Synthesis and Cardiac Imaging of ¹⁸F-Ligands Selective for β₁-Adrenoreceptors

Abstract

A series of potent and selective β₁-adrenoreceptor ligands were identified (IC₅₀ range, 0.04–0.25 nM; β₁/β₂ selectivity range, 65–450-fold), labeled with the PET radioisotope fluorine-18 and evaluated in normal Sprague–Dawley rats. Tissue distribution studies demonstrated uptake of each radiotracers from the blood pool into the myocardium (0.48–0.62% ID/g), lung (0.63–0.97% ID/g), and liver (1.03–1.14% ID/g). Dynamic μPET imaging confirmed the in vivo dissection studies.

The β₁- and β₂-adrenoreceptors (β-AR) play an important role in the regulation of heart function and have been extensively studied in recent decades,^1,2 as changes in the number and ratio of cardiac β-ARs have been associated with several diseases of the heart including myocardial ischemia,³ congestive heart failure,⁴ cardiomyopathy,⁵ and hypertention.⁶ In the healthy human heart, β₁-AR is the dominant subtype (4:1 ratio, β₁:β₂-AR) and widely distributed. However, in the failing heart, a predictable decrease in cardiac β₁-AR density (3:2 ratio, β₁:β₂-AR density observed in heart tissue of postmortem heart failure patients), coupled with enhanced sympathetic activity, boosting cardiac-derived noradrenaline and adrenaline levels, leads to further desensitization and subsequent down-regulation of β₁-ARs as well as a decrease in cardiac contractility, the hallmark of heart failure.⁷ In addition, it has been shown that β₁-AR down-regulation precedes clinical heart failure and may indeed be an early clinical marker of left ventricular dysfunction.⁸

To date, the human heart has been extensively studied with positron emission tomography (PET) and single photon emission computed tomography (SPECT) to assess myocardial blood flow and substrate metabolism.⁹ However, these imaging techniques have yet to realize the opportunity of noninvasive assessment of cardiac AR density, distribution, and occupancy using endogenous ligands or drugs,¹⁰ where the capacity of PET to be used for repeat measurements may significantly aid in monitoring and treatment of various heart diseases.¹¹ Identification of the population at risk for developing heart failure before clinical symptoms are noticed would greatly improve patient management as well as the stratification of a patient population that would benefit from placement of an implantable cardioverter-defibrillator (ICD).¹²

At present, however, no optimal radioligands for imaging cardiac β₁-AR using SPECT or PET are available,¹³ where the most extensively studied tracer in cardiovascular disease patients is the nonselective ligand [¹¹C]CGP-12177.¹⁴ Regrettably, its demonstrated capacity for visualization and quantification of β-AR density alone rather than the β₁-AR population specifically is less valuable for assessment of cardiac diseases.¹⁵

In contrast, the β₁-AR selective antagonist ICI 89,406 (1; Scheme 1) has previously been shown to produce effective blockage of β₁-AR during exercise in patients with angina pectoris.¹⁶ As such, Schaefer et al. chose this compound as their lead structure in the development of new β₁-AR selective ligands for PET and SPECT.¹⁷⁻¹⁹ Notably, the authors developed an O-methyl derivative of 1, labeled with carbon-11 ([¹¹C]OMe-ICI 89,406; [¹¹C]2), that exhibited high affinity and selectivity for β₁-AR in vitro. Unfortunately, in vivo studies did not demonstrate high specific binding to myocardial β₁-AR; rapid metabolism and high nonspecific binding in the myocardium were observed.²⁰

Scheme 1. Synthesis of ICI 89, 406, and Analogues.

Reagents and conditions: (a) HCl, dioxane, or TFA, CH₂Cl₂, 55–98%. (b) MeCN or i-PrOH, i-Pr₂NEt, Et₂BOMe, Yb(OTf)₃, 43–45%; method A, B, or C, 10–45%.

Our goal in the present study was to expand on the known structure–activity profile of 1 and evaluate multiple ¹⁸F-labeled derivatives through in vivo imaging studies.^21,22 Traditionally, ICI 89,406 along with a very small set of nonfluorinated derivatives were prepared via the acid-catalyzed epoxide opening of (S)-(oxiran-2-ylmethoxy)benzonitrile with an appropriately functionalized amine (Scheme 1).¹⁹ The various fluorinated β₁-ligands prepared in our study took advantage of this synthetic disconnection approach, where several functionalized amines were first prepared and then exposed to a series of substituted (S)-(oxiran-2-ylmethoxy)benzonitriles to create a diverse set of compounds with A and B regions as shown in Chart 1.

Chart 1. Diverse Moieties for the A and B Region Linked via the ICI 89,406 Backbone.

In conjunction, two complementary synthetic strategies were developed to generate the requisite functionalized amino-urea derivatives. Within each analogue, the central urea linkage was constructed through condensation of tert-butyl-2-aminoethylcarbamate with either a commercially available isocyanate or an in situ-generated p-nitrophenylcarbamate (Schemes 2 and 3 and Supporting Information).

Scheme 2.

Scheme 3. Synthesis of Carbamates 38, 40, 51–60, 63, and 65.

PG (protecting group) = Boc or Cbz. Reagents and conditions: (a) (i) 4-Nitrophenyl chloroformate, pyridine, solvent, overnight; (ii) NBoc-ethylenediamine, base, DMF or THF, 15 min −2 h, 29–94%. (b) H₂, Pd/C, EtOH, 90–96%. (c) 2-Fluoroethyl tosylate, K₂CO₃, DMF, 50–80 °C, 2–24 h, 42–96%.

In certain cases, additional functional group manipulations were required to complete the construct prior to reaction with the substituted (S)-(oxiran-2-ylmethoxy)benzonitriles, while in others, simple deprotection of the derived urea remained to generate the requisite coupling partner.²³

To generate multiple radioligands more efficiently, the radiosynthetic procedure that we chose for labeling compounds 5–7 utilized a well-established, two-step process that allows for labeling of multiple phenolic precursors with various ¹⁸F-labeled alkyl substituents (Scheme 4).²⁴ A robotic radiosynthesis station, which can conduct multistep radiosynthetic procedures, was developed and evaluated with our selected precursor phenols.

Scheme 4.

For preparation of the conjugate ¹⁸F-labeled alkylfluorides, precursor materials were required that contain activated leaving groups such as sulfonate esters (e.g., tosylates, mesylate). Primary derivatives were preferred because of their ease of displacement with fluorine-18 as compared to secondary derivatives, which often eliminate under the basic fluorination conditions. With these caveats in mind, we first generated the ¹⁸F-labeled radioligands [¹⁸F]fluoroethyl-(41) and [¹⁸F]fluoropropyl tosylate (62) from their respective bis-tosylate precursors using a combination of published methods.²⁵ Subsequent reaction of 41 and 62 with phenol precursors 47 or 49 under standard Williamson conditions afforded [¹⁸F]5–7 after high-performance liquid chromatography (HPLC) purification (11–31% conversion after 30 min; see the Supporting Information for details). Notably, for each radiotracer, ∼25 mCi of the final compound was isolated from ∼500 mCi of [¹⁸F]NaF in high radiochemical purity; the total synthesis time was 120 min. Specific activity values ranged from 750 to 2000 mCi/μmol depending upon the final radioactive concentration. Significantly, the single-step labeling process, ([¹⁸F]KF, K₂₂₂, MeCN, 90 °C, 15 min), using the derivative tosylate precursors of [¹⁸F]5–7, demonstrated rather poor chemical efficiency (<2% radiochemical yield).

Initial synthetic efforts were directed toward introduction of a fluorine atom into the B region of 1 (Table 1). Importantly, the aromatic fluorides 3 and 4 both exhibited potent β₁-AR binding activity (IC₅₀ = 0.2 and 0.1 nM, respectively) and β₁/β₂-AR selectivity (69- and 150-fold, respectively), revealing a modest preference for locating fluorine at the para-position of the aromatic ring. Alternatively, extension of the methoxy function in 2 to fluoroethoxy derivative 5 resulted in a 3-fold increase in selectivity while maintaining β₁-AR binding potency.²⁶ Interestingly, further extension to yield a fluoropropoxy function (6) resulted in decreased β₁-AR binding potency and subtype selectivity as compared to 2. Maintaining the fluoroethoxy group in 5 with transposition to the meta-position did, however, improve β₁-AR potency albeit with reduced selectivity (e.g., 5 and 7, Table 1). Relocating the fluorine atom from the B region (5) to the A region as in 12 induced a marked decrease in β₁-AR potency and selectivity. Lastly, when the fluoroethoxy group was replaced with a fluorinated amino derivative, β₁-AR potency and selectivity significantly decreased (8). Taken together, fluorine substitution was better tolerated on the B ring with regard to both β₁-AR potency and selectivity.

Table 1. Fluorinated Analogues of ICI 89,406 (1) and OMe-ICI 89,406 (2).

compound	β1a	β2/β1	cLogD
A1–B1 (2)	0.11	145	0.85
A1–B2 (3)	0.2	69	0.57
A1–B3 (4)	0.1	150	1.23
A1–B4 (5)	0.08	448	1.05
A1–B5 (6)	0.25	65	1.42
A1–B6(7)	0.04	265	1.03
A1–B7 (8)	1.5	13	0.39
A2–B8 (9)	21	164	–0.53
A3–B8 (10)	2	70	–1.87
A1–B9 (11)	1	511	–1.66
A2–B10 (12)	2	32	2.07
A1–B11 (13)	0.1	50	0.92
A1–B12 (14)	0.3	345	0.19
A1–B13 (15)	0.05	723	0.86
A1–B14(16)	0.05	>1000	0.47
A1–B15 (17)	0.35	163	0.35
A1–B16 (18)	26	8	1.04
A1–B17 (19)	10	10	–0.13
A1–B18 (20)	>100	0.05	–0.59
A1–B19 (21)	20	4	–0.74
A1–B20 (22)	10	4	0.92
A1–B21 (23)	3	30	0.42
A1–B22 (24)	0.14	6	1.28
A1–B23 (25)	5	0.1	0.91

^aIC₅₀ values are expressed in nanomolar concentrations.

Previous studies have identified potent β₁-AR selective ICI 89,406 analogues, which contain a carboxylic acid residue in the para-position of the B ring.²⁷ Hence, a set of fluorinated analogues (9–11), which contain a related substitution pattern, were prepared. Compounds 9 and 10 exhibit moderate binding potency to β₁-AR (IC₅₀ = 21 and 2 nM, respectively), while maintaining good β₁/β₂-AR selectivity (164- and 70-fold, respectively; Table 1). Transposing the fluorine atom from the A region to the B region (11) once again resulted in a significant increase in β₁-AR binding activity and selectivity as compared to 9 and 10. Although compounds 9–11 were comparable in their selectivity profile to 3–7, the β₁-AR binding potency was approximately 10-fold lower.

Despite numerous reports of β₁-AR ligands that contain an aryloxy propanolamine pharmacophore, many of these compounds are in fact not selective for β₁-AR.²⁸ Considering 5–7 contain both the aryloxy propanolamine backbone and a urea-substituted aromatic system (B region), one may conclude that the enhanced β₁-AR potency and selectivity is in fact attributed to these additional structural features. As such, supplemental structure–activity relationship (SAR) studies were conducted to evaluate structural requirements within the aryloxy propanolamine backbone of ligands 5–7 necessary to maintain potency and selectivity. For example, introduction of a heteroatom in the B ring of parent structures 4 and 5 afforded the fluorinated pyridine derivatives 14 and 13, respectively. Notably, 14 exhibited a 3-fold decrease in β₁-AR potency and a nearly 2-fold increase in β₁-AR selectivity as compared to 4, while compound 13 maintained β₁-AR potency, albeit with markedly decreased selectivity as compared to 5. Replacement of the fluorine atom in 14 with iodine or bromine yielded compounds 15 and 16, which exhibited increased β₁-AR potency and selectivity; 15 and 16 may prove interesting structural targets as ¹²⁴I and ⁷⁶Br-labeled PET radiotracers.

Introduction of an additional nitrogen atom in the B ring of 16 generated a fluorinated pyrimidine analogue, which resulted in a minor decrease in β₁-AR potency and selectivity (17; IC₅₀ = 0.35 nM; 163-fold), whereas replacement of the B ring with 5-membered heterocycles such as thiazole (18) or tetrazole (19) proved detrimental for both β₁-AR potency and selectivity. Not surprisingly, substitution of the B ring with saturated heterocycles as in 20 and 21 markedly decreased potency and selectivity and confirmed the aromatic structural requirements outlined above. A small structure–activity profile of the urea functionality within the linker connecting the A and B regions in 5–7 was also generated, where removal of either nitrogen atom within the urea functionality of 4 and 14 to afford the derived amides 24 and 23, respectively, resulted in >10-fold decrease in both potency and selectivity. Replacement of the urea moiety with a sulfonamide linker (25) also proved detrimental. Overall, no improved functional groups were identified.

In summary, compounds 3–7, 13, and 14, each containing a fluorine atom, exhibited β₁-AR binding (IC₅₀ = 0.04–0.30 nM) and good receptor selectivity (50–450-fold) suitable for imaging studies. Our criteria for radioligand selection were the following: (a) β₁-binding affinity, <1 nM; (b) β₁/β₂ selectivity >50; (c) moderate cLogD values; and (d) readily labeled with fluorine-18. Compounds 5–7 were thus selected as representative candidates for labeling with fluorine-18 as they achieved our selection criteria, provided a range of activity and selectivity values, and maintained a cLogD value near one.

Biodistribution studies of [¹⁸F]5–7 in Sprague–Dawley (SD) rats are summarized in Table 2. In general, high extraction of the radiotracers from the blood pool into tissues such as heart, lung, and liver was observed at 30 min postinjection. Heart uptake of compounds [¹⁸F]5–7 was 0.48, 0.42, and 0.62% ID/g, at 30 min, with corresponding heart:lung ratios of 1:2, 1:1.5, and 1:1.27, respectively. Similar values were also observed between the heart and the liver. In these cases, uptake of radioligands [¹⁸F]5–7 appears unaffected by the position or length of the fluoroalkyl side chain.

Table 2. Biodistribution of [¹⁸F]-ICI 89,406 Analogues in Rats^ab.

organ	[18F]5b	[18F]6a	[18F]7b
blood	0.18 ± 0.07	0.21 ± 0.01	0.14 ± 0.05
heart	0.48 ± 0.16	0.42 ± 0.15	0.62 ± 0.03
lung	0.97 ± 0.26	0.63 ± 0.27	0.79 ± 0.11
liver	1.14 ± 0.33	1.03 ± 0.52	1.14 ± 0.12
spleen	0.37 ± 0.02	0.40 ± 15	0.52 ± 0.09
kidney	1.26 ± 0.29	1.38 ± 0.16	2.01 ± 0.21
femur	0.27 ± 0.05	0.63 ± 0.16	0.16 ± 0.01
muscle	0.08 ± 0.02	0.09 ± 0.01	0.10 ± 0.01

^aData are expressed as the % ID/g ± SD with three animals per data point at 30 min postinjection.

^bData are expressed as the % ID/g ± SEM with two animals per data point at 30 min postinjection.

Considering low expression of the β₁ receptor subtype, the observed preferential uptake of these tracers in lung tissue is likely nonspecific in nature. While this detail is consistent with the published literature,²⁰ several authors report challenges associated with the in vivo demonstration of subtype selectivity, including achieving a proper balance of specific activity and receptor occupancy as well as mediating the pharmacologic effects of preferred blocking agents (i.e., bradycardia from metaprolol). Despite achieving appropriate levels of specific activity in the present study, the observed tissue distribution will likely complicate future in vivo blocking studies.

Despite low statistical power, the abbreviated in vivo dissection studies outlined above help to establish gross differences in compound distribution and facilitate SAR development. More extensive, statistically relevant studies are however required, to distinguish subtle distribution differences and finalize lead candidate selection.

Dynamic PET imaging studies carried out on SD rats supported the biodistribution results. All three radiotracers were clearly visualized in both the myocardium and the liver at 30 min postinjection; however, at 60 min, significant wash-out of all three radiotracers was observed. Importantly, no interference from uptake in the lung was detected (Figure 1). In vivo metabolism study of [¹⁸F]7 revealed progressive loss of the parent species over time beginning at 83% and ending at 44% (2 and 30 min, respectively); a 30 min urine sample contained 76% of the parent species. Two unique metabolites were observed by HPLC analysis.

Figure 1.

Images of the rat heart at 30 min. (a) Horizontal long axis, (b) short axis, and (c) vertical long axis.

Considering the proximity of data derived from [¹⁸F]5–7, compound [¹⁸F]7 was chosen as a representative example for more detailed in vivo evaluation. Time–activity plots generated through further study of [¹⁸F]7 showed rapid uptake of the radiotracer from the blood pool into the myocardium, lung, and liver <5 min postinjection, after which a steady level of the tracer was observed in blood, heart, and lung tissues as liver levels decreased markedly over time (Figure 2). Importantly, the time-dependent imaging study demonstrated [¹⁸F]7 does in fact accumulate in cardiac tissue within minutes after injection. Target:nontarget organ ratios remained relatively stable over time (Figure 3; Supporting Information).

Figure 2.

Time–activity curves from the imaging data of [¹⁸F]7.

Figure 3.

Images of [¹⁸F]7 in the rat heart at various time points. (a) Horizontal long axis, (b) short axis, and (c) vertical long axis.

In conclusion, multiple ¹⁸F-labeled derivatives of the β₁-selective ligand ICI 89,406 were developed and evaluated as potential cardiac imaging tracers. An extensive structure–activity study identified a series of fluorine-substituted β₁-selective ligands, with potent β₁-AR affinity and β₁/β₂ selectivity. In vivo evaluation of [¹⁸F]5–7 in SD rats demonstrated uptake of the radiotracers in heart (0.48–0.62% ID/g), lung (0.63–0.97% ID/g), and liver (1.03–1.14% ID/g) tissue; results were confirmed through μPET imaging study.

Despite clear visualization of the myocardium, the considerable uptake of these tracers in nontarget tissues intimates a lack of specific binding, which may limit overall application of current imaging data and general utility of existing pharmacophore design. Future studies will attempt to reconcile these issues of selectivity and be directed toward decreasing liver uptake and improving overall cardiac retention.

Supporting Information Available

Full experimental procedures and characterization data for all new compounds described in this study. This material is available free of charge via the Internet at http://pubs.acs.org.