Synthesis, in vitro and in vivo evaluation of 3β-[18F]fluorocholic acid for the detection of drug-induced cholestasis in mice

Introduction Drug-induced cholestasis is a liver disorder that might be caused by interference of drugs with the hepatobiliary bile acid transporters. It is important to identify this interference early on in drug development. In this work, Positron Emission Tomography (PET)-imaging with a 18F labeled bile acid analogue was introduced to detect disturbed hepatobiliary transport of bile acids. Methods 3β-[18F]fluorocholic acid ([18F]FCA) was prepared by nucleophilic substitution of a mesylated precursor with [18F]fluoride, followed by deprotection with sodium hydroxide. Transport of [18F]FCA was assessed in vitro using CHO-NTCP, HEK-OATP1B1, HEK-OATP1B3 transfected cells and BSEP & MRP2 membrane vesicles. Investigation of [18F]FCA metabolites was performed with primary mouse hepatocytes. Hepatobiliary transport of [18F]FCA was evaluated in vivo in wild-type, rifampicin and bosentan pretreated FVB-mice by dynamic μPET scanning. Results Radiosynthesis of [18F]FCA was achieved in a moderate radiochemical yield (8.11 ± 1.94%; non-decay corrected; n = 10) and high radiochemical purity (>99%). FCA was transported by the basolateral bile acid uptake transporters NTCP, OATP1B1 and OATP1B3. For canalicular efflux, BSEP and MRP2 are the relevant bile acid transporters. [18F]FCA was found to be metabolically stable. In vivo, [18F]FCA showed fast hepatic uptake (4.5 ± 0.5 min to reach 71.8 ± 1.2% maximum % ID) and subsequent efflux to the gallbladder and intestines (93.3 ± 6.0% ID after 1 hour). Hepatobiliary transport of [18F]FCA was significantly inhibited by both rifampicin and bosentan. Conclusion A 18F labeled bile acid analogue, [18F]FCA, has been developed that shows transport by NTCP, OATP, MRP2 and BSEP. [18F]FCA can be used as a probe to monitor disturbed hepatobiliary transport in vivo and accumulation of bile acids in blood and liver during drug development.


Introduction
Drug-induced cholestasis is a liver disorder that might be caused by interference of drugs with the hepatobiliary bile acid transporters. It is important to identify this interference early on in drug development. In this work, Positron Emission Tomography (PET)-imaging with a 18 F labeled bile acid analogue was introduced to detect disturbed hepatobiliary transport of bile acids.

Introduction
The production of bile is an important function of the liver. One of the primary constituents are bile acids: amphiphilic molecules synthesized by hepatocytes that play a vital role in digestion of lipids and uptake of fat-soluble vitamins [1]. Bile acids are excreted in the canals of Hering, stored in the gallbladder and excreted into the duodenum via the common bile duct. Bile acids are part of the enterohepatic recirculation and are reabsorbed from the small intestine into the portal vein and transported back to the hepatocytes, where uptake occurs primarily by the basolateral transport protein Na + -dependent Taurocholate Cotransporting Polypeptide (NTCP). However, the Organic Anion Transporting Polypeptide (OATP) is also capable of transporting bile acids into the hepatocyte. Hepatic efflux of bile acids towards the bile canaliculi is mediated mainly by the Bile Salt Export Pump (BSEP) and also by the Multidrug Resistance-associated Protein 2 (MRP2) [1].
Drug-induced liver injury (DILI) is an acquired liver disorder responsible for a significant amount of hospitalizations and a prime cause of rejecting new drug candidates during drug development [2,3]. A major part of DILI is represented by drug-induced cholestasis (DIC), which results from inhibition of the bile acid transporters by drugs, leading to a toxic accumulation of bile acids in the liver [4,5].
It is important to detect drug-induced cholestasis early on in drug development. In this regard, nuclear imaging is a powerful tool to investigate interference with the bile acid transporters on a molecular level [6]. Various radiotracers have already been developed that show hepatobiliary transport by these transporters. Single Photon Emission Computed Tomography (SPECT)-tracers such as 99m Tc Mebrofenin, [ 99m Tc]-DTPA-CDCA and [ 99m Tc]-DTPA-CA are substrates of OATP1B1, OATP1B3 and MRP2 [7,8]. Although the latter two are bile acid analogues, no transport by NTCP or BSEP was observed. [ 11 C]dehydropravastatin, [ 11 C]rosuvastatin, [ 11 C]TIC-Me, [ 11 C]glyburide and [ 11 C]telmisartan are Positron Emission Tomography (PET)-tracers that provide insight into (altered) transport function by OATP, NTCP or MRP2 [9][10][11][12][13]. However, in order to study bile acid transport and the corresponding disturbances, the desired tracer would be a radiolabeled bile acid, predominantly transported by NTCP, BSEP, and by OATP and MRP2 [14].
The synthesis and in vivo evaluation of different 11 C labeled bile acid analogues such as [ 11 C]cholylsarcosine was described [15,16]. Although the results are promising, the half-life of the 11 C-isotope can limit its use. Consequently, a 18 F labeled bile acid was developed and evaluated in vivo in mice by Jia et al. [17]. In this study the 18 F isotope was incorporated in the bile acid by modification of the carboxyl functional group. Due to this major structural modification however, questions were raised whether the transport mechanism of this tracer was still comparable to endogenous bile acids [18]. To our knowledge, none of these PET bile acid analogues has had its transport characterized in vitro.
Therefore, the aim of this work was to develop and evaluate a 18 F labeled bile acid with minor modifications on the endogenous bile acid structure. This tracer can represent endogenous bile acid transport and can be used as imaging probe for preclinical evaluation of drug interference with the hepatic bile acid transporters. In vitro assays were performed to determine the involved bile acid transporters for uptake in, -and efflux out of, the liver. As proof of this concept, imaging experiments in mice were performed with the hepatotoxic drugs rifampicin and bosentan. Rifampicin is a known inhibitor of human/rodent OATP/oatp and MRP2/mrp2 [9,19,20]; bosentan of NTCP/ntcp, and BSEP/bsep [21][22][23].
The precursor for radiosynthesis (Fig 1; 4 mg MsAcCAME dissolved in 200 μL anhydrous DMSO) was added to the vial. The reaction mixture was heated for 20 minutes at 120˚C. Radiofluorination of precursor molecule MsAcCAME was monitored by TLC (10 mM NH 4 Ac:AcN 1:4 v:v). The vial was subsequently cooled to room temperature and NaOH (100 μL 3 M) was added to allow deprotection of the labeled intermediate [ 18 F]FAcCAME. Deprotection took place for 10 minutes at 120˚C. The vial was cooled to room temperature and the reaction mixture was purified by means of semi-preparative HPLC (Grace Econosphere C18 10.0x250 mm, 10 μm; 6 mL/min AcN:H 2 O 10:90 v:v ->AcN 100% in 20 minutes as mobile phase; radiodetection (Ludlum Measurements Inc)). The desired HPLC-fraction was collected, diluted with ultrapure water and loaded onto a Sep-Pak C18 cartridge (preconditioned with 5 mL EtOH and 5 mL ultrapure water). The cartridge was washed with 5 mL ultrapure water and eluted with 500 μL EtOH. The solvent was evaporated under a gentle nitrogen flow and [ 18 F]FCA was reformulated with 500 μL phosphate buffered saline (PBS) pH 7.4. This formulation was subjected to quality control by means of analytical HPLC with radio-and UV detection (205 nm; Waters) using a Grace Alltima C18 (4.6x250 mm; 5 μm) column and 1 mL/min H 2 O:AcN 70:30 v:v as mobile phase. The retention time of [ 18 F]FCA was compared to that of a non-radioactive reference compound.
1 MBq of [ 18 F]FCA was used to spike an octanol/PBS 50:50 v:v pH 7.4 mixture in a test tube. After mixing and centrifugation (5 min; 1100 g), aliquots were taken from each layer and radioactivity was measured in a NaI (Tl) scintillation counter (Capintec; Ramsey, NJ, USA). The LogD value of [ 18 F]FCA was calculated as the logarithm of the ratio of counts in octanol and PBS.
In vitro uptake assay. The plates were incubated for 10 minutes at 37˚C and uptake was halted by the addition of 150 μL ice-cold buffer solution. The contents of each well were transferred to a glass fiber filter plate (Multiscreen HTS plates; Merck Millipore) and washed 3 times with 200 μL of icecold buffer solution. Vesicles were lysed by incubation with 100 μL 0.1 M NaOH for 10 minutes at room temperature. An 80 μL aliquot of this solution was used for liquid scintillation counting. Data points were collected in triplicate. IC50-values were calculated with Graphpad Prism v3.00.

In vitro investigation of [ 18 F]FCA stability
The stability or metabolization of [ 18 F]FCA was assessed in its formulation, in mouse serum and in presence of primary mouse hepatocytes. Mouse serum (Sigma Aldrich, Bornem, Belgium) was spiked with 37 MBq [ 18 F]FCA (1.0 mL total volume) and incubated at 37˚C for 5, 10, 30 and 60 minutes.
Primary mouse hepatocytes were isolated as described previously [24], formulated in DMEM medium at 1 million cells/mL and immediately incubated with 37 MBq [ 18 F]FCA (1.0 mL total volume) at 37˚C, 5% CO 2 for 5, 10, 30 and 60 minutes. A control sample without hepatocytes was included.
At the indicated timepoints, 100 μL serum or hepatocyte incubation medium was withdrawn and 100 μL AcN was added. Samples were then centrifuged at 13000 g for 5 minutes and the supernatant was analyzed by the same semi-preparative RP-HPLC method as for the radiosynthesis of [ 18 F]FCA.

In vivo evaluation of [ 18 F]FCA
Hepatobiliary transport of the tracer was evaluated in female wild-type FVB mice (5 weeks old, Charles River). The animals were housed in accordance with European Ethics Committee guidelines. The animal studies were approved by the Animal Experimental Ethical Committee of Ghent University (ECD 15/69). Food and water was provided ad libitum. Animals were fasted overnight before the PET-scan.
Mice were anesthetized with 1.5 v:v % isoflurane in 100% O 2 for the duration of the experiment. An intravenous polyethylene line was placed in the lateral tail vein. The animals were placed in the small animal PET/CT-scanner (FLEX Triumph II small animal PET/CT-scanner; PET field of view: 7.5 cm axial; 1.3 mm spatial resolution; TriFoil Imaging) on a heated bed. A CT-scan was acquired for anatomical correlation.
In PET-data were obtained in list-mode and were iteratively reconstructed (50 iterations) in frames of 15 seconds for the first 10 minutes, followed by 50 frames of 1 minute. ROI's were drawn manually over liver, gallbladder and intestines using PMod software. A ROI was drawn in the left ventricle on the static [ 18 F]FDG scan to obtain an image-derived arterial blood input curve [25]. The uptake of [ 18 F]FCA in arterial blood, liver, gallbladder and intestines was expressed as a % injected dose (% ID) and normalized for the weight of a 20 gram mouse. Time-activity curves (TAC's) were generated using these values.
AUC, maximum % ID, and time to peak values were determined for the TAC's with Graphpad Prism v3.00. Differences between 2 groups were determined with the non-parametric Mann-Whitney U test. P-values 0.05 were considered significant.

In vitro uptake
To determine the transport characteristics of 3β-fluorocholic acid, a competition experiment with 3 (Fig 2). IC50 values for this competition assay were 6.6 ± 0.8 μM for NTCP; 10.5 ± 1.1 μM for OATP1B1 and 25.2 ± 12.7 μM for OATP1B3.  (Fig 3). IC50 values of this inhibition were 252.6 ± 76.0 μM for BSEP and 577.2 ± 300.2 μM for MRP2.  The accumulation of the tracer in the liver reached a maximum % ID of 71.8 ± 1.2% ID with the time to peak being 4.5 ± 0.5 min, after which a decrease to baseline levels was observed, indicating excretion of radioactivity. A corresponding increase of [ 18 F]FCA was observed in the gallbladder and intestines. After 1 hour, almost all radioactivity was found in the gallbladder and intestines (93.3 ± 6.0% ID) (Fig 5, the corresponding metrics are shown in Table 1). No uptake was observed in the urinary bladder or other organs.

In vivo evaluation of [ 18 F]FCA
The hepatobiliary transport of [ 18 F]FCA was studied in mice who were dosed with rifampicin or bosentan. TAC's and metrics of [ 18 F]FCA distribution in rifampicin and bosentan treated mice are depicted in Fig 6 and Table 2. Rifampicin and bosentan treated mice showed an altered distribution of [ 18 F]FCA compared to control animals: time to peak of the liver TAC significantly increased 2.9 and 1.7-fold versus control for rifampicin and bosentan respectively. The maximum % ID in the liver was also significantly lower than control for both compounds (49.7 ± 1.6% and 54.4 ± 1.7% for rifampicin and bosentan respectively, compared
doi:10.1371/journal.pone.0173529.g002 to 71.2 ± 3.5% for control). Efflux towards the gallbladder and intestines was impaired: the AUC-value of the gallbladder & intestines TAC was 3.1 times lower than control for rifampicin and 1.7 times lower for bosentan. The AUC of the arterial blood TAC was 2.1-fold and 2.5-fold higher than control for rifampicin and bosentan respectively (see Fig 7).

Discussion
Bile acids are amphiphilic steroid derivatives that play a vital role in the digestion of lipids and uptake of fat-soluble vitamins. In many species, including rodents and humans, cholic acid makes up the larger part of the bile acid spectrum [26]. Cholic acid is therefore an opportune target to radiolabel and use as an imaging probe for hepatobiliary transport of bile acids in mice and humans alike. In this article, a 18 F labeled bile acid, 3β-[ 18 F]fluorocholic acid, is introduced to assess hepatobiliary transport of bile acids in mice. This 18 F PET-tracer shows remarkable structural resemblance to cholic acid, differing only in substitution of the 3α-OH with a 3β-fluorine. It was hypothesized that this modification would not have a major effect on transport by the bile acid transporters, as fluorine is isosteric and iso-electronic with a OHgroup. Moreover, the 3α-OH group is not an absolute requisite for active bile acid transport [27,28].
The radiofluorination of MsAcCAME with [ 18 F]fluoride showed an incorporation yield of 30%, which is in line with radiofluorination of a secondary mesylate directly attached to a cyclohexane ring [29]. This moderate yield can be attributed to the sterically hindered environment of the leaving group. After purification and formulation of [ 18 F]FCA in PBS, no radiochemical degradation products, metabolization or radiodefluorination was observed: the tracer remained stable for at least 6 hours in its formulation and at least 1 hour in presence of mouse serum and primary mouse hepatocytes. The logD value of [ 18 F]FCA was in accordance with the logD value of cholic acid (logD = 1.1) [30].
In vitro transport evaluation of [ 18 F]FCA revealed that the hepatic bile acid transporters involved in uptake and efflux are NTCP, OATP1B1, OATP1B3, BSEP and MRP2, which is also the case for endogenous bile acids [1,31,32]. Substitution of the 3α-OH of cholic acid with a 3β-fluorine did not have an effect on the ability to be transported by the same hepatobiliary transporters as endogenous bile acids.
The in vivo evaluation of [ 18 F]FCA in wild-type FVB mice showed that the tracer is rapidly and exclusively taken up into the liver after intravenous injection (4.5 min to reach 71.8 maximum % ID). Compared to 99m Tc-mebrofenin in FVB-mice (2.2 minutes to reach 51.8 maximum % ID) [33], uptake into the liver is slightly slower and reaches a comparable % ID. This behavior is similar to 11 C labeled bile acid analogues which also show a fast and exclusive hepatic uptake (<7 minutes), in pigs [15,16]. On the other hand, 99m Tc labeled bile acids 99m Tc CDCA and 99m Tc CA reach a lower peak value in the liver at reduced rate (12.7 minutes to reach 37.3 max % ID and 12.0 minutes to reach 25.7 max % ID respectively) [8]. Following rapid uptake into the liver, [ 18 F]FCA is excreted into the bile ducts, gallbladder and finally accumulates in the intestines. There was no accumulation of tracer visible in other organs for up to 6 hours post-injection. After 1 hour, activity in the gallbladder and intestines reached a Evaluation of a 18 F bile acid in mice for the detection of cholestasis plateau of 93.3% ID. This value is comparable to that of 99m Tc mebrofenin (78.1% ID) and higher than the maximal % ID in gallbladder and intestines of the 99m Tc labeled bile acids (62.2% ID for 99m Tc CDCA; 47.8% for 99m Tc CA). Because of their partial urinary clearance, 99m Tc bile acids show a lower uptake into the liver and a lower accumulation in gallbladder and intestines.
[ 18 F]FCA was developed to detect disturbed hepatobiliary transport of bile acids. As proof of concept, the oatp and mrp2 inhibitor rifampicin and the ntcp and bsep inhibitor bosentan were administered to FVB-mice and subjected to a PET-scan with [ 18 F]FCA. Both rifampicin and bosentan caused a significant delay in hepatic uptake and lower maximum % ID in the liver and both drugs gave rise to an increased amount of [ 18 F]FCA in arterial blood, which is in line with an observed increase in serum bile acids after administration of these drugs [19,22]. When an identical dose of rifampicin was administered to FVB-mice scanned with 99m Tc mebrofenin [33], a complete block of canalicular efflux was observed on the SPECTscan, whereas [ 18 F]FCA excretion into bile is only hindered. This difference in excretion can be attributed to the maximal blocking effect of rifampicin on 99m Tc mebrofenin transport, as this tracer is an exclusive oatp and mrp2 substrate [7]. [  F]FCA is rapidly taken up into the liver and excreted in gallbladder and intestines. Significant alterations in [ 18 F]FCA uptake and efflux parameters were observed with the hepatotoxic drugs rifampicin and bosentan. [ 18 F]FCA shows favorable in vitro and in vivo characteristics that are comparable to endogenous bile acids and can therefore be a promising imaging probe in preclinical drug development for the detection of drug-induced altered hepatobiliary transport of bile acids.