Synthesis and Evaluation of the First Fluorescent Antagonists of the Human P2Y2 Receptor Based on AR-C118925

The human P2Y2 receptor (hP2Y2R) is a G-protein-coupled receptor that shows promise as a therapeutic target for many important conditions, including for antimetastatic cancer and more recently for idiopathic pulmonary fibrosis. As such, there is a need for new hP2Y2R antagonists and molecular probes to study this receptor. Herein, we report the development of a new series of non-nucleotide hP2Y2R antagonists, based on the known, non-nucleotide hP2Y2R antagonist AR-C118925 (1), leading to the discovery of a series of fluorescent ligands containing different linkers and fluorophores. One of these conjugates, 98, displayed micromolar affinity for hP2Y2R (pKd = 6.32 ± 0.10, n = 17) in a bioluminescence-energy-transfer (BRET) assay. Confocal microscopy with this ligand revealed displaceable membrane labeling of astrocytoma cells expressing untagged hP2Y2R. These properties make 98 one of the first tools for studying hP2Y2R distribution and organization.


■ INTRODUCTION
P2Y receptors (P2YRs) are G-protein-coupled receptors (GPCRs) that are activated by extracellular nucleotides. The P2Y family is composed of eight members, encoded by distinct genes, that can be subdivided into two groups on the basis of their primary signaling through specific G proteins 1 and sequence homology. The first subgroup includes the P2Y 1,2,4,6,11 receptors, which primarily signal though G q , whereas the second subgroup, signaling through G i , encompasses the P2Y 12,13,14 receptor subtypes. 2 Notably, the P2Y 2 receptor (P2Y 2 R) is activated by the endogenous agonists uridine-5′-triphosphate (UTP, hP2Y 2 EC 50 = 140 nM) and adenosine-5′-triphosphate (ATP, hP2Y 2 EC 50 = 230 nM). 3 As P2Y 2 R is predominately G qcoupled, receptor activation leads to the stimulation of phospholipase C, IP 3 release, and the elevation of the intracellular Ca 2+ concentration, as well as the initiation of protein kinase C and the activation of the mitogen-activated proteinkinase cascade.
Defining the clinical role for P2Y 2 R antagonism has been hampered by the lack of high-affinity and druglike receptor antagonists. 4 However, it has been reported that ATP released from tumor-cell-activated platelets induces the opening of the endothelial barrier, leading to the migration of tumor cells and hence cancer proliferation. More importantly, P2Y 2 R was identified as the primary mediator of this effect; a strong reduction of tumor cell metastasis was observed in P2Y 2 R-deficient mice, revealing a therapeutic potential of P2Y 2 R antagonists as antimetastatic agents. 5,6 Recently, it has been reported that both inflammation and fibrosis were reduced in P2Y 2 R-deficient mice compared with those in wild-type animals. In addition, mechanistic studies have demonstrated that the recruitment of neutrophils into the lungs, the proliferation and migration of lung fibroblasts, and IL-6 production are all key P2Y 2 R-mediated processes. These studies clearly demonstrate the involvement of P2Y 2 R subtypes in the pathogenesis of fibrotic lung diseases in humans and mice and support the development of selective P2Y 2 R antagonists for the treatment of idiopathic pulmonary fibrosis (IPF). 7 To date, the only reported high-affinity P2Y 2 R antagonists were those developed by scientists from AstraZeneca resulting in the non-nucleotide P2Y 2 R antagonist AR-C118925 (1). 8,9 Several in vivo and ex vivo studies using 1 have been reported that further validate the therapeutic benefits of P2Y 2 R antagonists. Importantly, it was shown that 1, which was reported to be inactive at 10 μM against a panel of 37 other receptors, was able to concentration-dependently antagonize ATPγS-induced mucin secretion in an ex vivo model of human bronchial epithelial cells. 10 In addition, Muller et al. recently demonstrated that 1 was a selective, high-affinity, reversible antagonist of P2Y 2 R. 11 We were drawn to the exciting possibility of using 1 as a chemical template to design new P2Y 2 R antagonists 12 and synthesize fluorescently labeled chemical tools to further probe P2Y 2 R function. 13 Using fluorescence as a means to study GPCRs allows scientists access to a large range of pharmacological techniques that can capture dynamic processes in living cells. 14 In particular, fluorescently labeled receptor antagonists have been developed to target GPCRs, allowing the visualization of GPCR function at the cellular level. 15 −17 In addition, fluorescent ligands can be used in resonance-energy-transfer (RET) techniques, in particular those that utilize nanoluciferase (NLuc), to quantify ligand−receptor interactions and determine the affinities of unlabeled ligands. 18 This offers advantages for receptors such as P2Y 2 R for which there are currently no commercially available radio ligands. In addition, as the reported antagonists for P2Y 2 R have mid to high affinities, it is proposed that the fluorescent ligands designed from these might also have affinities in this range. This may prove problematic for techniques which directly monitor fluorescent-ligand binding, but NanoBRET has been shown to display low, nonspecific binding at high fluorescent-ligand concentrations. 18,19 ■ RESULTS AND DISCUSSION Synthesis and Evaluation of the hP2Y 2 R Antagonists. The medicinal-chemistry strategy involved an initial exploration of the structure−activity relationship (SAR) around 1 in order to enable the design of structural analogues with improved predicted physicochemical properties and to guide our design strategy by highlighting suitable linking sites to attach the fluorophore groups ( Figure 1 and Table 1).
demanding substituent, such as that in 19, resulted in a complete loss of affinity for hP2Y 2 R. We therefore explored the substitution of the amino group and showed that both compounds with sterically demanding amino groups (8, 9, and 10) and compounds with cyclic tertiary amines (11, 12, and 13) were inactive. However, the linear, less sterically demanding alkyl amino groups, such as those in 14,15,16, and 17, increased hP2Y 2 R affinity, although the bulkier 2-phenoxyethan-1-amino substituent in 18 resulted in the compound being inactive. Thus far, all of the compounds synthesized were tested as racemic mixtures. To try and determine whether the activity resided in one enantiomer, the resolution of 14 and 16 was achieved through semipreparative chiral HPLC, and the biology of each of the resolved enantiomers was independently assessed (Table 3).
From these results, it is possible to see that the hP2Y 2 Rantagonist affinities observed for the racemic compounds 14 and 16 reside largely in enantiomers 20 and 23, respectively. Although some antagonist activity is observed for 21, this may be attributed to residual active enantiomer 20, which constitutes 11% of the sample. Unfortunately, the resolved enantiomers proved to be amorphous powders, and so singlecrystal X-ray determination of their absolute chiralities could not be used for structural determination. However, vibrational circular dichroism was used for 22 and 23, from which spectra were acquired for both samples and fitted to the calculated spectra. 22−24 The results (Supporting Information) showed that there was a good match between the spectrum of 22 and the calculated spectrum for the (S)-enantiomer; therefore, 23 was assigned as the active (R)-enantiomer ( Figure 2).
In an attempt to increase affinity within the new series of compounds, we the incorporated key structural features of 14 and 5 to generate compound 24.
However, 24 did not demonstrate the expected increase in affinity from the combination of features from 14 and 5 and instead showed a level of hP2Y 2 R affinity ( pK d = 7.02 ± 0.05, n = 4) similar to those of 1 and 3, demonstrating that the SARs within the series of compounds were nonadditive. 25 The synthesis of compounds 2−4 is illustrated in Scheme 1. The alkylation of 5- (2,8-dimethyl-5H-dibenzo[a,d] [7]annulen-5-yl)pyrimidine-2,4(1H,3H)-dione 8 with ethyl 2-(bromomethyl)thiazole-4-carboxylate followed by a treatment with Lawesson's reagent and saponification gave 2, 8 which was reacted with 2-amino tetrazole via benzotriazol-1-yl-oxytripyrrolidinophosphonium hexafluorophosphate activation to afford 3. In a similar manner, alkylation with methyl iodide followed by a conversion to the thiouracil gave 4 in a good overall yield. Scheme 2 shows the synthetic route to compounds 5 and 6. The first step involved performing a Heck reaction coupling 3-chloroiodobeneze, 25, to allyl alcohol, and this successfully isomerized in situ yielding the desired aldehyde, 26. This compound was reacted with ethyl dichloroacetate in a Darzens condensation 25 to generate an α-chloro epoxide, which was reacted directly with thioacetamide to afford the desired 2-methylthiazole, 27, with moderate yields achieved over two steps. Freshly prepared sodium ethoxide, generated from sodium metal in dried ethanol, was found to be the optimal base for the Darzens condensation. Saponification gave the carboxylic acid, 28, and treatment with oxalyl chloride generated the acyl chloride, which was immediately cyclized to give the tricyclic ketone, 29, as a single regioisomer. Lithiation of the di-tert-butyl etherprotected uracil, 30, 25 was readily achieved with n-butyllithium, and this compound underwent a 1,2-addition to the ketone, 29, to give the tertiary alcohol, 32. Concomitant deprotection and dehydration resulted in uracil, 33, via heating in trifluoroacetic acid. Although the yield for this reaction was poor, other acidic conditions were ineffective. Alkylation at the N1-position of the uracil was achieved in a one-pot process of silylation with N,O-bis(trimethylsilyl)trifluoroacetamide (BSTFA), alkylation with iodomethane, and subsequent desilylation, which give 34. Finally, a reaction with Lawesson's reagent gave 6. From the uracil intermediate, 33, alkylation with ethyl 2-(bromomethyl) thiazole-4-carboxylate 26 gave 35, which was subsequently reacted with Lawesson's reagent to give 36. Hydrolysis was  (6) a The compounds were separated using Phenomenex's Lux 5 μm amylose-2 stationary phase. b The estimated affinity value for each antagonist (pK b ) was calculated using the Gaddum equation from the shift of the UTPγS concentration−response curve brought about by the addition of a single concentration of the antagonist. The data shown are the means ± SEM, and the numbers of separate experiments are given in parentheses. c IA = inactive; i.e., less than a 50% inhibition of the response to 0.1 μM UTPγS in the presence of a 10 μM concentration of the compound (n = 3).

Journal of Medicinal Chemistry
Article followed by a reaction of the resulting carboxylic acid with 5-aminotetrazole and benzotriazol-1-yl-oxytripyrrolidinophosphonium hexafluorophosphate to give 5.
In a similar sequence to Scheme 2, aldehyde 26 was reacted with ethyl dichloroacetate, and the resulting crude α-chloro epoxide reacted with thiourea to afford the 2-aminothiazole, 37, which was converted to the 2-chlorothiazole, 38 (Scheme 3). The ethyl ester was hydrolyzed to afford the carboxylic acid, 39, which was converted to the acid chloride and cyclized to the tricyclic ketone, 40. Lithiation of the tert-butyl ether-protected uracil, 31, and a reaction with 40 gave the tertiary alcohol, 41. After screening a range of milder acidic conditions, heating to 140°C with microwave irradiation in acetic acid/1,4-dioxane (1:1) for 10 min was found to be optimal for the formation of the desired uracil intermediate, 42, which was subsequently methylated to give 43. The chlorine atom in compound 43 was readily displaced through a nucleophilic aromatic substitution with a range of primary and secondary amines upon heating under basic conditions. These conditions were unsuccessful when aniline was used, and in this instance, heating in the microwave with hydrochloric acid (catalytic) proved successful in giving 46. Microwave-based conditions were employed for the substitution with ammonia in the synthesis of 45. A Suzuki reaction with phenylboronic acid gave 44. Through the use of Lawesson's reagent, it was then possible to convert these uracil derivatives (44−54) to the respective 4-thiouracil derivatives (6−19). Through a route analogous to the synthesis of compound 7, it was possible to generate the desired tetrazole analogue, 24. Uracil intermediate 42 was alkylated at the N1-position with ethyl 2-(bromomethyl)thiazole-4-carboxylate to give 57. This compound was reacted with 2-methoxyethylamine to afford 58, which was subsequently reacted with Lawesson's reagent to give the 4-thiouracil, 59. Hydrolysis followed by benzotriazol-1yl-oxytripyrrolidinophosphonium hexafluorophosphate activation and a reaction with 5-aminotetrazole afforded 24.
Synthesis of the hP2Y 2 R Fluorescent Ligands. With a view to developing a series of fluorescent conjugates suitable for both a bioluminescence-resonance-energy-transfer (BRET) ligand-binding assay 18,27−29 and imaging via confocal microscopy, we embarked on a strategy to synthesize BODIPY conjugates, specifically with the dyes BODIPY A (628 nm absorption max, 642 nm emission max) or BODIPY B (503 nm absorption max, 509 nm emission max) as this would allow us the opportunity of ligand choice in future imaging work.
Two positions on the P2Y 2 R-antagonist core structure were considered for the attachment of the linker and fluorophore ( Figure 3). In order to simplify the synthetic chemistry and increase the SARs within the series, we examined the replacement of the furan ring of 1 and the thiazole ring of 2 with 1,3,5trisubstituted phenyl rings. This would allow the attachment of the acyl tetrazole group in addition to providing a second free carboxylic acid group to which the fluorescent conjugates could be attached. Fortunately, the 3,5-dicarboxylic acid on 60 (hP2Y 2 R pK d = 6.53 ± 0.04, n = 7) was well tolerated with no loss of affinity for P2Y 2 R compared with the affinity of compound 2. Therefore, the first series of compounds had the linker fluorophore attached via the phenyl ring of 60.
Having established that alkoxyalkyl amines are tolerated in terms of activity in the 2-position of the thiazole in the 4H-benzo [5,6]cyclohepta [1,2-d]thiazol-4-yl) tricyclic rings of compounds of the type shown in Figure 2, this position was chosen as the second point of attachment of the linker fluorophore. Finally, a third generation of fluorescent ligands were explored that contained the optimal second-generation fluorescent ligand with an incorporated acyl tetrazole functional group.
First-Generation hP2Y 2 R Fluorescent Ligands. The general synthetic route for the first-generation P2Y 2 R fluorescent antagonists is shown in Scheme 4.
To determine whether any of these conjugates had affinities for P2Y 2 R and consequently if they could be used in

Article
NanoBRET binding assays, a 1321N1 astrocytoma cell line expressing recombinant P2Y 2 R tagged on its N-terminus with NLuc (NLuc-P2Y 2 R) was generated. The NLuc-tagged P2Y 2 receptors exhibited normal calcium signals (UTPγS EC 50 = 91 ± 12 nM, n = 3). These NLuc-P2Y 2 cells were treated with increasing concentrations of 66−69 and then treated with the NLuc substrate, furimazine, before the resulting BRET signal was monitored. All four compounds showed moderate to low affinities for the NLuc-P2Y 2 R (Table 4), with conjugates 68 and 69 having higher affinities. However, this did illustrate the power of using NanoBRET to monitor ligand binding to low-affinity receptors.
Second-Generation hP2Y 2 R Fluorescent Ligands. The general synthetic route for the second-generation P2Y 2 R fluorescent antagonists, in which the linker fluorophores are attached to the thiazole rings, is illustrated in Scheme 5.

Journal of Medicinal Chemistry
Article (2-(2-aminoethoxy)ethyl)carbamate to generate the corresponding amides (74−77). The Boc protecting groups were removed using TFA, and the resulting amines were coupled with the appropriate commercially available BODIPY succinimidyl ester (SE), generating a small library of 14 fluorescent conjugates (78−91). These fluorescent conjugates were initially tested using the aforementioned NanoBRET binding assay at fixed concentrations of 10 μM in the presence or absence of 10 μM 1  (see Figure S1 in the Supporting Information). It was found that 80, 85, 86, and 87 generated the largest specific NanoBRET signals; therefore, their affinities were determined from saturation binding assays, demonstrating the excellent signal-to-noise ratios observed from NanoBRET assays even for low-affinity conjugates ( Figure 4 and Table 5).
Reassuringly, the affinities determined for the nonfluorescent P2Y 2 R antagonist, 1, with either 86 (pK i = 7.45 ± 0.13, n = 4) or 87 (pK i = 7.32 ± 0.13, n = 4) were consistent with the affinities determined using the P2Y 2 R functional assay (pK b = 7.51 ± 0.09), demonstrating that the P2Y 2 R fluorescent ligands could be used in a NanoBRET assay for determining the affinities of nonfluorescent P2Y 2 R antagonists. The clear demonstration of the saturable specific binding of these low-affinity fluorescent ligands confirmed the utility of the NanoBRET binding format and the ability to exploit the good signal-tonoise ratio of this proximity-based assay. To explore the opportunity to develop higher-affinity fluorescent conjugates, we embarked on a synthetic strategy to incorporate affinityenhancing acyl-tetrazole functional groups into the fluorescent compounds.
Third-Generation P2Y 2 R Fluorescent Ligands. The general synthetic route for the third-generation P2Y 2 R fluorescent antagonists is illustrated in Scheme 6. The displacement of the chlorine atom in compound 57 with tert-butyl 3-aminopropanoate gave the tert-butyl ester, 92. Treatment with TFA afforded the conversion of the tert-butyl ester to the corresponding acid, which was immediately activated and coupled to tert-butyl (2-aminoethyl)carbamate, 93, with HATU. Hydrolysis of the ethyl ester, activation of the carboxylic acid by benzotriazol-1-yl-oxytripyrrolidinophosphonium hexafluorophosphate, and coupling with 5-aminotetrazole gave the amidotetrazole, 94. The Boc protecting group was removed, a See Scheme 4. b A and B correspond to the R groups in Scheme 4. c The pK d values were derived from saturation binding curves. The data shown are the means ± SEM, and the numbers of separate experiments are given in parentheses.

Journal of Medicinal Chemistry
Article and the resulting amine was attached to the fluorophore with the appropriate BODIPY succinimidyl ester (Scheme 6). In contrast to the nonfluorophore compounds 14 and 24, whose affinities hardly changed when their uracil N1-substituents were changed, in three of the four compounds (95−98), there were significant increases in their affinities relative to those of the nonamidotetrazole series (compare 98 with 86, 96 with 85, and 97 with 80; Tables 5 and 6).
Pharmacological Evaluation of Third-Generation P2Y 2 R Fluorescent Ligands. To further evaluate the utility of the fluorescent ligands in studying the pharmacology of P2Y 2 R, one BODIPY A and one BODIPY B linked fluorescent ligand (97 and 98, respectively) were chosen for further studies. Initially, we confirmed that 97 and 98 still retained the ability to functionally antagonize P2Y 2 R ( Figure 5). In a Ca 2+ -mobilization assay, modest rightwards shifts of the agonist-dose−response curves were observed for both 10 μM 97 (pK b = 5.69 ± 0.05, n = 8) and 10 μM 98 (pK b = 5.87 ± 0.05, n = 7).
To further evaluate the utility of fluorescent conjugate 98 in the NanoBRET ligand-binding assay, the affinities of a selection of P2Y 2 R antagonists (1, 3, 6, 22, 23, 60, and 86) and the previously reported stabilized-triphosphate P2Y 2 R antagonist, 99, 8 over a range of P2Y 2 R affinities, were determined in competition binding experiments.
All eight compounds induced concentration-dependent decreases in the specific binding of 98, which enabled their affinities to be determined. There was a good correlation between the values obtained in the NanoBRET assay and those determined in the Ca 2+ -mobilization assay ( Table 7). In addition to those of the antagonists, the NanoBRET assay was also used to estimate the affinity of UTPγS. As there have been no reports of radio ligands for P2Y 2 R, this measurement has not previously been possible.
The availability of both green (98) and red (97) fluorescent P2Y 2 R ligands with reasonable affinities for hP2Y 2 R suggested that they may both have utility in imaging the receptor in living cells. Confocal-microscope images of fluorescent ligands 97 and 98 with 1321N1 astrocytoma cells expressing hP2Y 2 R (Figure 7a,c) showed localized membrane fluorescence and very little intracellular fluorescence. When the cells were pretreated with 1, the membrane-specific fluorescence of 97 and 98 was substantially reduced (Figure 7b,d), indicating that the majority of the membrane fluorescence observed was specific labeling of P2Y 2 R.

Journal of Medicinal Chemistry
Article ■ CONCLUSION We have described the synthesis and evaluation of new examples of acidic hP2Y 2 R antagonists based on the known hP2Y 2 R antagonist, 1. In addition, we have shown the discovery of a new series of neutral hP2Y 2 R antagonists and demonstrated SAR leading to the identification of potent hP2Y 2 R antagonists (such as 20 and 23). In addition, we have shown a stereochemical preference for biological activity within this series, as typified by the resolved examples of 20 versus 21 and 22 versus 23. Vibrational circular dichroism has suggested that in the case of 16, all of the hP2Y 2 R biological activity resides in the (R)-enantiomer (23), although single-crystal X-ray work will be required to confirm this initial stereochemical assignment. The SAR studies led to the identification of suitable linking sites for the attachment of the fluorescent ligands, thus generating three distinct series of fluorescently labeled hP2Y 2 R antagonists. From this extensive synthetic work, two examples (97 and 98) were identified as demonstrating both functional antagonist activities (in Ca 2+ -mobilization assays) and sufficient affinities for hP2Y 2 R through a new bioluminescenceresonance-energy-transfer (BRET) assay. In addition, confocal microscopy revealed clear, displaceable membrane labeling of astrocytoma cells expressing hP2Y 2 R. These excellent imaging properties make 97 and 98 ideal tools for studying hP2Y 2 R distribution and organization. Finally, the discovery of the new hP2Y 2 R-antagonist fluorescent ligands (97 and 98) became realized as a result of an extensive program of synthetic chemistry, in which it proved essential to explore the parallel changes of linker-attachment points, fluorophores, and linking groups. 30 From this study, only a few fluorescent conjugates were shown to possess sufficient affinities to enable the establishment of a new NanoBRET-based fluorescent assay for the identification of new hP2Y 2 R fragments and ligands.

Journal of Medicinal Chemistry
Article respectively. The NMR data was processed using iNMR (version 5.5.7), which referenced the spectra to those of the residual solvents. Chemical shifts (δ) are quoted as values in parts per million, and coupling constants (J) are given in hertz. Multiplicities are described using the following abbreviations: s, singlet; d, doublet; t, triplet; q, quartet; qi, quintet; sep, septet; m, multiplet; app, apparent; and br, broad. All compounds submitted for biological screening had purities >95%.

Journal of Medicinal Chemistry
Article using a 40× c-Apochromat 1.2NA water-immersion objective. For 97, the images were collected using a 633 nm excitation wavelength and a 488/561/633 dichroic, and the emissions were collected through a 650LP filter. For 98, a 488 nm excitation wavelength was used with the same dichroic, and the emissions were collected using an LP575 filter. In each case, a pinhole diameter of 1 airy unit was used, and the laser power, gain, and offset were kept the same for all of the samples within each experiment. For both 97 and 98, the images presented are as representative of an individual experiment with matched conditions. Linear adjustments to the image brightness and contrast have been applied equally across all of the comparative images using Zen software in order to prepare the images for presentation.
Data Analysis. All of the data were analyzed using GraphPad Prism 6.
For the calcium-mobilization experiments, as none of the compounds synthesized as part of this study showed any partialagonist actions, the estimated affinity values (pK b ) were calculated from the shifts of the agonist-concentration−response curves in the presence of the fluorescent antagonists using eq 1: where DR (dose ratio) is the ratio of the agonist concentration required to stimulate an identical response in the presence and absence of the antagonist, b. The pK b is −log K b . The total and nonspecific saturation binding curves were fitted simultaneously using eq 2: where B max is the maximum specific BRET signal, [b] is the nanomolar concentration of the fluorescent ligand, K d is the equilibrium dissociation constant in nanomolar, M is the slope of the nonspecificbinding component, and C is the intercept with the Y-axis. The pK d is −logK d . The competition binding curves were fitted using eq 3: where [L] is the nanomolar concentration of 98, and K d is the equilibrium dissociation constant of 98 in nanomolar. The IC 50 was calculated as in eq 4:

Journal of Medicinal Chemistry
Article