Discovery of Highly Potent Adenosine A₁ Receptor Agonists: Targeting Positron Emission Tomography Probes

Abstract

Adenosine receptor (AR) radiotracers for positron emission tomography (PET) have provided knowledge on the in vivo biodistribution of ARs in the central nervous system (CNS), which is of therapeutic interest for various neuropsychiatric disorders. Additionally, radioligands that can image changes in endogenous adenosine levels in different physiological and pathological conditions are still lacking. The binding of known antagonist adenosine A₁ receptor (A₁R) radiotracer, [¹¹C]MDPX, failed to be inhibited by elevated endogenous adenosine in a rodent PET study. Since most of the known AR PET radiotracers were antagonists, we propose that an A₁R agonist radioligand may possess higher sensitivity to measure changes in endogenous adenosine concentration. Herein, we report our latest findings toward the development of a full agonist adenosine A₁ radioligand for PET. Based on a 3,5-dicyanopyridine template, 16 new derivatives were designed and synthesized to optimize both binding affinity and functional activity, resulting in two full agonists (compounds 27 and 29) with single-digit nanomolar affinities and good subtype selectivity (A₁/A_2A selectivity of ~1000-fold for compound 27 and 29-fold for compound 29). Rapid O-[¹¹C]methylation provided [¹¹C]27 and [¹¹C] 29 in high radiochemical yields and radiochemical purity. However, subsequent brain PET imaging in rodents showed poor brain permeability for both radioligands. An in vivo PET study using knockout mice for MDR 1a/a, BCRP, and MRP1 indicated that these compounds might be substrates for brain efflux pumps. In addition, in silico evaluation using multiparameter optimization identified high molecular weight and high polar surface area as the main molecular descriptors responsible for low brain penetration. These results will provide further insight toward development of full agonist adenosine A₁ radioligands and also highly potent CNS A₁AR drugs.

Graphical Abstract

INTRODUCTION

Adenosine is an important neuromodulator in the central nervous system (CNS) implicated in many physiological processes and neurological disorders.^1–4 For the last two decades, various adenosine A₁ and A_2A receptors (A₁R and A_2AR) radiotracers for positron emission tomography (PET) have been developed to investigate the in vivo CNS adenosine system in both animals and humans. In general, antagonist radiotracers, such as [¹¹C]preladenant (1),⁵ [¹¹C]MDPX (2),⁶ [¹⁸F]CPFPX (3),⁷ (Figure 1) allow to measure adenosine receptor density, but as of now have been unable to measure endogenous adenosine,^8,9 which would be valuable to investigate neuropathology.^1,10,11 For [¹¹C]MDPX, the studies to assess sensitivity in competition with endogenous adenosine showed negative results.^8,9 In general, PET agonist radiotracers are more sensitive to changes in endogenous agonist concentrations.¹² In the case of AR radiotracers, the only reported agonist [¹⁸F]FNECA (4) derived from endogenous adenosine did not cross the blood—brain barrier (BBB),^13,14 presumably due to its high polar surface area resulting from the hydrophilic ribose substituents.

Figure 1.

Representative A₁R PET ligands.

To overcome this problem, we aimed to develop agonist radiotracers based on a known 3,5-dicyanopyridine scaffold.¹⁵ Our previous study demonstrated that the 3,5-dicyanopyridine template could offer a wide range of chemical and functional properties, which could be fine-tuned en route to develop subtype selective agonist radioligands.¹⁵ For example, [¹¹C]MMPD (5) is a potent (K_i = 0.49 nM), selective A₁R partial agonist with high BBB permeability.¹⁴ These properties also make them a great starting point for developing selective partial agonist drug candidates without adverse cardiovascular side effects, such as high-grade atrioventricular block, extensive bradycardia, atrial fibrillation, and vasodilation.¹⁶

Herein, we report our recent progress toward the second generation of adenosine A₁ agonist radioligands based on a 3,5-dicyanopyridine pharmacophore. Previously, through the systematic comparison of two similar templates, namely 3,5-dicyanopyridine and 5-cyanopyrimidine, we concluded that the 3,5-dicyanopyridine pharmacophore is superior in terms of binding affinity toward A₁R and A_2AR. Therefore, we focused solely on the 3,5-dicyanopyridine core and synthesized a library of molecules, studying their structure–activity relationship (SAR) for binding and functional activity, along with PET imaging in rodents and in silico estimation of BBB permeability with multiparameter optimization (MPO).

RESULTS AND DISCUSSION

Design, Synthesis, and SAR Study.

Taking into consideration physicochemical guidelines for CNS PET radiotracers,^17,18 we explored the SAR of 3,5-dicyanopyridine derivatives to optimize both functional activity and binding affinity, as shown in Scheme 1. The A-ring was modified with an acetamido substituent in the para-position and hydroxy or alkoxy group in the meta-position, since they were shown to influence both binding and functional activities.¹⁵ A strategy of introducing heterocycles in the B-ring led to the synthesis of compounds 10–30, where most compounds were designed for facile radiolabeling with [¹¹C] or [¹⁸F]. We screened the binding affinity of compounds 10–30 through a detailed SAR study for A₁R, A_2AR, and A₃R subtypes and investigated their selectivity for the A₁R subtype to further refine our prediction of target molecules as PET imaging probes (Table 1). It is noteworthy that compounds 10, 12,¹⁹ 14, 15,²⁰ and 24¹⁵ have been synthesized elsewhere, while compounds 11, 13, and 16–29 were newly synthesized and reported herein. Moreover, compound 30 in this series is a 5-cyanopyrimidine congener of the 3,5-dicyanopyridine 27.

Scheme 1.

Parallel Synthesis of 3,5-Dicyanopyridines for a Potential Adenosine Agonists^a

^aReagents and conditions: (i) Malononitrile, piperidine, EtOH, reflux, 2 h; (ii) malononitrile, PhSH, Et₃N, EtOH, reflux, 5 h; (iii) Na₂S, DMF, 80 °C; (iv) HCl (aq); (v) Et₃N, MeCN, 50 °C, overnight.

Table 1.

SAR of Compounds 10–30 at Adenosine Receptors

Entry	R1	R2	R3	Affinity, Ki, nMa (or %inhib.)b			A1 functional efficacy (Emax %)c
				A1	A2A	A3
10		-H		53±18	2.8±0.1	22±6.7	122±5.2%
11		-H		67±23	38±7.3	35±11%	−5.2±1.9%
12		-H		217±44	10±2.0	8.2±1.1	98±2.5%
13		-H		109±30	27±6.6	156±45	141±13%
14	-H	-OH		9.6±4.0	32±0.4	30±15	102±2.0
15	-H	-OMe		3.1±0.4	38±5.5	71±13	23±3.9%
16	-H	-OH		41±19	81±15	55±14%	13±0.1%
17	-H	-OMe		4.2±0.4	155±36	45±10%	−3.0±2.4%
18	-H	-F		13±2.6	43±5.9	261±82	32±5.4%
19	-H	-OMe		13±1.9	44±5.8%	44±8.4%	14±7.5%
20	-H	-OMe		2.7±1.0	74±3.3	52±9.3%	2.1±3.1%
21	-H	-OMe		2.7±0.3	173±24	38±16%	−1.2±1.7%
22	-H	-OMe		6.2±0.1	318±72	42±12%	2.6±2.9%
23	-H	-OMe		39±5.9	34±14%	32±11 %	3.2±2.7%
24	-H	-OH		3.5±1.3	15±3.1	74±3.4	109±7.8%
25	-H			18±4.8	709±198	378±81	−1.0±3.6%
26	-H	-F		16±4.6	72±24	50±11%	20±3.5%
27		-OCH2O-		1.6±0.4	54±8.5%	47±16%	104±5.9%
28		-OCH2O-		2.3±0.7	47±12%	26±13%	92±2.9%
29	-H	-H		6.1±1.1	175±31	507±35	97±18%
30				29±4.2	13±6.8%	25±12%	54±27%

^aCompetition radioligand binding assays were conducted with membranes prepared from HEK-293 cells expressing recombinant A₁, A_2A; or A₃Rs (human). The incubation was performed for 1 h at 25 °C. The radioligands used were: for A₁R, [³H]8-cyclopentyl-1,3-dipropylxanthine (0.5 nM); for A_2AR, [³H]ZM241385 (0.8 nM); and for A₃R, [¹²⁵I]N⁶-(4-amino-3-iodobenzyl)adenosine-5′-N-methyluronamide ([¹²⁵I]I-AB-MECA, 0.1 nM). Nonspecific binding was determined using 10 μM 8-[4-[[[[(2-aminoethyl)amino]carbonyl]-methyl]oxy]phenyl]-1,3-dipropylxanthine, 10 μM for A₁ and A_2A and 100 μM for A₃. Values are expressed as the mean ± SEM from 2–4 independent experiments.

^bPercent inhibition at 1 μM.

^ccAMP accumulation assay (N⁶-cyclopentyladenosine expressed as 100%).

Previously, compound 10 was reported as a potent A_2AR agonist and a weak A₁R agonist.¹⁵ When the pyridyl ring was replaced with an imidazolyl or phenyl group (12 and 13), binding affinity was significantly reduced for both A₁R and A_2AR, while their full A₁R agonism was maintained. This is presumably due to the presence of the acetamido group in para-position of the A-ring. Since we had previously reported¹⁵ that 3,5-dicyanopyridine compounds bearing an acetamido group showed inadequate BBB permeability due to a high polar surface area (PSA), we also synthesized N-methyl acetamido derivative (11) to reduce the PSA, considering the possible radiolabeling route as well as BBB penetration. Interestingly, agonistic activity was completely abolished, indicating the sensitive interaction of this position with A₁R.

A similar trend on affinity was also observed at the meta-position of the A-ring in this compound series with an 2-imidazolyl group in the B-ring (14–17). For instance, the A₁R binding affinity was improved by replacement of phenol with m-anisole as the A-ring, but the agonistic activity decreased significantly, which was consistent with our previously published results.¹⁵ In compounds 16 and 17, methylation on a nitrogen of the imidazole ring also lowered binding affinity, particularly for A₁R and A₃R subtypes, along with lower A₁R agonistic activity.

It became clear that a hydroxy group and a methoxy group on the A-ring were critical for functional and binding activities, respectively. We therefore investigated replacing those groups in compound 18 with fluoro, which is a bioisosteric replacement for hydroxy and methoxy groups and is also a site for the introduction of a radioisotope. Unfortunately, the A₁ binding affinity was not improved compared with the parent compounds 14 and 15, and functional activity was only slightly improved, compared with the methoxy substituent.

In the next step, the imidazolyl group in B-ring was replaced with various heteroaryl groups (19–23) to examine their influence on functional activity at the A₁R, while the OMe group was retained to maintain the higher binding potency. In all these cases, A₁R functional activity was not improved and the A₁R binding affinity worsened with the large benzothiazole group in compound 23.

Since the methoxy group in MMPD (5) decreased functional activity despite its increased binding potency,¹⁵ we also replaced this substituent with the bioisosteric F and fluoroethoxy groups to give compounds 25 and 26. Although the A₁R agonism was slightly improved, the binding affinities were significantly reduced.

Lastly, inspired by a previously reported A₁R full agonist,²¹ 30, it was reasoned that a phenylthiazole group in the B-ring might improve functional activity. Surprisingly, while 30 turned out a weak A₁R partial agonist (K_i = 29 nM, A₁R E_max = 54%) in our assay setup, all three new compounds (27–29) showed full agonism. Furthermore, the binding affinity was also significantly improved, reflecting that the 3,5-dicyanopyridine platform was superior to its 5-cyanopyrimidine congener. We chose to pursue PET experiments of 27 since they demonstrated full agonism with a single digit nanomolar affinity (K_i = 1.6 nM, A₁R E_max = 104%) along with compound 29, a smaller variant of 27 with low molecular weight and yet full A₁R agonist functional activity. Based on known procedures,^22,23 both A₁R full agonists turned out to be highly subtype selective not only against A₂A and A₃ but also against the A_2B subtype (27, K_i = 436 ± 115 nM; 29, K_i = 272 ± 99 nM).

It is also worth mentioning that compound 12 was serendipitously discovered to possess high binding affinity toward the A₃R subtype (8.2 nM) with partial agonism (A₃R E_max = 60%). Since most known A₃R agonists are nucleoside derivatives,⁴ compound 12 could be a structural basis for BBB penetrable molecular probes as well as for the development of pain medications targeting A₃R.⁴

Radiochemistry.

Two full agonists (27 and 29) were chosen for radiolabeling with [¹¹C] for further in vivo imaging studies. The precursors (34 and 36) were synthesized by coupling of the corresponding free thiols (33 and 35) with 4-(chloromethyl)-2-(4-hydroxyphenyl)thiazole (32). Compound 32 was synthesized through a demethylation reaction with excess boron tribromide from the commercially available 4-(chloromethyl)-2-(4-methoxyphenyl)thiazole (31), as shown in Scheme 2.

Scheme 2.

Synthetic Reagents and Conditions for Synthesis of [¹¹C]27 and [¹¹C]29^a

^aReagents and conditions: (i) Anhydrous DCM, 10 equiv BBr₃, 0 °C to rt, overnight; (ii) Et₃N, MeCN, 50 °C, overnight; (iii) [¹¹C]MeOTf, MeCN, TBAOH, 80 °C, 3 min; (iv) Et₃N, MeCN, 50 °C, overnight; (v) [¹¹C]MeOTf, MeCN, TBAOH, 80 °C, 3 min.

Both [¹¹C]27 and [¹¹C]29 were successfully radiolabeled via O-methylation using [¹¹C]methyl triflate²⁴ in moderate RCY ([¹¹C]27, 22.2 ± 5.8% nondecay corrected, n = 6; and [¹¹C] 29, 24.3 ± 6.7%, nondecay corrected, n = 5) and high molar activities ([¹¹C]27, 832 ± 411 GBq/μmol @ EOB; and [¹¹C] 29, 1195 ± 492 GBq/μmol @ EOB). Radiochemical purity was also high ([¹¹C]27, 99%<, n = 6; and [¹¹C]29, 99%<, n = 5), and the averaged total synthesis time was 32 min. It is worth noting that O-[¹¹C]methylation using [¹¹C]methyl iodide also led to the formation of [¹¹C]27 and [¹¹C]29, albeit in significantly lower RCYs of 3–7% (n = 2).

PET Imaging Studies in Rodents.

Initially, preclinical PET scans were performed on male Wistar rats with [¹¹C]27 (Figure 2A). Compared with [¹¹C]MMPD (5),¹⁵ the averaged whole brain uptake of [¹¹C]27 was very low (standard uptake value (SUV) = 0.28 ± 0.03 g/mL, Figure 3A) with homogeneous distribution, which was reasoned to be due to the high polar surface area of 27 (tPSA, 127.1). We then tested compound 29 (Figure 2B) as its PSA is lower than that of 27 and is similar to that of MMPD (108.6). However, [¹¹C]29 also a showed lack of BBB permeability, though it was slightly improved from that of [¹¹C]27 (Figure 3A).

Figure 2.

In vivo PET imaging using [¹¹C]27 and [¹¹C]29. Averaged PET images (0–70 min) were normalized with injection dose and body weight into a single representative SUV image (upper panels) and coregistered to a Wistar rat MRI brain template (lower panels).

Figure 3.

Time–activity curves of (A) [¹¹C]27 and [¹¹C]29 baseline. (B) [¹¹C]29 baseline and pretreatment with DPCPX (2 mg/kg) in the same subject. No significant reduction of brain uptake with DPCPX pretreatment was observed, indicating little binding specificity of [¹¹C]29.

To investigate lack of BBB permeability, ex vivo studies of [¹¹C]27 were performed using triple knockout (tKO) mice for the three most abundant brain efflux pumps, MDR 1a/a, BCRP, MRP1, along with control mice. As shown in Figure 4, the averaged brain uptake of [¹¹C]27 of tKO mice was four times higher than that of control mice, reflecting that [¹¹C]27 is a substrate of these cell membrane proteins. In short, [¹¹C] 27 is likely to be a substrate for either MDR 1a/a or MRP1, not BCRP (refer to SI).

Figure 4.

Comparison of whole brain SUV of [¹¹C]27 in control mice (n = 3) and tKO mice (n = 3) for the three most abundant brain efflux pumps MDR 1a/a, BCRP, and MRP1. The difference between average tKO mice SUV (0.165 ± 0.026) and control mice SUV (0.037 ± 0.006) was statistically significant using a two-sample t test of unequal variance with one tail.

In Silico Calculation: BBB Permeability.

MPO analysis was also used to uncover molecular properties which may be responsible for their lack of BBB permeability. According to Wager’s method,²⁵ CNS scores were generated and compared for all the compounds in this report including previously published radioligands by our group (Figure 5A).¹⁵ The CNS MPO score was calculated using the StarDrop software package by Optibrium Inc.^26,27 as the sum of the six physicochemical parameters, namely molecular weight (M_w), pK_a of the most basic center, calculated logP (clogP), calculated logD at pH 7.4 (clogD), topological polar surface area (tPSA), number of hydrogen-bond donors (HBD).

Figure 5.

(A) MPO CNS scores for selected adenosine A₁R (radio)ligands. [¹¹C] Compounds shown as white circles, and all other compounds shown as dark circles. [¹¹C]A = N-(4-(2-amino-3,5-dicyano-6-((methyl-¹¹C)thio)pyridin-4-yl)phenyl)acetamide, [¹¹C]B = N-(4-(6-amino-5-cyano-2-((methyl-¹¹C)thio)pyrimidin-4-yl)phenyl)acetamide, [¹¹C]C = 2-amino-4-(4-methoxyphenyl)-6-((methyl-¹¹C)thio)pyridine-3,5-dicarbonitrile, [¹¹C]D = 4-amino-6-(4-methoxyphenyl)-2-((methyl-¹¹C)thio)pyrimidine-5-carbonitrile, and [¹¹C]5 = 2-amino-4-(3-(methoxy-¹¹C)phenyl)-6-(((6-methylpyridin-2-yl)methyl)thio)pyridine-3,5-dicarbonitrile. (B) Correlation of whole brain uptake area under the curve (AUC) vs MPO CNS score of [¹¹C] labeled 3,5-dicyanopyridine derivatives synthesized in our laboratory. AUC was calculated from time–activity curves presented as the SUV using the trapezoidal sum from 0 to 30 min. Linear regression calculated using sum of least-squares residuals (R² = 0.911).

While most of the previously reported brain penetrable radiolabeled compounds based on the 3,5-dicyanopyridine template ([¹¹C]C, [¹¹C]D, [¹¹C]5)¹⁵ showed high MPO CNS scores (>5), both [¹¹C]27 and [¹¹C]29 showed relatively low (<4) scores (Figure 5A). Overall, the brain uptake was highly correlated with the CNS score (R² = 0.91) when compared with AUCs for 30 min of each radiolabeled compound (Figure 5B). Initial brain uptake (t < 2.5 min) was also highly correlated with the CNS score (R² = 0.78, SI). Therefore, we concluded that the main molecular property that limits the BBB permeability of [¹¹C]27 and [¹¹C]29 is their molecular weights despite their favorable calculated lipophilicity (clogP).

CONCLUSION

In summary, we developed A₁R full agonist ligands with nanomolar affinity based on a 3,5-dicyanopyridine pharmacophore through comprehensive SAR studies and optimized both binding affinity and functional activity. Compounds 27 and 29 were chosen for labeling with [¹¹C]MeOTf for preclinical brain imaging. Our rodent PET imaging studies indicated a lack of BBB permeability of these full agonists, consistent with low scores in predictions of physicochemical properties for CNS radioligand candidates. Although no successful CNS adenosine full agonist radioligand was discovered herein, this report provides valuable SAR information that might be broadly applicable in further studies of AR PET ligands including peripheral adenosine imaging as well as in potential pharmaceutical development.

METHODS

Detailed synthesis and characterization of all the new compounds are presented in the SI.

Radiosynthesis of Compound [¹¹C]27 and [¹¹C]29.

A solution of precursor (0.8–1.0 mg) was suspended in 200 μL of anhydrous MeCN, and 2.5 μL of tetrabutylammonium hydroxide (1 M in methanol) was added through the reaction flask wall. The mixture was then vortexed for 1 min which led to formation of a clear yellow solution. The mixture was allowed to react with [¹¹C]CH₃I in a stream of helium at 80 °C for 3 min before being injected into a semipreparative HPLC column (Phenomenex Onyx Monolithic C18 LC Column 100 × 10 mm). The mixture was eluted at 5 mL/min with an isocratic mixture of 60% solvent A (90% 0.01 M phosphate buffer, 10% EtOH pH = 7.2–7.4) and 40% solvent B (100% EtOH) and monitored for absorbance at 280 nm and radioactivity using the flow count detector (NaI(Tl)) built into the FX-M. The product [¹¹C]27 was collected in 10.0 min ([¹¹C]27), while the product [¹¹C] 29 was collected in 11.9 min (Figure S2), and radioactivity was measured by a dose calibrator (Capintec, CRC 712M) to determine RCY. RCY for [¹¹C]27 was calculated to be 22.2 ± 5.8% (n = 6) and molar activity (832 ± 411 GBq/μmol @ EOB, n = 4). The collected product solution was formulated with 2.5 mL of sterile water for rodent PET studies (final ethanol content ≤10%). RCY for [¹¹C]29 24.3 ± 6.7%, n = 5) and molar activity (1195 ± 492 GBq/μmol @ EOB, n = 5). The collected product (1 mL) solution was formulated with 3.0 mL of sterile water (final ethanol content ≤10%) for rodent PET studies. Radiochemical synthesis data obtained from the Tracerlab FXM for compounds [¹¹C]27 and [¹¹C]29 are shown in Figures S1 and S2.

PET Imaging.

For PET studies, rats were anesthetized under isoflurane (Forane, 99.9%; 5.0% induction for 5 min, 1.0–2.5% maintenance) prior to catheter placement and for the duration of scanning. A catheter was inserted into the penile vein for radiotracer injection. Subjects were placed prone position side-by-side into a Siemens microPET Focus 220 scanner. Vitals (heart rate, respiratory rate, spO₂, body temperature) were monitored using a PhysioSuite (Kent Scientific no. PS-04). Temperature was maintained close to 36 °C with a homeothermic blanket with negative feedback control (Harvard Apparatus no. 507222F). A 10 min transmission scan with a Co-57 point source was collected for attenuation correction prior to 90 min emission scans. Radiotracer was injected as a bolus over 1 min using a syringe pump (Harvard Apparatus no. HA1100WD) and immediately flushed with heparinized saline (250 μL). The injected dose was 16.6 ± 3.2 MBq. For the [¹¹C]29 blocking study, DPCPX (2.0 mg/kg, 550 μL) was injected intraperitoneally 10 min prior to radiotracer injection. PET data were collected in list mode and reconstructed into 23 frames (6 × 20 s, 5 × 60 s, 4 × 120 s, 3 × 300 s, 3 × 600 s, and 2 × 1200 s), and sinogram reconstruction was performed using 2D Filtered Back Projection. Time–activity curves were generated in PMOD (version 3.807) and normalized to subject weight and injected activity and represented as SUV. Averaged image (0–70 min) was generated to evaluate BBB permeability for each compound.

Ex Vivo Studies.

For ex vivo studies, all mice were placed under anesthesia as described above. Catheters were constructed using 20 cm BPTE-10 polyethlyene tubing (Instech Las), sharp tip needles (27 GA, Becton Dickinson), and blunt tip needles (30 GA, Component Supply). Catheters were flushed with heparinized saline (0.6% heparin, 0.9% HCl saline) and placed in the tail vein. [¹¹C]27 (2.63 ± 1.54 MBq, 100 μL) was administered intravenously and immediately flushed with heparinized saline (100 μL). Animals were euthanized 15 min postinjection via decapitation. Whole brain tissue and ventricular whole blood were extracted, stored in preweighed glass vials, and placed in an automatic well-type gamma counter (Wallac Wizard 3″; PerkinElmer) to measure radioactivity. Vials were postweighed to obtain tissue mass, and the SUV was generated using total injected activity and subject weight.

ABBREVIATIONS

AR

adenosine receptors

PET

positron emission tomography

CNS

central nervous system

BBB

blood—brain barrier

[¹¹C]MDPX

8-dicyclopropylmethyl-1-¹¹C-methyl-3-propyl-xanthine

MDR

multidrug resistance

BCRP

breast cancer resistance protein

MRC1

multidrug resistance-associated protein 1

[¹⁸F]CPFPX

8-cyclopentyl-3-(3-(fluoro-¹⁸F)-propyl)-1-propyl-1,2,3,7-tetrahydro-6H-purin-6-one

FNECA

(2S,3S,4R,5R)-5-(6-amino-9H-purin-9-yl)-N-(2-(fluoro-¹⁸F)-ethyl)-3,4-dihydroxytetrahydrofuran-2-carboxamide

[¹¹C]MMPD

2-amino-4-(3-(methoxy-¹¹C)phenyl)-6-(((6-methylpyridin-2-yl)methyl)thio)pyridine-3,5-dicarbonitrile

SAR

structure–activity relationship

SUV

standard uptake value

AUC

area under the curve

RCY

radiochemical yield

MPO

multiparameter optimization

PSA

polar surface area

tPSA

topological polar surface area

clogP

calculated logP

clogD

calculated logD at pH 7.4

HBD

number of hydrogen-bond donors