the highest sensitivity as well as the best fold over basal ratio was selected and maintained with 300 μg/mL G418 and 300 μg/mL Hygromycin B (Invitrogen). Transient transfection was done using FuGENE 6 (Promega) according to the manufacturer’s instructions with a 5:1 ratio of the reagent over the amount of plasmid DNA. All of the assays were performed at ∼48 h after transfection.

DNA-Encoded Library Synthesis.

Libraries were built by a tagged-split-and-pool chemistry approach (Chemetics; Nuevolution) at Lexicon Pharmaceuticals.

Affinity Selection.

The library selection procedure is schematically illustrated in Fig. S1. In detail, 25 μg of detergent (DDM)-solubilized, Flag-tagged β₂AR was immobilized on 50 μL of M1 anti-Flag beads. The β₂AR beads were then incubated with library molecules in 200 μL of a binding buffer (20 mM Hepes, pH 7.4, 150 mM NaCl, 2 mM CaCl₂, 0.1% DDM), supplemented with 20 μg of sheared salmon sperm DNA (ssDNA) (Ambion), for 1 h at room temperature, while vigorously agitated. Before incubation, 1 μL of library molecules were set aside for quantitation using quantitative PCR (Q-PCR). Following incubation, the beads were repeatedly washed (five times) with 1 mL of the ice-cold binding buffer supplemented with ssDNA using centrifugation at 10,000 × g for 1 min at 4 °C to remove unbound library molecules. Bound compounds were then eluted by twice incubating the beads with 250 μL of the preheated binding buffer containing 12.5 μg of ssDNA at 72 °C for 30 min followed by centrifugation at 10,000 × g for 1 min. The combined supernatant was first purified using RapidClean (Advansta) according to the manufacturer’s instructions to remove denatured protein, and then desalted using an Amicon Ultra-3K centrifugal filter (EMD Millipore), exchanging to 200 μL of the fresh binding buffer. One microliter of purified material was again set aside for quantitation by Q-PCR, and the entire remaining volume of the sample was either subjected to additional rounds of affinity selection with fresh protein, or was subjected to DNA sequencing.

Q-PCR.

Library DNA was quantitated at each round of the affinity selection process and before sequencing using Q-PCR. Library DNA copy number was determined by absolute quantitation using a standard curve generated with plasmid DNA prepared from a representative library fragment that had been subcloned and sequence verified. In brief, library DNA samples were amplified directly or following dilution in TE buffer supplemented with ssDNA using oligonucleotide primers 5′-AAGGCCTAGATTCACTCACG-3′ and 5′-TAGTAGAGTCAGCAGTGAGC-3′, custom FAM/TAMRA probe 5′-GCGACCGTTGACGATGCCGAG-3′, and TaqMan Universal PCR Mix (Thermo Fisher Scientific) according to manufacturer’s guidelines. Samples were processed in triplicate, read using an ABI 7900HT, and assessed using at least two concentrations.

Next-Generation Sequencing.

Sequencing was carried out using the Ion Torrent PGM (Thermo Fisher Scientific) using manufacturer’s procedures for single-direction amplicon sequencing. In brief, affinity-selected material was amplified by two rounds of PCR to first increase yield and then to append the necessary sequencing adapters for emulsion PCR. The first round of amplification was performed using library forward primer 5′-AAGGCCTAGATTCACTCACG-3′ and a unique reverse primer that was appended to individual library aliquots as they were designated for use in affinity selection. This unique primer ensures precise sample tracking and concomitantly provides a distinct identifier for each library. The second round of PCR used oligonucleotide primers that fused the Ion Torrent adapter sequences to the library forward primer and the unique library sample reverse primer, where a 5-nt sorting code was inserted between the sequencing adapters and the library template to allow sample pooling. PCR products were gel-purified and quantitated by Q-PCR as described above, except that the standard curve was derived from plasmid DNA representing a sequence-verified Ion Torrent library template. A total of 4 × 10⁸ DNA molecules were used for each emulsion PCR. Emulsion PCR samples were processed on the Ion OneTouch 2 system (Thermo Fisher Scientific) and were pipetted manually onto an Ion 318 chip for sequence detection. FASTA sequence output was analyzed by a custom BLAST algorithm to match each sequence coding element to the synthetic design template for the specified library, by which these sequences were subsequently deconvoluted into their requisite chemical structures. The frequency of compound detection postselection versus a control selection performed without protein was used as a measure of significance.

Reconstitution of the β₂AR into High-Density Lipoprotein Particles.

FLAG-tagged β₂AR was incorporated into high-density lipoprotein (HDL) particles (nanodiscs) using membrane scaffolding protein 1 (MSP1), a derivative of apolipoprotein A-1, as previously described (21, 22). Briefly, FLAG-tagged β₂AR was incubated with a 50-fold molar excess of MSP1, 8 mM POPC:POPG (3:2 molar ratio; 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine and 1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-[1′-rac-glycerol]) lipids (Avanti Polar Lipids), respectively, for 1 h at 4 °C. After removal of detergent using BioBeads SM-2 (Bio-Rad) overnight at 4 °C, receptor-containing nanodiscs were purified, from free nanodiscs, by M1 anti-FLAG antibody affinity column followed by size exclusion chromatography.

Radioligand Binding.

Assays were done essentially as described (10, 22). Briefly, competition binding was performed with ∼0.7 ng of the β₂AR in nanodiscs diluted in an assay buffer (20 mM Hepes, pH 7.4, 100 mM NaCl, 0.1% BSA, and 1 mM ascorbic acid). The reaction mixture (250 μL) contained the β₂AR, 60 pM [¹²⁵I]-cyanopindolol (CYP) (2,200 Ci/mmol; PerkinElmer), 15 at different concentrations, and a serial dilution of isoproterenol. Nonspecific binding was determined using 20 μM propranolol. Following a 90-min incubation at room temperature, binding assays were terminated by rapid filtration onto GF/B glass-fiber filters (Brandel) treated with 0.3% polyethyleneimine (PEI) and washed with 8 mL of a cold binding buffer (20 mM Hepes, pH 7.4, 100 mM NaCl) using a harvester (Brandel). Bound [¹²⁵I] was quantified using a Packard Cobra Quantum gamma counter (Packard) and expressed as specific binding.

For [³H]-ICI 118,551 (American Radiolabeled Chemicals) binding, 10 ng of β₂AR in nanodiscs were incubated with the radioligand at 0.3 nM and 15 in a serial dilution, and were harvested as described above. Bound [³H] was quantified using a Tri-Carb 2800TR liquid scintillation counter (PerkinElmer).

[³H]-methoxyfenoterol (³H-Fen) (25) binding was carried out with isolated membranes from Sf9 cells exogenously expressing the β₂V₂R (10). Membranes with phosphorylated forms of the receptor were obtained from Sf9 cells coexpressing the membrane targeted from of GRK2 (24), followed by stimulation with isoproterenol (10 µM) for 20 min before harvesting cells. Membranes were prepared essentially following the procedures described earlier (24). Approximately 40 μg of the isolated membranes in a binding buffer (50 mM Tris, pH 7.4, 2 mM EDTA, 12.5 mM MgCl₂) supplemented with 0.05% BSA and 1 mM l-ascorbic acid, were incubated with 15 at different concentrations (or its analogs), 4.3 nM ³H-Fen, and transducers (100 nM heterotrimeric Gs or 250 nM β-arrestin1 together with 0.5 µM Fab30). Following incubation for 90 min at room temperature, binding assays were terminated by harvesting the reaction mixture onto PEI-soaked GF/B filters. Bound [³H] was extracted overnight with 5 mL of scintillation fluid and quantified as described above.

Isothermal Titration Calorimetry.

Isothermal titration calorimetry (ITC) analysis was performed as described (22) with detergent (DDM)-solubilized, purified β₂AR. ITC experiments were carried out using the MicroCaliTC200 system (Malvern). The β₂AR was dialyzed against a buffer consisting of 20 mM Hepes, pH 7.5, 100 mM NaCl, 0.1% DDM, and 0.001% cholesteryl hemisuccinate. Protein concentrations were determined by measurement of the absorbance at 280 nm, using molar extinction coefficient per cm parameters of the β₂AR (ε: 66,350). Compound 15 at a 200 μM concentration in the same buffer (40 μL) was loaded into the syringe and titrated into the 200-μL sample cell containing β₂AR (20 μM). Titrations were performed at 25 °C using an initial injection of 0.5 μL, followed by 2.3-μL injections (1-s duration, 300-s spacing, and 5-s filter period). Reference power was set to 10 μcal⋅s⁻¹ and stirring speed to 750 rpm (MicroCaliT200). Reference titrations were obtained by injecting 15 alone into sample cells containing the buffer alone. To obtain the stoichiometry (N) and association constant (K_A) of the interaction, the raw data were baseline corrected, peak area integrated, and fitted to a one-site nonlinear least-squares fit model using the Origin7 software program.

Nanobody ELISA.

Preparation of nanobodies (Nbs) and their binding to the β₂AR, determined by ELISA, were done as essentially described (10). To passively adsorb Nbs, Maxisorp (NUNC) 96-well plates (Thermo Fisher Scientific) were coated with 100 µL of 10 µg/mL purified Nb in buffer A (20 mM Hepes, pH 7.4, 100 mM NaCl) overnight at 4 °C. Each new reagent addition was preceded by three 5-min washes with 300 µL of a wash buffer (buffer A supplemented with 0.02% DDM). The plate was blocked for 60 min at room temperature in the wash buffer with 3% (wt/vol) nonfat milk. Purified full-length human β₂AR was preincubated with dimethyl sulfoxide (DMSO) or ligands for 30 min in buffer A supplemented with 0.1% DDM, and 0.5% BSA. Immobilized Nbs were incubated with 100 µL of 0.5 µg/mL purified β₂AR for 90 min at room temperature. Captured β₂AR was detected using M2-HRP (1:5,000) diluted in buffer A supplemented with 0.02% DDM, and 0.5% BSA. Antibody was incubated for 1 h at room temperature, plates were subsequently treated with 100 µL of Ultra-TMB (Pierce), and absorbance was measured at 450 nm.

Measurements of cAMP Production.

The level of cAMP was monitored primarily using Glosensor, a chemiluminescence-based cAMP biosensor (28) (Promega) as previously described (29) with slight modifications. HEK-293 cells stably expressing Glosensor were plated in 96-well white, clear-bottomed plates at a density of 80,000 cells per well. On the following day, Glosensor reagents (Promega) were added, and cells were incubated in a 27 °C humidifying incubator for ∼1 h. Subsequently, cells were treated with either 0.5% DMSO or 15 at different concentrations in Hanks’ balanced solution (Sigma), supplemented with 20 mM Hepes, pH 7.4, and 0.05% BSA, together with 100 μM 3-isobutyl-1-methylxanthine (IBMX) (Sigma), and further incubated for 20 min. At the end of incubation, cells were stimulated with a serial dilution of Gs-coupled receptor agonists including isoproterenol for 5 min at room temperature. Changes in luminescence were read using a NOVOstar microplate reader (BMG Labtech).

The fluorescence resonance energy transfer (FRET)-based ICUE2 reporter assay (32) was also used to monitor cAMP levels. In brief, HEK-293 cells stably expressing ICUE2 together with the β₂AR were plated in 96-well black, clear-bottomed plates at a density of 50,000 cells per well. On the following day, the growth medium was substituted with an imaging buffer (125 mM NaCl, 5 mM KCl, 1.5 mM MgCl₂, 1.5 mM CaCl₂, 10 mM glucose, 0.2% BSA,10 mM Hepes, pH 7.4) at 37 °C. After the first preread of FRET signals using a NOVOstar microplate reader (BMG Labtech), cells were treated with either 0.5% DMSO or 15 at different concentrations in the imaging buffer for 20 min at 37 °C. After the second preread of FRET signals, cells were stimulated with a serial dilution of isoproterenol for 3 min at 37 °C, and FRET signals were read again. Intracellular cAMP levels were determined as loss of ICUE2 FRET signals upon stimulation, expressed as a FRET ratio as follows: cyan fluorescent protein (CFP) intensity (438/32 emission bandpass filters; Semrock) relative to FRET intensity (542/27 emission filter).

In Vitro GTPase Activity Assay.

β₂AR-induced activation of Gs protein was monitored by measuring GTP hydrolysis using chemiluminescence-based GTPase-Glo Assay (Promega), following manufacturer’s protocol. The β₂AR reconstituted into HDL particles at 4 nM was incubated with 2.5 μM GTP and 15 at various concentrations for 20 min at room temperature in 96-well flat white-bottomed plates, followed by stimulation with isoproterenol in a concentration-dependent fashion. The GTP hydrolysis reaction was initiated by addition of 443 nM purified Gs protein and 1 mM DTT to the preincubated receptor mixture together with ligands at total volume of 25 μL per well. The enzymatic reaction of Gs protein was proceeded at room temperature for 2 h. The reaction was quenched by addition of the reconstituted GTPase-Glo reagent (25 μL per well) and incubated at room temperature for 30 min with shaking. Then detection reagent (50 μL per well) was applied and incubated at room temperature for 10 min in dark. The luminescence signal was detected by a SpectraMax M5 plate reader (Molecular Devices).

Measurements of β-Arrestin Recruitment.

The extent of β-arrestin recruitment was monitored primarily using PathHunter, a chemiluminescence-based enzyme fragment complementation assay (30) (DiscoverX), as previously described (29) with slight modifications. U2OS cells, stably expressing enzyme acceptor-tagged β-arrestin2 together with ProLink-tagged receptors (either stably expressed β₂V₂R or transiently expressed other receptors; DiscoverX), were plated in 96-well white, clear-bottomed plates at a density of 25,000 cells per well. On the following day, cells were pretreated with 0.5% DMSO or 15 as described for the Glosensor assay. Subsequently, cells were stimulated with agonists for 30–45 min at 37 °C, which was terminated by adding PathHunter detection reagents (DiscoverX). After further incubation for 1 h at 27 °C, luminescence signals were read using a NOVOstar microplate reader (BMG Labtech).

Changes in β-arrestin recruitment were also determined by monitoring FRET signals between β-arrestin2-mYFP and either β₂AR-mCFP or AT₁R-mCFP as previously described (43) with slight modifications. Briefly, HEK-293 cells stably expressing β-arrestin2-mYFP and either β₂AR-mCFP or AT₁R-mCFP were plated and treated with DMSO or 15, and FRET signals were read as described for the ICUE2 cAMP assay after stimulation with a serial dilution of either isoproterenol or angiotensin II, respectively, for 7 min at 37 °C. The extent of β-arrestin recruitment to the receptor was measured as a fluorescence resonance FRET ratio as follows: FRET intensity (542/27 emission filter) relative to CFP intensity (438/32 emission bandpass filters; Semrock).

β-Arrestin Endocytosis Assay.

The level of β-arrestin endocytosis was measured by the active endocytosis assay (DiscoverX) as described (24) with slight modifications. Individual receptors examined were transiently transfected into U2OS cells stably expressing an Enzyme Acceptor-tagged β-arrestin2 and endosome-localized ProLink tag protein. On the following day, transfected cells were plated, treated with DMSO or 15 as described for the recruitment assay, and stimulated with a serial dilution of respective agonists for 60 min. β-Arrestin endocytosis was detected as chemiluminescence signals resulting from the complementation of β-galactosidase fragments (Enzyme Acceptor and ProLink) within endosomes. Luminescence signals were detected on a NOVOstar plate reader (BMG Labtech) using the PathHunter Detection kit (DiscoverX).

Data Analyses.

All of the dose–response curve fits, except from the ITC experiment, were obtained using the computer program GraphPad Prism. Statistical analyses for the shift of curves in Figs. 2A and 3, and Figs. S2 and S3 were performed using a two-way ANOVA. *P < 0.05; **P < 0.01; ***P < 0.001 compared with the control curve obtained in the presence of the vehicle (DMSO). Statistical analyses for the results in Fig. 2 E and F and Table S4 as well as percentage decreases in the maximal response and fold shifts of the EC₅₀ value shown in Tables S2 and S3 were performed using a one-way ANOVA with Bonferroni posttests. *P < 0.05; **P < 0.01; ***P < 0.001, compared with the control value obtained in the presence of the vehicle (DMSO) except analyses done in Table S4, where the data were compared with the value obtained with 15 that was normalized to 100%.

Synthesis and characterization of compound 15 and its derivatives.

General chemistry.

All chemicals and solvents, unless otherwise stated, were purchased from standard suppliers (Sigma-Aldrich, Thermo Fisher Scientific, Santa Cruz Biotechnology, TCI America) and were used without further purification. Precoated silica gel 60 F254 aluminum plates were used for analytical TLC. Course of reactions was followed by visualization under UV (254 or 366 nm) and/or using standard staining procedures such as KMnO₄. Flash column chromatography was performed on using silica gel 60 (SiO₂; 230–400 mesh; Merck). Structural characterization of compounds was performed by NMR spectroscopy (¹H and ¹³C) and mass spectrometry (MS). ¹H NMR and ¹³C NMR spectra were recorded on a FT-NMR Bruker Avance Ultra Shield Spectrometer at 400.13 (or 300) and 100.62 (or 75) MHz, respectively. Deuterated solvents (DMSO-d₆ and CDCl₃) were purchased from Cambridge Isotope Laboratories. Chemicals shifts (δ) are reported in parts per million (ppm) downfield from tetramethylsilane (TMS). Coupling constants (J) values are in hertz, and the splitting patterns are described as follows: singlet (s); doublet (d); triplet (t); quartet (q); multiplet (m). High-resolution time-of-flight mass spectra (HRMS ESI-TOF) were obtained on a Waters LCT Premier XE (TOF) using electrospray ionization.

(9H-Fluoren-9-yl)methyl (S)-(3-(3-bromophenyl)-1-(methylamino)-1-oxopropan-2-yl)carbamate (2).

To an ice-cold stirred solution of Fmoc-3-bromo-l-phenylalanine (466 mg, 1 mmol), 1-hydroxybenzotriazole (HOBt) (270 mg, 2 mmol) and O-benzotriazol-1-yl-N,N,N′,N′-tetramethyluronium hexafluorophosphate (HBTU) (758 mg, 2 mmol) in DMF (6 mL), methylamine hydrochloride (135 mg, 2 mmol) was added followed by N,N-diisopropylethylamine (388 mg, 3 mmol) under nitrogen stream. The reaction mixture was stirred at 0 °C for 30 min, and then allowed to warm to ambient temperature while the stirring was continued for 12 h. After removal of DMF under reduced pressure, the residue was dissolved in EtOAc (120 mL), and then washed with 0.5 M NaHSO₄ solution (30 mL), saturated NaHCO₃ solution (30 mL), and brine (2 × 30 mL), respectively, and finally dried over Na₂SO₄. The solvent was removed with a rotary evaporator, and the tan solid residue was crystallized from EtOAc to give a white fluffy solid (450 mg, 94% yield). ¹H NMR (300 MHz, DMSO-d₆): δ2.60 (d, J = 4.5 Hz, 3H), 2.72–3.02 (dd, J = 14.0, 4.4 Hz, 2H), 4.20 (m, 4H), 7.19–8.01 (m, 13H); ¹³C NMR (75 MHz, DMSO-d₆): δ25.5, 25.7, 37.4, 55.8, 56.1, 110.0, 120.2, 121.5, 127.4, 128.5, 128.6, 129.1, 129.2, 130.3, 130.4, 132.1, 132.2, 137.6, 139.5, 141.4, 141.6, 142.7, 157.0, 172.2. ESI-MS (positive mode): m/z 479 [M + H]⁺ and m/z 501 [M + Na]⁺.

(S)-2-Amino-3-(3-bromophenyl)-N-methylpropanamide (3).

To a stirred solution of (9H-fluoren-9-yl)methyl (S)-(3-(3-bromophenyl)-1-(methylamino)-1-oxopropan-2-yl)carbamate (450 mg, 0.94 mmol) in DMF (4 mL) was added piperidine (2 mL) at room temperature. The reaction mixture was stirred at ambient temperature and under nitrogen atmosphere for 2 h or until the starting material completely disappeared, as checked by TLC. After removal of solvent by a rotary evaporator, the residue was purified by flash column chromatography (eluting with DCM/MeOH = 20:1) to afford light yellow solid (180 mg, 75% yield). ¹H NMR (300 MHz, DMSO-d₆): δ2.56 (d, J = 4.7 Hz, 3H), 2.61 (dd, J =13.4, 8.3 Hz, 1H), 2.89 (dd, J = 13.3, 5.0 Hz, 1H), 3.35 (dd, J = 8.2, 5.0 Hz, 1H), 7.17–7.82 (m, 5H); ¹³C NMR (75 MHz, DMSO-d₆): δ25.5, 40.8, 56.3, 121.5, 128.5, 129.0, 130.2, 132.1, 142.1, 174.7. ESI-MS (positive mode): m/z 257 [M + H]⁺ and m/z 289 [M + Na]⁺.

(9H-Fluoren-9-yl)methyl ((S)-1-(((S)-3-(3-bromophenyl)-1-(methylamino)-1-oxopropan-2-yl)amino)-3-(4-carbamoylphenyl)-1-oxopropan-2-yl)carbamate (4).

To a stirred solution of Fmoc-l-4-carbamoylphenylalanine (301 mg, 0.7 mmol) in DMF (4 mL) was added HOBt (113 mg, 0.84 mmol) and HBTU (318 mg, 0.84 mmol) at room temperature, and the mixture was cooled to 0 °C before a solution of (S)-2-amino-3-(3-bromophenyl)-N-methylpropanamide (180 mg, 0.7 mmol) in DMF (2 mL) was added. The reaction mixture was stirred at 0 °C for 10 min, and then N,N-diisopropylethylamine (272 mg, 2.1 mmol) was introduced via a syringe. After the whole reaction mixture was stirred at ambient temperature for 12 h, the solvent, DMF, was removed under reduced pressure. A tan solid residue was crystallized from dichloromethane to give white solid (365 mg, 78% yield). ¹H NMR (300 MHz, DMSO-d₆): δ2.57 (d, J = 4.4 Hz, 3H), 2.72–3.05 (m, 4H), 4.11–4.50 (m, 5H), 7.18–8.23 (m, 18H); ¹³C NMR (75 MHz, DMSO-d₆): δ25.6, 46.6, 53.9, 54.2, 56.0, 65.8, 120.2, 121.5, 125.4, 126.4, 127.2, 127.4, 127.7, 128.2, 128.4, 129.2, 129.3, 130.3, 132.0, 132.4, 137.7, 140.7, 140.8, 141.6, 141.7, 143.8, 143.9, 155.8, 167.9, 170.9, 171.2, 171.3. ESI-MS (positive mode): m/z 691 [M + Na]⁺.

4-((S)-2-Amino-3-(((S)-3-(3-bromophenyl)-1-(methylamino)-1-oxopropan-2-yl)amino)-3-oxopropyl)benzamide (5).

To a stirred solution of (9H-fluoren-9-yl)methyl ((S)-1-(((S)-3-(3-bromophenyl)-1-(methylamino)-1-oxopropan-2-yl)amino)-3-(4-carbamoylphenyl)-1-oxopropan-2-yl)carbamate (365 mg, 0.52 mmol) in DMF (4 mL) was added piperidine (2 mL) at room temperature. The reaction mixture was stirred at ambient temperature and under nitrogen atmosphere for 2 h or until the starting material completely disappeared, as checked by TLC. After removal of the solvent by a rotary evaporator, the tan solid residue was crystallized from EtOAc to afford light brown solid (236 mg, 95% yield). ¹H NMR (300 MHz, DMSO-d₆): δ2.57 (d, J = 4.5 Hz, 3H), 2.86–3.35 (m, 5H), 4.47 (m, 1H), 7.14–8.14 (m, 10H); ¹³C NMR (75 MHz, DMSO-d₆): δ25.7, 37.7, 40.6, 53.3, 56.2, 121.5, 127.5, 128.5, 129.2, 129.3, 130.3, 132.1, 132.3, 140.7, 142.3, 170.9, 171.0, 174.0. ESI-MS (positive mode): m/z 447 [M + H]⁺ and m/z 469 [M + Na]⁺.

4-((2S)-3-(((S)-3-(3-Bromophenyl)-1-(methylamino)-1-oxopropan-2-yl)amino)-2-(2-cyclohexyl-2-phenylacetamido)-3-oxopropyl)benzamide (compound 15).

To a stirred solution of 2-cyclohexyl-2-phenylacetic acid (218 mg, 1.0 mmol) in DMF (4 mL) was added HOBt (135 mg, 1.0 mmol) and HBTU (379 mg, 1.0 mmol) at room temperature. The mixture was cooled to 0 °C, and a solution of 4-((S)-2-amino-3-(((S)-3-(3-bromophenyl)-1-(methylamino)-1-oxopropan-2-yl)amino)-3-oxopropyl)benzamide (230 mg, 0.5 mmol) in DMF (2 mL) was slowly added. Then N,N-diisopropylethylamine (194 mg, 1.5 mmol) was added dropwise to this solution as it stirred at 0 °C. After the reaction mixture was stirred at ambient temperature for 10 h, the solvent and volatiles were evaporated under the reduced pressure, and a tan solid residue was crystallized from EtOAc to finally generate a white solid (240 mg, 72% yield). ¹H NMR (400 MHz, DMSO-d₆): δ0.86–1.54 (m, 11H), 2.55 (d, J = 4.5 Hz, 3H), 2.78–3.17 (m, 5H), 4.43 (m, 2H), 6.95–8.15 (m, 13H); ¹³C NMR (100 MHz, DMSO-d₆): δ25.6, 26.2, 30.3, 31.3, 37.6, 53.4, 53.5, 53.8, 53.9, 57.9, 121.5, 126.4, 127.2, 128.1, 128.3, 128.4, 128.5, 129.0, 129.4, 130.3, 132.0, 132.2, 139.4, 140.8, 167.8, 170.7, 170.8, 172.5. HRMS (ESI-TOF): calcd. for C₃₄H₃₉BrN₄O₄, 647.2155 [M+H]⁺; found: 647.2226.

Compound 15 derivatives.

(S)-3-(3-Bromophenyl)-2-((2S)-2-(2-cyclohexyl-2-phenylacetamido)-3-phenylpropanamido)-N-methylpropanamide (compound 15A1).

The title compound, a derivative of compound 15 lacking the formamide group on the aromatic core of the methylbenzamide, was prepared in a manner analogous to parent compound, except in the third step shown in Scheme S1, the intermediate Fmoc-l-phenylalanine (79 mg, 0.2 mmol) was used instead of Fmoc-l-4-carbamoylphenylalanine. The compound was purified as white solid (70% yield). ¹H NMR (300 MHz, DMSO-d₆): δ0.61–1.89 (m, 11H), 2.55 (d, J = 4.5 Hz, 3H), 2.64–3.19 (m, 5H), 4.30–4.45 (m, 2H), 6.90–8.17 (m, 20H). ESI-MS (positive mode): m/z 604 [M + H]⁺.

Scheme S1.

Synthesis of compound 15.

3-((S)-2-Amino-3-(((S)-3-(3-bromophenyl)-1-(methylamino)-1-oxopropan-2-yl)amino)-3-oxopropyl)benzamide (compound 15A2).

The title compound, bearing the formamide group in the meta position rather than in para on the aromatic core of the methylbenzamide, was prepared in a manner analogous to compound 15. Except in this case in the third step in Scheme S1, the intermediate Fmoc-l-3-carbamoylphenylalanine (175 mg, 0.40 mmol) was used instead of Fmoc-l-4-carbamoylphenylalanine. The compound was purified as light brown solid (81% yield). ¹H NMR (300 MHz, DMSO-d₆): δ0.45–1.99 (m, 11H), 2.56 (d, J = 4.4 Hz, 3H), 2.68–3.19 (m, 5H), 4.43 (m, 2H), 6.95–8.15 (m, 17H); ¹³C NMR (75 MHz, DMSO-d₆): δ25.5, 25.8, 37.6, 53.4, 53.7, 54.0, 58.0, 121.4, 124.5, 126.9, 127.4, 127.6, 127.8, 128.1, 128.3, 128.5, 128.8, 129.1, 129.6, 132.1, 139.4, 140.4, 140.6, 140.9, 167.7, 167.8, 170.7, 170.8, 171.0, 172.3, 172.5. HRMS (ESI-TOF): calcd. for C₃₄H₃₉BrN₄O₄, 647.2155 [M+H]⁺; found: 647.2229.

4-((2S)-2-(2-Cyclohexyl-2-phenylacetamido)-3-(((S)-1-(methylamino)-1-oxo-3-phenylpropan-2-yl)amino)-3-oxopropyl)benzamide (compound 15A3).

The title compound, lacking the bromine group on the aromatic core bromo-benzyl was prepared in a manner analogous to compound 15, except in the first step in Scheme S1, the intermediate Fmoc-3-l-phenylalanine (1 g, 2.6 mmol) was used instead of Fmoc-3-bromo-l-phenylalanine. The compound was purified as white solid (81% yield); ¹H NMR (300 MHz, DMSO-d₆): δ0.86–1.54 (m, 11H), 2.53 (d, J = 4.6 Hz, 3H), 2.68–3.24 (m, 5H), 4.43 (m, 2H), 6.95–8.42 (m, 15H). ESI-MS (positive mode): m/z 569 [M + H]⁺.

4-((2S)-3-(((S)-3-(3-Fluorophenyl)-1-(methylamino)-1-oxopropan-2-yl)amino)-2-(2-cyclohexyl-2-phenylacetamido)-3-oxopropyl)benzamide (compound 15A4).

The title compound, bearing a 2-fluoro group instead of 2-bromo on the aromatic core bromobenzyl of compound 15, was prepared in a manner analogous to it, except in the first step in Scheme S1, the intermediate Fmoc-3-fluoro-l-phenylalanine (650 mg, 1.55 mmol) was used instead of Fmoc-3-bromo-l-phenylalanine. The compound was purified as white solid (41% yield). ¹H NMR (300 MHz, DMSO-d₆): δ1.09–1.98 (m, 11H), 2.54 (d, J = 4.3 Hz, 3H), 2.67–3.20 (m, 5H), 4.42 (m, 2H), 6.95–8.20 (m, 14H). HRMS (ESI-TOF): calcd. for C₃₄H₃₉FN₄O₄ [M+Na]⁺ 609.2955; found: 609.2851.

4-((2S)-3-(((S)-3-(3,5-Dibromophenyl)-1-(methylamino)-1-oxopropan-2-yl)amino-2-(2-cyclohexyl-2-phenylacetamido)-3-oxopropyl)benzamide (compound 15A5).

The title compound, bearing di-bromine groups instead of a single bromine group on the aromatic core bromo-benzyl scaffold of compound 15, was prepared in a manner analogous to it, except in the first step in Scheme S1, the intermediate (9H-fluoren-9-yl)methyl (S)-(3,5-dibromo)phenylalanine (351 mg, 0.64 mmol) was used instead of Fmoc-3-bromo-l-phenylalanine. The compound was purified as white solid (50% yield). ¹H NMR (400 MHz, DMSO-d₆): δ0.92–1.89 (m, 11H), 2.55 (d, J = 4.7 Hz, 3H), 2.78–3.19 (m, 5H), 4.37 (m, 2H), 6.94–8.51 (m, 12H); ¹³C NMR (100 MHz, DMSO-d₆): δ25.7, 30.3, 31.3, 37.2, 37.4, 53.8, 53.9, 53.8, 53.9, 58.0, 122.2, 126.5, 127.2, 127.4, 128.1, 128.3, 128.5, 128.9, 129.3, 131.5, 132.0, 132.1, 139.5, 139.6, 141.0, 141.5, 142.7, 142.9, 167.8, 167.9, 170.7, 170.9, 171.2, 172.2, 172.5. HRMS (ESI-TOF): calcd. for C₃₄H₃₈Br₂N₄O₄ [M+Na]⁺, 749.1248; found: 749.1139.

4-((2S)-3-(((S)-3-(3-Bromophenyl)-1-(methylamino)-1-oxopropan-2-yl)amino)-2-(2-cyclohexyl-2-(4-hydroxyphenyl)acetamido)-3-oxopropyl)benzamide (compound 15A6).

The title compound, bearing hydroxyl group on the aromatic core of the cyclohexylmethyl-benzene region of compound 15, was prepared in a manner analogous to it, except in the fifth step in Scheme S1, 2-cyclohexyl-2-(4-hydroxyphenyl)acetic acid (51 mg, 0.22 mmol) was used instead of 2-cyclohexyl-2-phenylacetic acid. The compound was purified as white solid (74% yield). ¹H NMR (400 MHz, DMSO-d₆): δ0.83–1.58 (m, 11H), 2.55 (d, J = 4.4 Hz, 3H), 2.68–3.07 (m, 5H), 4.40 (m, 2H), 6.99–9.46 (m, 14H). HRMS (ESI-TOF): calcd. for C₃₄H₃₉BrN₄O₅ [M+Na]⁺ 685.2104; found: 685.1992.

4-((2S)-3-(((S)-3-(3-Bromophenyl)-1-(methylamino)-1-oxopropan-2-yl)amino)-2-(2-cyclohexyl-2-(4-methoxyphenyl)acetamido)-3-oxopropyl)benzamide (compound 15A7).

The title compound, bearing methoxy group on the aromatic core of the cyclohexylmethyl-benzene portion of compound 15, was prepared in a manner analogous to it, except in the fifth step in Scheme S1, 2-cyclohexyl-2-(4-methoxyphenyl)acetic acid (54 mg, 0.22 mmol) was used instead of 2-cyclohexyl-2-phenylacetic acid. The compound was purified as white solid (70% yield); ¹H NMR (400 MHz, DMSO-d₆): δ0.85–1.6 (m, 11H), 2.55 (d, J = 4.5 Hz, 3H), 2.67–3.11 (m, 5H), 4.39 (s, 3H), 4.40–4.43 (m, 2H), 6.99–8.15 (m, 15H). HRMS (ESI-TOF): calcd. for C₃₅H₄₁BrN₄O₅ [M+Na]⁺ 699.2260; found: 699.2160.