Non-Invasive Assessment of Locally Overexpressed Human Adenosine 2A Receptors in the Heart of Transgenic Mice

A2A adenosine receptors (A2A-AR) have a cardio-protective function upon ischemia and reperfusion, but on the other hand, their stimulation could lead to arrhythmias. Our aim was to investigate the potential use of the PET radiotracer [18F]FLUDA to non-invasively determine the A2A-AR availability for diagnosis of the A2AR status. Therefore, we compared mice with cardiomyocyte-specific overexpression of the human A2A-AR (A2A-AR TG) with the respective wild type (WT). We determined: (1) the functional impact of the selective A2AR ligand FLUDA on the contractile function of atrial mouse samples, (2) the binding parameters (Bmax and KD) of [18F]FLUDA on mouse and human atrial tissue samples by autoradiographic studies, and (3) investigated the in vivo uptake of the radiotracer by dynamic PET imaging in A2A-AR TG and WT. After A2A-AR stimulation by the A2A-AR agonist CGS 21680 in isolated atrial preparations, antagonistic effects of FLUDA were found in A2A-AR-TG animals but not in WT. Radiolabelled [18F]FLUDA exhibited a KD of 5.9 ± 1.6 nM and a Bmax of 455 ± 78 fmol/mg protein in cardiac samples of A2A-AR TG, whereas in WT, as well as in human atrial preparations, only low specific binding was found. Dynamic PET studies revealed a significantly higher initial uptake of [18F]FLUDA into the myocardium of A2A-AR TG compared to WT. The hA2A-AR-specific binding of [18F]FLUDA in vivo was verified by pre-administration of the highly affine A2AAR-specific antagonist istradefylline. Conclusion: [18F]FLUDA is a promising PET probe for the non-invasive assessment of the A2A-AR as a marker for pathologies linked to an increased A2A-AR density in the heart, as shown in patients with heart failure.


Introduction
Multiple effects of adenosine in humans and animals have been described for many years. In the heart, adenosine, potentially released from myocardial ATP, reduces the heart With that regard, the aim of this study was to evaluate the potential of [ 18 F]FLUDA for A2A-AR imaging by PET in a mouse model with a functional myocardial overexpression of the human A2A-AR [21].

Impact of FLUDA on the Atrial Force of Contraction (FOC) in Electrically Stimulated Atrial Preparations
The yet unknown agonistic or antagonistic action of FLUDA towards the human A2A-AR was investigated in electrically stimulated atrial preparations of wild-type mice (WT) and hA2A-AR TG ( Figure 2). As shown in Figure 2B, the FOC of WT was not changed after adding 10 µM of the A2A-AR agonist CGS 21680 (positive inotropic effect) to the organ bath. An anti-inotropic effect towards the human A2A-AR was observed after the subsequent adding of 1 µM FLUDA. The FOC was decreased by 15.0 ± 2.4% in the A2A-AR TG atria. Hence, these results revealed an antagonistic effect of FLUDA towards a stimulated A2A-AR.

Impact of FLUDA on the Atrial Force of Contraction (FOC) in Electrically Stimulated Atrial Preparations
The yet unknown agonistic or antagonistic action of FLUDA towards the human A 2A -AR was investigated in electrically stimulated atrial preparations of wild-type mice (WT) and hA 2A -AR TG ( Figure 2). As shown in Figure 2B, the FOC of WT was not changed after adding 10 µM of the A 2A -AR agonist CGS 21680 (positive inotropic effect) to the organ bath. An anti-inotropic effect towards the human A 2A -AR was observed after the subsequent adding of 1 µM FLUDA. The FOC was decreased by 15.0 ± 2.4% in the A 2A -AR TG atria. Hence, these results revealed an antagonistic effect of FLUDA towards a stimulated A 2A -AR. With that regard, the aim of this study was to evaluate the potential of [ 18 F]FLUDA for A2A-AR imaging by PET in a mouse model with a functional myocardial overexpression of the human A2A-AR [21].

Impact of FLUDA on the Atrial Force of Contraction (FOC) in Electrically Stimulated Atrial Preparations
The yet unknown agonistic or antagonistic action of FLUDA towards the human A2A-AR was investigated in electrically stimulated atrial preparations of wild-type mice (WT) and hA2A-AR TG ( Figure 2). As shown in Figure 2B, the FOC of WT was not changed after adding 10 µM of the A2A-AR agonist CGS 21680 (positive inotropic effect) to the organ bath. An anti-inotropic effect towards the human A2A-AR was observed after the subsequent adding of 1 µM FLUDA. The FOC was decreased by 15.0 ± 2.4% in the A2A-AR TG atria. Hence, these results revealed an antagonistic effect of FLUDA towards a stimulated A2A-AR.

In Vitro Binding of [ 18 F]FLUDA to the A 2A -AR in Heart Samples
We performed competition assays to determine the A 2A -AR specificity of [ 18 F]FLUDA towards the heart tissue of WT and A 2A -AR TG ( Figure 3). In cardiac cryosections of WT, non-homologous competition with ZM 241385 revealed a specific A 2A -AR binding of [ 18 F]FLUDA of 24.6 ± 9.6% ( Figure 3A). However, the low signal of total binding, as well as the homologous competition with FLUDA, suggests a very low A 2A -AR density, preventing the determination of the endogenous A 2A -AR receptor density B max . In cardiac cryosections of A 2A -AR TG, non-homologous competition with ZM241385 revealed a specific binding of 69.0 ± 6.6% ( Figure 3A). A B max of 455 ± 78 fmol/g wet weight and a K D of 5.9 ± 1.6 nM was determined ( Figure 3B). Notably, the specific binding of [ 18 F]FLUDA in muscle and lung tissues were comparable in WT and A 2A -AR TG ( Figure S1). adenosine by 1 µg/mL adenosine deaminase (ADA). Induction of atrial contraction was achieved by the A2A-AR agonist CGS 21680 (10 µM). (B) Quantification of left atrial force. Data are means ± SEM; numbers in columns are numbers of atrial preparations and + marks the added substance; * p < 0.05 vs. WT; + p < 0.05 vs. Ctr; # p < 0.05 vs. CGS (ANOVA).

In Vitro Binding of [ 18 F]FLUDA to the A2A-AR in Heart Samples
We performed competition assays to determine the A2A-AR specificity of [ 18 F]FLUDA towards the heart tissue of WT and A2A-AR TG ( Figure 3). In cardiac cryosections of WT, non-homologous competition with ZM 241385 revealed a specific A2A-AR binding of [ 18 F]FLUDA of 24.6 ± 9.6% ( Figure 3A). However, the low signal of total binding, as well as the homologous competition with FLUDA, suggests a very low A2A-AR density, preventing the determination of the endogenous A2A-AR receptor density Bmax. In cardiac cryosections of A2A-AR TG, non-homologous competition with ZM241385 revealed a specific binding of 69.0 ± 6.6% ( Figure 3A). A Bmax of 455 ± 78 fmol/g wet weight and a KD of 5.9 ± 1.6 nM was determined ( Figure 3B). Notably, the specific binding of [ 18 F]FLUDA in muscle and lung tissues were comparable in WT and A2A-AR TG ( Figure S1).

In Vitro Binding Studies of [ 18 F]FLUDA in Human Atrial Samples
In a preliminary study, we used cryosections of human atrial samples of subjects without any diagnosed heart failure for autoradiography with [ 18 F]FLUDA and determined an A 2A -AR-specific binding of 36.3 ± 5.3% (n = 2) (Figure 4), which is comparable to the values determined in the WT group of mice, representing a low A 2A -AR availability in the healthy heart tissue.
In a preliminary study, we used cryosections of human atrial samples of subjects without any diagnosed heart failure for autoradiography with [ 18 F]FLUDA and determined an A2A-AR-specific binding of 36.3 ± 5.3% (n = 2) (Figure 4), which is comparable to the values determined in the WT group of mice, representing a low A2A-AR availability in the healthy heart tissue.

In Vitro Binding Studies of [ 18 F]FLUDA in Human Atrial Samples
In a preliminary study, we used cryosections of human atrial samples of subjects without any diagnosed heart failure for autoradiography with [ 18 F]FLUDA and determined an A2A-AR-specific binding of 36.3 ± 5.3% (n = 2) (Figure 4), which is comparable to the values determined in the WT group of mice, representing a low A2A-AR availability in the healthy heart tissue.   Under baseline conditions, we observed an increased initial uptake (1 to 10 min p.i.) of [ 18 F]FLUDA in the myocardium of the A 2A -AR TG compared to WT ( Figure 5). The analysis of the time-activity curves (TACs) ( Figure 5, Table 1) revealed an earlier timeto-peak value in the blood compartment (−0.2 min, p = 0.011) and a higher TAC peak value in the myocardium of A 2A -AR TG (+10.4%) compared to WT. Hence, the integrated activity concentration over time was higher in both compartments. In the initial phase, a 2.3 times higher AUC value (AUC 1-10 , p < 0.001) was found in the myocardium, while it was 1.6 times higher (AUC 1-10 , p = 0.004) in the blood, as a result of the functional overexpression of the hA 2A -AR. To normalise for unspecific physiological effects, AUC 1-10 ratios (SUVr) of the myocardium to blood ( Figure 6C, Table 3) and myocardium to muscle (Table 3)    For blocking studies, the A 2A -AR-specific inhibitor istradefylline (Figure 1) was injected 5 min before the radiotracer to prove the A 2A -AR-specific uptake of [ 18 F]FLUDA into the myocardium (Figure 6, Tables 2 and 3). In WT, the time to peak in the blood compartment of the left ventricle was earlier compared to the baseline conditions (p = 0.017), whereas the TAC peak value in the muscle was slightly increased (SUV of 0.5 ± 0.1 vs. 0.8 ± 0.1, p = 0.027) compared to baseline WT. However, we found no significant changes in the AUC 1-10 and AUC 0-60 for these and other investigated tissues in comparison to the untreated WT. In A 2A -AR TG, pre-injection of istradefylline abolished the increased uptake of [ 18 F]FLUDA into the myocardium, as shown by the significant reduction of the AUC 1-10 , about 0.6 times (p = 0.032 vs. WT). These results were validated by the normalisation of the AUC 1-10 of the tissues to the blood compartment, as well as to muscle. It was apparent that tissue normalisation to the blood compartment was less prone to fluctuation compared to the muscle, as shown by the SEM values (Table 3); thus, the blood compartment seems more suitable as a reference tissue. Additionally, in the more sensitive ex vivo autoradiography studies, the increased uptake of [ 18 F]FLUDA in the murine hearts of A 2A -AR TG was found even fifteen minutes post-injection and was blocked by pre-administration of istradefylline ( Figure S3).     Hence, the high specific binding of [ 18 F]FLUDA towards the overexpressed hA 2A -AR in the myocardium of mice, as shown by the in vitro autoradiography studies, could be validated in vivo by PET imaging.

Discussion
The present work demonstrates the usefulness of the new radiotracer [ 18 F]FLUDA for specific A 2A -AR imaging by PET in cardiac tissue. We present evidence that (1)  Ishiwata and colleagues provided evidence of the successful non-invasive assessment of A 2A -AR in a human subject by PET imaging using the methylxanthine derivative of KF17837 [ 11 C]TMSX [20]. [ 11 C]TMSX was later used in subsequent studies to compare the A 2A -AR density in hearts between endurance athletes and untrained men [22,23]. The non-xanthine derivative [ 18 F]FESCH was one of the first 18 F-labelled A 2A -AR PET-imaging probes [24,25], and its deuterated isotopologue [ 18 F]FLUDA was recently developed and evaluated in vitro as well as in vivo by our group [18]. In the present study, we could show a reduction of the FOC in atrial preparations (stimulated with CGS 21680) from transgenic, but not from wild type mice, after adding FLUDA ( Figure 2). Hence, we could confirm the antagonistic action of FLUDA towards the A 2A -AR by a functional assay.
In a former study with [ 18 F]FLUDA, we determined a K D value of 4.30 ± 0.73 nM and a B max value of 556 ± 143 fmol/mg wet weight in the striatum of healthy CD-1 mice and a K D value of 0.68 nM, as well as a B max value of 218 fmol/mg wet weight in the striatum of piglets [18], a brain region with a high A 2A -AR density. In the human heart, the A 2A -AR is localised at the level of the Z-line of atrial myocytes, where it is co-expressed with α-actinin and the ryanodine receptor [26]. In the present study, we determined comparable A 2A -AR binding kinetics of [ 18 F]FLUDA in cardiac cryosections of A 2A -AR TG (K D of 5.9 ± 1.6 nM and a B max of 455 ± 78 fmol/mg protein) as in the striatum of healthy male CD-1 mice, although reliable binding kinetic parameters for cardiac cryosections of female FVB/N mice (WT) could not be determined (Figure 3). We assume that the A 2A -AR density in the cardiac tissue of WT is low and at the edge of the detection limit of [ 18 F]FLUDA, confirming the findings of the functional assays. Binding kinetic studies from other groups with [ 3 H]ZM241385 revealed a higher A 2A -AR density in human heart membrane preparations from patients with chronic heart failure (NYHA functional class III and IV, B max = 210 ± 8 fmol/mg protein), which was accompanied with a decreased ligand binding affinity (K D = 2.4 ± 0.1 nmol/L) when compared to control (B max = 135 ± 5 fmol/mg protein, K D = 0.9 ± 0.0 nmol/L) [27]. As an initial step in the present study, we determined a low A 2A -AR-specific binding with [ 18 F]FLUDA in the human atrial samples of patients with no diagnosed heart failure (Figure 4), which is in accordance with the finding in binding studies in cardiac cryosections of WT mice in this study. Hence, we would also expect a good signal-to-background ratio in humans with an increased A 2A -AR receptor density in the heart. A direct comparison to heart samples of patients with heart failure and atrial fibrillation should be performed in future studies.
A 2A -AR agonists are often used in the clinic to dilate the coronary arteries in patients and to assess the severity and functional consequences of impaired vasodilation in angina pectoris [28]. However, [ 18 F]FLUDA is not expected to be useful to image the coronary A 2A -AR density, specifically, as the amount of smooth muscle cells and endothelial cells is much lower compared to the number and volume of cardiomyocytes. Thus, we would expect a binding of [ 18 F]FLUDA mainly to the A 2A -AR in cardiomyocytes. The in vivo studies in mice revealed increased uptake of [ 18 F]FLUDA under baseline conditions into the myocardium of A 2A -AR TG compared to WT. However, this was restricted to the initial phase after administration of the radioligand. In a single ex vivo autoradiography, we could confirm increased A 2A -AR-specific uptake into the myocardia of A 2A -AR TG even after 15 min p.i. (Figure S3). An accumulation over time of [ 18 F]FLUDA into the cardiac tissue of mice could not be observed as it was shown in other studies for [ 11 C]TMSX [20]. Interestingly, in that study, the cardiac uptake of [ 11 C]TMSX was displaceable by just 40% with carrier and 8-(3-chlorostyryl)caffeine (each in A 2A -AR-saturable concentrations), which may be caused by a higher amount of unspecific binding of this tracer, as it was shown for the binding in brain regions with low A 2A -AR expression [29,30]. Additionally, it was shown that 60 min post-injection [ 11 C]TMSX was very stable in human plasma (> 90% intact radioligand), whereas in mice, only 54% in plasma and 76% in heart tissue were measured [20]. In a recent study, we found that 71% of [ 18 F]FLUDA was non-metabolised in plasma samples of healthy mice 15 min post-injection [18]. Hence, further studies are needed to clarify a potential enrichment of [ 18 F]FLUDA-derived metabolites in the cardiac tissue over time.
Clinical outlook: [ 18 F]FLUDA PET imaging could be useful to assess the receptor occupancy in the heart by A 2A -AR-targeting drugs. This might be useful when treating Parkinsonian patients with A 2A antagonists to avoid adverse side effects. Another potential application could be the non-invasive determination of an elevated A 2A AR density in patients with fibrillation as a diagnostic marker. Hence, it would be possible to establish a cause-effect relationship between A 2A -AR density and atrial fibrillation in a non-invasive manner in long-term follow-up. The next step will be to test the eligibility of [ 18 F]FLUDA in patients and to determine the uptake of [ 18 F]FLUDA in the heart.
In summary, we describe a novel radioligand to label and quantify A 2A receptors for diagnostic purposes in the living mammalian heart.

General Information
All chemicals and reagents were purchased from commercially available sources and used without further purification.

Animals
For the present study, female FVB/N mice (A 2A -AR TG and WT, age: 4-6 months, weight 27 ± 2 g) were used. The generation of the transgenic FVB/N mice overexpressing the human A 2A -AR under control of an alpha myosin heavy chain promoter in cardiac tissue (A 2A -AR TG), were described elsewhere [21]. In the present study, female A 2A -AR TG and wild type (WT) mice were used in the indicated number. The investigation conforms to the Guide for the Care and Use of Laboratory Animals published by the National Research Council (2011). Animals were handled and maintained according to approved protocols of the animal welfare committee of the University of Münster, Germany. All procedures performed in studies involving animals were in accordance with the ethical standards of the institution or practice at which the studies were conducted (Landesdirektion Sachsen, TVV 18/18).

Human Atrial Preparations
Right atrium samples were obtained from patients undergoing open-heart surgery with coronary artery bypass grafts and electrically stimulated in organ baths, as described previously [31,32]. This study complied with the Declaration of Helsinki and was approved by the local ethics committee (hm-bü04.08.2005). All patients gave informed consent.
Ex vivo autoradiography of the murine hearts was performed immediately after euthanising the animals at 15 min p.i. of the radioligand. The hearts were isolated, frozen by immersion in isopentane at −20 • C, cryosectioned (16 µm; MICROM HM560), and the dried sections were exposed to a phosphor imaging plate for 120 min and processed as described above.

Small-Animal PET/MR Experiments
The animals were initially anaesthetised with 5% isoflurane and were positioned prone into a small-animal PET/MR (nanoScan, MEDISO, Budapest, Hungary) on a temperaturecontrolled bed system (37 • C) while respiration rate was continuously monitored. The anaesthesia (Anaesthesia Unit U-410, agntho's, Lidingö, Sweden) was maintained at 2.1-1.3% isoflurane in a 60% oxygen/40% air gas mixture (Gas blender 100 series, MCQ Instruments, Rome, Italy) with 250 mL/min airflow. Prior to the 60 min PET scan, a scout image MR sequence was performed to outline the animal dimensions. Animals (A 2A -TG and WT) received an i.v. injection of [ 18 F]FLUDA (3.5 to 11.3 MBq; 1.5 to 11.3 fmol/g bodyweight). Control animals (WT: n = 6; A 2A -TG: n = 8) received a vehicle solution containing DMSO/Kolliphor/NaCl, 1:2:7 (v/v/v) 10 min prior to [ 18 F]FLUDA i.v. injection. For the determination of the A 2A AR-specific uptake, animals received an i.v. injection of 2.5 mg/kg bodyweight tozadenant (WT and A 2A -TG: n = 4) or 1mg/kg istradyfelline (WT and A 2A -TG: n = 3) 10 min prior the radiotracer administration. The data were collected in list mode (11 × 10; 1 × 20; 5 × 30; 1 × 45; 4 × 60; 1 × 180; 6 × 300; 2 × 900 s). Subsequently, after the PET scan, a T1-weighted whole-body MR scan (gradient echo sequence, TR = 20 ms, TE = 3.2 ms) was performed for anatomical orientation and attenuation correction at the reconstruction step (3D-OSEM, 4 iterations, 6 subsets; MR-based attenuation correction). The reconstructed images were analysed with PMOD (Version 3.802). For analysing the [ 18 F]FLUDA uptake into the myocardium and the activity concentration in the left ventricle (blood compartment), VOIs were delineated in an averaged image of the first 10 min of the PET imaging after injection of [ 18 F]FLUDA, and co-registered T1 images from MR. Non-parametrical analyses of achieved time-activity curves (TACs) were performed with Microsoft Excel to determine the time to peak, the TAC peak value, and the area-under-the-curve (AUC): where c (radioactivity) is expressed as a standardised uptake value normalised to the bodyweight in g (SUV). GraphPad Prism 9 (GraphPad Inc.; La Jolla, CA, USA) was used for graphical presentation.

Statistical Analysis
Values are represented as mean ± standard error of the mean (SEM). Statistical analyses were performed with Microsoft Excel and GraphPad Prism (v9, San Diego, CA, USA), with p-values ≤ 0.05, calculated by variance analysis (ANOVA) with Bonferroni's post-hoc test or Student's t-test if applicable, were considered as significant.