Evaluating Hepatobiliary Transport with 18F-Labeled Bile Acids: The Effect of Radiolabel Position and Bile Acid Structure on Radiosynthesis and In Vitro and In Vivo Performance

Introduction An in vivo determination of bile acid hepatobiliary transport efficiency can be of use in liver disease and preclinical drug development. Given the increased interest in bile acid Positron Emission Tomography- (PET-) imaging, a further understanding of the impact of 18-fluorine substitution on bile acid handling in vitro and in vivo can be of significance. Methods A number of bile acid analogues were conceived for nucleophilic substitution with [18F]fluoride: cholic acid analogues of which the 3-, 7-, or 12-OH function is substituted with a fluorine atom (3α-[18F]FCA; 7β-[18F]FCA; 12β-[18F]FCA); a glycocholic and chenodeoxycholic acid analogue, substituted on the 3-position (3β-[18F]FGCA and 3β-[18F]FCDCA, resp.). Uptake by the bile acid transporters NTCP and OATP1B1 was evaluated with competition assays in transfected CHO and HEK cell lines and efflux by BSEP in membrane vesicles. PET-scans with the tracers were performed in wild-type mice (n = 3 per group): hepatobiliary transport was monitored and compared to a reference tracer, namely, 3β-[18F]FCA. Results Compounds 3α-[18F]FCA, 3β-[18F]FGCA, and 3β-[18F]FCDCA were synthesized in moderate radiochemical yields (4–10% n.d.c.) and high radiochemical purity (>99%); 7β-[18F]FCA and 12β-[18F]FCA could not be synthesized and included further in this study. In vitro evaluation showed that 3α-FCA, 3β-FGCA, and 3β-FCDCA all had a low micromolar Ki-value for NTCP, OATP1B1, and BSEP. In vivo, 3α-[18F]FCA, 3β-[18F]FGCA, and 3β-[18F]FCDCA displayed hepatobiliary transport with varying efficiency. A slight yet significant difference in uptake and efflux rate was noticed between the 3α-[18F]FCA and 3β-[18F]FCA epimers. Conjugation of 3β-[18F]FCA with glycine had no significant effect in vivo. Compound 3β-[18F]FCDCA showed a significantly slower hepatic uptake and efflux towards gallbladder and intestines. Conclusion A set of 18F labeled bile acids was synthesized that are substrates of the bile acid transporters in vitro and in vivo and can serve as PET-biomarkers for hepatobiliary transport of bile acids.


Introduction
Bile acids are steroid derivatives that are produced by the hepatocytes of the liver and excreted in bile. These molecules play an important role in micelle formation for lipid digestion and uptake of fat-soluble vitamins in the intestines [1]. The majority of bile acids (95%) is then reabsorbed and transported back to the liver by the portal circulation.
Bile acid homeostasis in the liver is maintained by specific bile acid transporters on the hepatocytes. Uptake at the basolateral side relies predominantly on the Na + -dependent Taurocholate Cotransporting Polypeptide (NTCP) and also the Organic Anion Transporting Polypeptides (OATPs). Once in the hepatocyte, excretion in the bile canaliculi is mainly mediated by the Bile Salt Export Pump (BSEP) [2].

Contrast Media & Molecular Imaging
However, this highly efficient hepatobiliary transport of bile acids can be disturbed by xenobiotics that inhibit the aforementioned bile acid transporters or in certain liver diseases such as primary biliary cirrhosis or progressive familial intrahepatic cholestasis (PFIC) [3][4][5]. A toxic buildup of bile acids in the hepatocytes, termed cholestasis, can then present itself. Clinical features may consist of nausea, abdominal pain, jaundice and pruritus [6]. An accurate in vivo determination of bile acid transport efficiency can therefore be valuable for detection or evaluation of liver disease in both drug development and clinic.
To visualize physiological processes on a molecular level in vivo, Positron Emission Tomography (PET) is the imaging modality of choice. The possibility for close resemblance of the PET-radiotracer and the endogenous substrate under investigation gives this imaging technique an important asset [7]. Recently, a number of studies with 11 C or 18 F labeled bile acid PET-tracers have been published. Frisch et al. evaluated 11 C labeled bile acids, conjugated with sarcosine or N-methyl taurine in pigs [8,9]. These tracers were able to visualize hepatobiliary transport and can provide valuable information in hepatobiliary diseases, although the short half-life of the 11 C isotope limits their use. Several 18 F labeled bile acids were developed to overcome this issue. Jia et al. developed a 18 F labeled bile acid for studying Farnesoid X Receptor-(FXR-) related diseases, using a click reaction of 1,3-dipolar cycloaddition of terminal alkynes and organic azides [10]. However, this modification removes the terminal carboxylic acid moiety that is critical for recognition by the bile acid transporters [11]. Testa et al. improved this click chemistry approach by retaining the carboxylic acid functional group [12]. Nevertheless, for both 18 F labeled compounds the introduction of a long 1,2,3-triazole linked fluoroalkyl sidechain on the carboxylic acid terminus encompasses a substantial addition to the steroid structure. This might result in altered hepatobiliary transport, compared to endogenous bile acids.
Because of the need for a 18 F labeled bile acid with only a minimal modification in molecular structure compared to an endogenous bile acid, 3 -[ 18 F]fluorocholic acid (3 -[ 18 F]FCA) was developed by our research group [13]. The bile acid structure for 3 -[ 18 F]FCA is cholic acid, of which the 3 alpha OH group was substituted for a 3-beta fluorine. This tracer showed transport by NTCP, OATP, and BSEP in vitro and could visualize hepatobiliary transport in vivo and druginduced alterations thereof.
Given the increased interest in bile acid PET-imaging, a further understanding of the impact of fluorine-OH substitution on bile acid handling in vitro and in vivo can be beneficial. Therefore, in this study a number of fluorinated analogues of cholic, chenodeoxycholic, and glycocholic acid were synthesized, evaluated, and compared in vitro and in vivo in mice. 18 F-Labeled Bile Acids. The 18 Fisotope was introduced on the bile acid skeleton by a nucleophilic substitution on a suitable mesylate, protected precursor molecule of which the synthesis can be found in Supplementary Data (available ). The conceived 18 F labeled bile acids are shown in Figure 1. The radiosynthesis was performed as described earlier [13]. In short, [ 18 F]fluoride (1.3 GBq) was trapped on a Sep-Pak QMA-column (Waters, Zellik, Belgium) that was preconditioned with 5 mL 0.01 M K 2 CO 3 and 5 mL ultrapure water. The activity was eluted in a radiosynthesis vial with 1 mL 9 : 1 AcN : H 2 O Cryptand-2.2.2 (Acros Organics, Geel, Belgium) and K 2 CO 3 (20 mg and 2 mg, resp.) solution. The solvents were removed by evaporation under a gentle nitrogen flow at 100 ∘ C. The residue was dried further by adding and evaporating 2 × 500 L AcN.

Radiosynthesis of the
The precursor for radiosynthesis was dissolved in 200 L anhydrous DMSO and added to the radiosynthesis vial. A fixed amount (6.84 mol; 3.4-4.4 mg) of precursor for 3 - Afterwards, the vial was cooled and 100 L 5 M NaOH was added. The mixture was shaken and heated again at 120 ∘ C for 10 minutes. After cooling and neutralization of the basic reaction mixture, purification was performed by a semipreparative HPLC-system (Grace Econosphere C18 10.0 × 250 mm, 10 m; 6 mL/min 10% AcN in H 2 O -> 100% AcN in 20 minutes; radiodetection (Ludlum Measurements Inc.)). The desired HPLC-fraction was collected, diluted with ultrapure water to 50 mL, and loaded on a Sep-Pak C18 Plus short cartridge (preconditioned with 10 mL EtOH and 10 mL ultrapure water). After washing with 10 mL ultrapure water, the 18 F-labeled bile acid was eluted off the column with 1 mL EtOH in a separate vial. This fraction was evaporated under a gentle N 2 -flow and heating. Finally, 500 L phosphate buffered saline was added to reformulate the tracer for in vivo use.
The logD-value, stability (in formulation for use and mouse serum), and identity of the tracer were assessed as described earlier [13]. Radiochemical purity and percentage 18 F-labeling of precursor were determined by radio-TLC (silica gel TLC; with 10% MeOH in CH 2 Cl 2 and 4 : 1 AcN : H 2 O, resp.).

In Vitro
where refers to inhibition constant; IC50 refers to concentration of fluorinated bile acid that causes a 50% decrease in All dosing solutions were formulated in the washing buffer and dosing was executed in triplicate. The incubation was stopped by placing the plates on ice and adding 1 mL ice-cold 1% BSA HBSS-solution. The supernatant was aspirated and the cells were washed twice with 2 mL ice-cold HBSS-solution. NaOH (250 l, 0.1 M) was added to lyse the cells and the plates were shaken (15 minutes, 37 ∘ C). Aliquots of this lysate were subjected to liquid scintillation counting (TriCarb 2900 TR; Perkin Elmer) and protein determination with a BCA-assay (ThermoFisher Scientific).

In Vitro Efflux
Assays. Transport of the bile acids by BSEP was evaluated by their inhibition of [ 3 H]TC uptake in BSEP membrane vesicles (Pharmtox). Cholic acid (CA) was added to the test compound panel as a reference. Ki-values of the different bile acids were calculated using the experimentally determined IC50-and Km-parameters in the Cheng-Prusoff equation (GraphPad Prism v3.00 Software) (see (1)). The assay was performed in V-tip 96-well plates; each well con-  Food and water were provided ad libitum, but the animals were fasted overnight before a PET/CT-scan. They were anesthetized with 1.5 v : v% isoflurane in O 2 and placed on a heated bed. To allow injection of the tracer, a polyethylene intravenous line was inserted in the lateral tail vein and fixed. After the animals were transferred to the scanner animal bed, a 1 hour PET-scan was started and 9 MBq tracer was injected directly after starting the scan. Following this PETscan, 9 MBq [ 18 F]FDG was injected and twenty minutes later, a second PET-scan with a 20-minute acquisition time was started.
All PET-scans were obtained in list-mode and were iteratively reconstructed (50 iterations). The 1-hour scan was reconstructed in the following frames: 8 × 15 s; 16 × 30 s; 10 × 60 s; 20 × 120 s. For presentation purposes, Maximum Intensity Projection PET/CT images were generated in Amide software. The data were analyzed using Pmod software v3.405 (PMOD Technologies): Regions Of Interest (ROIs) were drawn manually over the liver, gallbladder, and intestines. On the static [ 18 F]FDG scan, the left ventricle was delineated and this ROI was pasted on the dynamic scan to obtain an image-derived arterial blood concentration. The uptake of radioactivity in liver, gallbladder, and intestines was expressed as a percentage of the injected dose (% ID) and normalized for the weight of a 20 g mouse. The % ID in these organs was monitored in function of time to obtain timeactivity curves (TACs). Biliary clearance of the tracers was determined with equation 2 from Ghibellini et al. [14] Biliary clearance = cumulative amount of tracer in gallbladder&intestines AUC blood concentration 0→60 min .
The Area Under the Curve (AUC), % ID, and time-topeak values of the TACs were determined in GraphPad Prism v3.00 Software. The obtained parameters of the different 18 F labeled bile acids were compared to 3 -[ 18 F]FCA-values in SPSS Statistics 23 Software. Differences between two groups were analyzed with the nonparametric Mann-Whitney U test. A value ≤ 0.05 was considered significant. 18 (Figure 3 and Table 2). Figure 2   in the BSEP membrane vesicles. A sigmoidal dose-response curve was fitted on the acquired data points to determine the IC50 and calculate the Ki-value ( Figure 3 and Table 2). 18  Time-activity curves of the 18 F labeled bile acids were generated and the relevant parameters were extracted from these graphs (Figure 4 and Table 3). Representative PET/CT images of 3 -[ 18 F]FGCA and 3 -[ 18 F]FCDCA are displayed in Figure 5.

Discussion
Cholestasis, a toxic accumulation of bile acids in the liver or blood, may occur in certain liver diseases or can be triggered as a result of a xenobiotic interfering with the hepatobiliary transport of bile acids [15]. It is therefore important to have an adequate tool to evaluate the efficiency of bile acid transport in vivo for clinical use or in drug development.
To that end, PET-imaging of hepatobiliary transport with radiolabeled bile acids is gaining importance. A number of studies have been published in which bile acids were labeled with either 11 C or 18 F on different bile acid structures [8][9][10]12].  Table 2. All data are mean ± SD ( = 3).
Ki-values of CA, 3 FCA, 3 FGCA, 3 FCDCA, and 12 FCA for the bile acid transporters NTCP, OATP1B1, and BSEP were determined as an affinity measure. It was demonstrated that these compounds caused a concentration dependent decrease in uptake of the tritium labeled model substrates of NTCP, OATP1B1, and BSEP. The substitution of the 3-OH function by a more lipophilic fluorine atom on the 3 -and 3 -position of CA causes an increase in affinity for NTCP and OATP1B1, yet gives rise to a decrease in affinity for the BSEP-transporter compared to CA. Although both 3-FCA epimers are a substrate for the bile acid transporters, 3 -FCA displays a slightly higher affinity than 3 -FCA for NTCP and OATP1B1 (2.53 ± 0.76 M and 5.22 ± 0.56 M, resp., versus 6.18 ± 0.59 and 9.67 ± 0.83 M, resp.). The affinity for BSEP does not change. This difference in affinity for 3 / cholic acid epimers of bile acid transporters was already uncovered in cell lines that express the Apical Sodium-dependent Bile acid Transporter (ASBT; responsible for basolateral uptake of bile acids in the enterocytes). It was found that 3 -OH-bile acids have a lower affinity than 3 OH-bile acids for ASBT [17]. In the present study, the hepatic basolateral uptake transporters NTCP and OATP1B1 also reveal a slight preference in affinity for the 3 -fluorocholic acid epimer. Compound 3 -FGCA, 3 -cholic acid conjugated with the amino acid glycine, showed an increase in affinity for all bile acid transporters under investigation. This is in line with literature data: conjugated bile acids have a higher affinity for NTCP, OATP, and BSEP than nonconjugated bile acids [2,18].