Blocking of efflux transporters in rats improves translational validation of brain radioligands

Background Positron emission tomography (PET) is a molecular imaging technique that can be used to investigate the in vivo pharmacology of drugs. Initial preclinical evaluation of PET tracers is often conducted in rodents due to the accessibility of disease models as well as economic considerations. Compared to larger species, rodents display a higher expression and/or activity of efflux transporters such as the P-glycoprotein (P-gp). Low brain uptake could, therefore, be species-specific and uptake in rodents not be predictive for that in humans. We hypothesized that a better prediction from rodent data could be achieved when a tracer is evaluated under P-gp inhibition. Consequently, we compared the performance of eight neuroreceptor tracers in rats with and without P-gp inhibition including a specific binding blockade. This data set was then used to predict the binding of these eight tracers in pigs. Methods PET tracers targeting serotonin 5-HT2A receptors ([18F]MH.MZ, [18F]Altanserin, [11C]Cimbi-36, [11C]Pimavanserin), serotonin 5-HT7 receptors ([11C]Cimbi-701, [11C]Cimbi-717 and [11C]BA-10) and dopamine D2/3 receptors ([18F]Fallypride) were used in the study. The brain uptake and target-specific binding of these PET radiotracers were evaluated in rats with and without inhibition of P-gp. Rat data were subsequently compared to the results obtained in pigs. Results Without P-gp inhibition, the amount of target-specific binding in the rat brain was sufficient to justify further translation for three out of eight evaluated tracers. With P-gp inhibition, results for five out of eight tracers justified further translation. The performance in pigs could correctly be predicted for six out of eight tracers when rat data obtained under P-gp inhibition were used, compared to four out of eight tracers without P-gp inhibition. Conclusions P-gp strongly affects the uptake of PET tracers in rodents, but false prediction outcomes can be reduced by evaluating a tracer under P-gp inhibition.

translational research and drug development to noninvasively quantify biological targets, dose-occupancy relationships or therapeutic response [2,3].
During preclinical development of novel PET tracers, it is important to use animals that represent human physiology as closely as possible. Rodents are usually tested first, for a number of reasons including lower costs compared to larger animals and the availability of disease models [4]. Compared to other species, rodents have been demonstrated to have increased brain expression of the efflux transporter, P-glycoprotein (P-gp), both in absolute terms and relative to other drug efflux transporters [5,6]. P-gp is an ATP-dependent efflux pump, localized at the luminal side of the brain capillary endothelium which forms the blood-brain barrier (BBB). Substrates of P-gp are pumped out into the lumen of the brain capillaries and thus removed from the brain tissue [7]. P-gp activity is known to be a major factor limiting the brain uptake of PET tracers in rodents [8,9]. If tracers for central nervous system (CNS) targets are initially evaluated in rodents and show low brain uptake, they could be de-selected for further translation, even though they would have worked in higher species, including humans.
We have previously demonstrated that the brain uptake of two PET tracers used for brain receptor imaging, [ 11 C]GR205171 and [ 18 F]Altanserin, was three-sixfold lower in rats than in humans, which is unsatisfactorily low [10]. In the same study, the uptake of [ 18 F]Altanserin in the minipig brain was 3.8-fold higher than in the rat brain, apparently due to a lower P-gp activity in pigs. Detecting brain uptake of this magnitude during initial evaluation of a tracer would warrant further translation to humans.
In this study, we investigate if rodents generally display higher efflux transporter activity than pigs by studying the brain uptake of eight structurally different PET tracers in both rodents and pigs. Furthermore, we propose an experimental design which accounts for elevated P-gp activity in rodents and facilitates the translation of tracers from rodents to higher species. Specifically, we supplement the traditional evaluation of tracer uptake at baseline and after selective target block with concomitant inhibition of P-gp activity. Figure 1 summarizes our set-up. In short, rats were scanned under four experimental conditions: drug-naïve or "baseline" (1), target block (2), P-gp inhibition (3) and target block combined with P-gp inhibition (4). The advantage of this set-up is that, in addition to assessing the specific binding of the tracer, it determines whether the tracer is a P-gp substrate or not. Importantly, this setup is compatible with the higher-throughput scanning approach earlier reported by us [11] Results of the tracer evaluation via both the traditional workflow (baseline and target block conditions) and the new workflow (P-gp inhibition alone and combined with target block) were used to predict the performance of the same tracer in pigs. The prediction algorithm is illustrated in Additional file 1: Fig. S1. We hypothesized Step 1 and 2: Traditionally, specific binding can be determined by comparing tracer uptake at baseline and after target block (Case A). This work proposes to perform the same two experiments with simultaneous inhibition of P-gp (Case B).
Step 3: If specific binding is observed in Case A or B, a successful translation from rodent to pig is predicted Shalgunov et al. EJNMMI Res (2020) 10:124 that additional evaluation of a tracer's P-gp dependency would give a better prediction of its performance in pigs and ultimately in humans than the traditional workflow.

Pig PET experiments
Description of PET experiments in pigs and analysis of obtained data can be found in the Additional file 1.

Rat PET scanning protocol
On the day of scanning, rats were transported to the scanner at least one hour prior to the experiment. Isoflurane, 3-3.5% in 0.6% oxygen, was used to induce anaesthesia, while anaesthesia was maintained with 2.0-2.5% isoflurane during the scans. The PET tracers were administered as intravenous (i.v.) bolus injections via tail vein catheters (BD Neoflon 25G, Stockholm, Sweden) at the beginning of the scan with doses being between 5-20 MBq. The rats were subsequently scanned in the High-Resolution Research Tomograph (HRRT) scanner (Siemens AG, Munich, Germany) using a custom-made 2 × 2 rat holder, which enabled simultaneous scanning of four rats ( Fig. 1) [11]. The animals were scanned for either 60 min ( 11 C) or 90 min ( 18 F), followed by a transmission scan at speed 10 (acquisition time approximately 6 min). The animals were scanned at baseline (no pre-treatment before tracer injection) and after receiving P-gp inhibition with or without target block. Details of inhibition and blocking regimens are described in Table 1. For combined inhibition and target block scans, rats were pre-treated with the P-gp inhibitor elacridar (5 mg/kg, Carbosynth, Compton, UK) and the 5-HT 7 receptor antagonist SB-269970 (3 mg/kg, Tocris Bioscience, Abingdon, UK), the sigma and dopamine D 2/3 receptor antagonist haloperidol (1 mg/kg, Janssen-Cilag, Birkerød, Denmark) or the 5-HT 2A receptor antagonist ketanserin (3 mg/kg, Sigma-Aldrich, Saint Louis, Missouri, USA). Elacridar was given 30 min prior to tracer injection through the intravenous catheter, and receptor blocking drugs were given 15 min before tracer injection. Chosen dosages and pre-treatment intervals were based on literature data [20][21][22][23]. Studies with [ 18 F]Altanserin were performed in a MicroPET Focus 120 scanner (Siemens Medical Solutions, Malvern, PA, USA). Hyponorm/midazolam (VetaPharma Ltd., Leeds, UK/Hameln Pharmaceuticals, Hameln, Germany) was used as anaesthesia. The rats received 11 ± 2 MBq of the tracer as a bolus injection. Three of the rats were administered with 22.5 mg/ kg bolus of cyclosporin A (Sandimmun, Novartis, Basel, Switzerland) followed by a constant infusion of 7.5 mg/ kg/h, starting 20-25 min before radiotracer administration, as previously described [10].

Quantification of rat PET data
For data analysis, the software PMOD 3.7 (PMOD Technologies, Zürich, Switzerland) was used. Summed PET images were generated based on all counts recorded in the time intervals specified in Table 1. Images were then aligned to a standardized MRI-based atlas of the rat brain [24] from where pre-defined regions of interest (ROIs) were extracted. Regions known to possess high densities of relevant receptors were selected as target ROIs: medial prefrontal cortex (mPFC) and frontal cortex (FC) for 5-HT 2A receptor PET tracers [25], thalamus (Tha) for 5-HT 7 receptor PET tracers [16] and striatum (Str) for dopamine D 2/3 receptor tracer [26]. Cerebellum region (Cb), having low densities of receptors targeted by all investigated tracers, was chosen as a reference region (Additional file 1: Fig. S3). In addition, whole brain (Wb) ROI was chosen to monitor overall tracer penetration into the brain. The time-activity curves (TACs) for target ROIs were extracted from the PET images, and the activity was converted into standardized uptake values (SUV). SUV, expressed in g/mL, is equal to the concentration of radioactivity measured in the ROI divided by the injected radioactivity dose per body weight. Area under the curve (AUC) values were calculated from the TACs using GraphPad Prism 7 (GraphPad Software, California, USA) and expressed in min × g/mL. As scanning durations were different for 11 C-labelled and 18 F-labelled tracers (60 min and 90 min, respectively), AUC values of 18 F-labelled tracers were calculated for both the full duration and the first 60 min of the scan.

Rat data analysis
Apparent target-specific binding of the tracers in the target regions was expressed as specific binding ratios (SBR), which were calculated from mean full scan length AUC values applying Eq. 1; cerebellum was used as a reference region for all tracers. Changes in apparent specific binding of the tracers in response to target receptor blockade were calculated from SBR values using Eq. 2; SBR changes were calculated for "baseline-target block" and "P-gp inhibition alone-combined P-gp inhibition and target block" condition pairs. Changes in tracer uptake in the target-rich region in response to P-gp inhibition were calculated from mean AUC values for the first 60 min of the scan as shown in Eq. 3.
Evaluation of a tracer was considered successful if (whether with or without P-gp inhibition) the tracer showed an SBR value of at least 0.15 (15% higher uptake in the target region relative to reference region), and this SBR value decreased by at least 30% under target block condition.

Results
Of the four 5-HT 2A receptor tracers investigated, only [ 18 F]MH.MZ showed substantial uptake in the rat brain at baseline: AUC for the target receptor-rich mPFC was 60 ± 5 min × g/mL (here and further AUC values refer to the first 60 min of the scan unless stated otherwise; full scan length AUC values for all tracers, ROIs and experimental conditions are presented in Additional file 1: Table S1; numbers of rats scanned per experimental condition are shown in Additional file 1: Table S2). Baseline SBR for mPFC equalled 0.71. Pre-treatment with ketanserin, a 5-HT 2A receptor antagonist, decreased SBR to 0.02 (− 97% change). Under P-gp inhibition with elacridar, the AUC for the mPFC rose to 147 min × g/ mL (+ 143%), and the SBR value for the mPFC reached 1.04. Combining target blockade with P-gp inhibition decreased the SBR to 0.04 (− 96%). Thus, the 5-HT 2A receptor-specific binding and the blocking of it were C]Pimavanserin, AUC in the cerebellum after combined ketanserin and elacridar pre-treatment was greater than AUC in the mPFC, resulting in a negative SBR value of -0.10. However, a change in SBR from 0.03 to -0.10, although high in relative terms (-395%), can hardly be interpreted as a sign of target-specific binding because both values are very low.
All in all, for [ 18 F]Altanserin, 5-HT 2A receptor-specific binding and the blocking effect became observable after P-gp inhibition, for [ 11 C]Cimbi-36 the use of P-gp inhibition led to an amplification of both apparent specific binding at baseline and the blocking effect, while [ 11 C] Pimavanserin did not show any sign of specific 5-HT 2A receptor binding either with or without P-gp inhibition (Fig. 2a).
Combined P-gp inhibition and target (D 2/3 ) block decreased striatal SBR to 0.22 (− 95%). Therefore, targetspecific binding was clearly visible both without and with P-gp inhibition. Evaluation

Discussion
In this work, we propose a higher-throughput approach to improve the translation of CNS PET tracers from rodents to higher species. Our data suggest that one of the main factors limiting translation of CNS tracers from rats to larger animals is the higher activity of efflux pumps, including P-gp, in rats. We used cyclosporine A and elacridar to inhibit the action of P-gp, the dominant drug efflux transporter in the rodent brain and study the influence of P-gp on the accumulation of a set of eight tracers [5]. Cyclosporine A is a selective P-gp inhibitor, while elacridar inhibits both P-gp and BCRP (breast cancer resistance protein) efflux pumps [29]. Within our data set, all investigated tracers showed higher brain uptake in rats when the action of P-gp was inhibited. The smallest increase in target-rich region AUC values after P-gp inhibition was observed for [ 11 C]Cimbi-717 (47%) and [ 18 F] Altanserin (93%), while for all other tracers, AUC values increased by 100-700%. Brain uptake of [ 18 F]Altanserin was previously reported to increase 2.6-fold (by 160%) in response to P-gp inhibition [10], which is higher but still comparable to the results obtained in this work (+ 93%), using the same P-gp inhibition condition. Higher increase reported in [10] could have been the consequence of a higher body weight of the rats used for the experiments (370 g vs 200-300 g in this work). It should be noted that all compounds evaluated by us are secondary and/or tertiary amines, which tend to be P-gp substrates [30,31]. However, an amino group is often present in CNS PET tracers.
In  15) in target-rich regions and more than 30% reduction in SBR values after a selective target block (Figs. 2 and 4). [ 18 F]Altanserin demonstrated a decent SBR value (0.33) at baseline without P-gp inhibition, but ketanserin pre-treatment led to a drastic decrease of tracer uptake in both frontal cortex and cerebellum, with uptake in the cerebellum falling even more drastically (Additional file 1: Table S1). As a consequence, [ 18 F]Altanserin´s SBR after target blockade was higher compared to baseline conditions. Likewise, brain uptake in the reference ROI decreased more than in the target ROI in response to target blockade for [ 11 C]BA-10. These paradoxical observations may be explained by the hindered diffusion of the tracers across the BBB or their accelerated washout from the brain caused by the vasodilatory [20] or vasoconstrictive [32] properties of the respective blocking drugs (ketanserin and SB-269970). Saturation of P-gp or other efflux transporters by the blocking drugs is an unlikely explanation, because this would have increased the uptake of the tracers in both target and reference ROIs. All in all, further research is required to unequivocally resolve this issue.   Table 1 for details on drug dosing and PET image summation). Numbers of rats scanned per experimental condition are shown in Additional file 1: Table S2. Wb whole brain, Tha thalamus, Cb cerebellum, AUC area under the curve, SBR specific binding ratios, P-gp P-glycoprotein We hypothesized that the target binding of tracers that are actively transported out of the rat brain can nevertheless be assessed once the increased activity of efflux pumps is inhibited. Therefore, all eight tracers were tested under P-gp inhibition with and without a simultaneous target block. P-gp inhibition led to an increase in baseline SBR values for all tracers (Additional file 1: Fig. S8 Fig. 3). For the latter two tracers, their uptake in the target and reference ROIs under P-gp inhibition either did not change in response to target receptor blockade or slightly increased in both ROIs, which led to an increase in SBR (Fig. 3). This could have been caused by the perturbations in the cerebral blood flow due to the pharmacological action of the blocking drug (SB-269970, see above) or by the influx into the brain of the excess of the tracers displaced from 5-HT 7 receptors in peripheral tissues by the blocking drug.
Based on our data in rats, we predicted whether a tracer would work in pigs or not (see Additional file 1: Fig. S1) and compared these predictions to actual results obtained in pigs, as well as to the evaluation results in humans for those tracers that reached the clinical evaluation stage ([ 18 Table 2 and Additional file 1: Table S3 summarize the outcomes. In contrast to rats (Figs. 2 and 3), low brain uptake was not an issue for any of the investigated tracers during evaluation in pigs and humans (Additional file 1: Table S3 and [33][34][35][36]). This confirms the notion that rats generally have a highly efficient brain efflux transporter system, which can limit the uptake of tracers in the brain, whereas in larger/higher species the efflux transporter system has less influence on the tracer uptake. Seven out of eight tracers displayed target-specific binding in pigs, i.e. pre-treatment with a specific receptor blocking agent reduced the binding potential (BP ND ) of the tracer in the target region (by 30% or more). For all tracers except [ 11 C]Cimbi-701 and [ 11 C] Cimbi-717, which had shown specific binding in pigs but not in rats, baseline BP ND values in pigs highly correlated with baseline SBR values in rats, both with and without P-gp inhibition (Additional file 1: Fig. S9). [ 11 C]Pimavanserin did not show any specific binding in pigs.
In our study, blocking experiments in rats without P-gp inhibition detected specific binding for three tracers ([ 18 F]MH.MZ, [ 11 C]Cimbi-36 and [ 18 F]Fallypride) and without additional experiments in non-rodent species, the remaining tracers would likely have been discarded. The use of P-gp inhibition helped to additionally identify two tracers ([ 18 F]Altanserin and [ 11 C]BA-10) that showed specific binding. [ 11 C]Pimavanserin did not show any specific binding either with or without P-gp inhibition in rats, and behaved in the same way in pigs. Agreement between pig and rat data on [ 11 C]Pimavanserin is a positive finding within the framework of this study, even though it may be perceived as disappointing that a ligand with demonstrated sub-nanomolar affinity and high selectivity towards 5-HT 2A receptors [37] turned out to be unsuitable for receptor imaging. Rodent data for [ 11 C]Cimbi-701 and [ 11 C]Cimbi-717, on the other hand, demonstrate the limitations of brain radioligand evaluation in rodents: even with the use of P-gp inhibition, these tracers would have been discarded based on rodent  Table 1 for details on drug dosing and PET image summation, n = 2). Str striatum, Cb cerebellum, Wb whole brain, AUC area under the curve, SBR specific binding ratios, P-gp P-glycoprotein data only. Apart from efflux transporter expression and activity, inter-species differences in receptor abundances or metabolic pathways are factors that may have a crucial impact on the translational validation of PET tracers [38,39]. Densities of 5-HT 2A , 5-HT 7 and D 2/3 receptors in the respective receptor-rich regions of rodent, pig and human brain are summarized from literature data in Additional file 1: Table S4 [28,[40][41][42]. Densities of 5-HT 7 receptors in rat and pig thalamus are very close, so the differences in metabolism between rats and pigs are a more likely explanation for the discrepant results of [ 11 C] Cimbi-701 and [ 11 C]Cimbi-717 evaluation in these two species.

Conclusions
The inclusion of P-gp inhibition into the workflow helped to predict the outcome of six out of eight cases (75% success rate), even though all tracers were strong substrates of P-gp. A traditional evaluation workflow without P-gp inhibition could only predict the outcome for four out of eight tracers (50% success rate). Our results demonstrate how addition of P-gp inhibition can aid in the interpretation of initial tracer evaluation in rats, improve the translatability and minimize unjustified discontinuations of promising tracers. We believe that the proposed workflow allows for a more effective initial in vivo screening using rats.
Additional file 1. Detailed overview of the experimental design, description of experiments in pigs, chemical structures of all studied compounds, investigated brain regions on the MRI template, time-activity curves for rat and pig ([ 18 F]Fallypride and [ 11 C]BA-10) scans, SBR and AUC values for all tracers, numbers of rats scanned per experimental condition are detailed in the Additional file.

Table 2 Short summary of tracer evaluation in rats, pigs and humans
Concrete estimates of target-specific binding at baseline and its changes in response to target receptor blockade in rats and pigs, as well as estimates of targetspecific binding in humans are summarized in Additional file 1: Table S3 a Traditional evaluation workflow without P-gp inhibition b P-gpI = P-gp inhibition c Prediction outcome indicates if the evaluation outcome in pigs could (thumb up) or could not (thumb down) be correctly predicted from evaluation outcome in rats