In vivo PET Imaging of [11C]CIMBI-5, a 5-HT2AR Agonist Radiotracer in Nonhuman Primates.

PURPOSE
5-HT2AR exists in high and low affinity states. Agonist PET tracers measure binding to the active high affinity site and thus provide a functionally relevant measure of the receptor. Limited in vivo data have been reported so far for a comparison of agonist versus antagonist tracers for 5-HT2AR used as a proof of principle for measurement of high and low affinity states of this receptor. We compared the in vivo binding of [11C]CIMBI-5, a 5-HT2AR agonist, and of the antagonist [11C]M100907, in monkeys and baboons.


METHODS
[11C]CIMBI-5 and [11C]M100907 baseline PET scans were performed in anesthetized male baboons (n=2) and male vervet monkeys (n=2) with an ECAT EXACT HR+ and GE 64-slice PET/CT Discovery VCT scanners. Blocking studies were performed in vervet monkeys by pretreatment with MDL100907 (0.5 mg/kg, i.v.) 60 minutes prior to the scan. Regional distribution volumes and binding potentials were calculated for each ROI using the likelihood estimation in graphical analysis and Logan plot, with either plasma input function or reference region as input, and simplified reference tissue model approaches.


RESULTS
PET imaging of [11C]CIMBI-5 in baboons and monkeys showed the highest binding in 5-HT2AR-rich cortical regions, while the lowest binding was observed in cerebellum, consistent with the expected distribution of 5-HT2AR. Very low free fractions and rapid metabolism were observed for [11C]CIMBI-5 in baboon plasma. Binding potential values for [11C]CIMBI-5 were 25-33% lower than those for [11C]MDL100907 in the considered brain regions.


CONCLUSION
The lower binding potential of [11C]CIMBI-5 in comparison to [11C]MDL100907 is likely due to the preferential binding of the former to the high affinity site in vivo in contrast to the antagonist,  [11C]MDL100907, which binds to both high and low affinity sites.

([ 11 C]M100907 aka [ 11 C]volinanserin) and [ 18 F]altanserin have been so far the most commonly used antagonist ligands for in vivo PET studies of 5-HT2AR (17)(18)(19). Antagonist ligands bind to the high affinity (HA) and low affinity (LA) conformations of 5-HT2AR with equal affinity (10). In contrast, agonists ligands at tracer doses, as used in PET, bind preferentially to the HA state of the receptor with high affinity, which is coupled to G-protein and thereby provides a more meaningful functional measure of the 5-HT2AR (10,(20)(21)(22)(23). Sequential PET scans performed with both an agonist and antagonist 5-HT2AR tracer in the same subject may enable the quantification of binding to active or G-protein-coupled receptors (GPCR) and help estimate the ratio of coupled to uncoupled receptors (21,22). This ratio reflects the capacity of the receptor for signal transduction. The classical approach to determine relative levels of high and low affinity 5-HT2ARs is by measuring the binding maximum (Bmax) and dissociation constant (KD) using a two-site model (21,22). This can be achieved in vitro by administering the agonist radiotracer in a sufficient concentration range to determine the Bmax for both sites. However, this approach may not be feasible in vivo since it would require injection of high pharmacological doses of the radioligand, which may be prohibited in humans. As an alternative approach, we set out to determine the HA and total 5-HT2AR binding in vivo by comparing agonist and antagonist radiotracer bindings using PET imaging.

Materials
The commercial chemicals and solvents used in the synthesis were purchased from Sigma-Aldrich Chemical Co. (St. Louis, MO), Fisher Scientific Inc. (Springfield, NJ), or Lancaster (Windham, NH) and were used without further purification. MDL100907 and desmethyl-MDL100907 were purchased from ABX Advanced Biochemical Compounds. Analytical grade reagents were purchased from standard commercial sources. HPLC analyses were performed using a Waters 1525 binary HPLC system. The parent fractions and metabolites were collected from HPLC system coupled with -detector and measured using Packard Instruments Gamma Counter (Model E5005, Downers Grove, IL). [ 11 C]CO2 was produced from RDS112 cyclotron (Siemens, Knoxville, TN) or PET Trace GE cyclotron. For detection of radiolabeled products, gamma ray detector (Bioscan Flow-Count fitted with a NaI detector) was used in series with the UV detector (Waters Model 996 set at 254 nm). Data acquisition for both the analytical and preparative systems was accomplished using a Waters Empower Chromatography System. The specific activities were determined at the end of synthesis (EOS) based on the UV absorption and concentration standard curves (λ = 254 nm). PET imaging were performed in baboon using an ECAT EXACT HR+ scanner (Siemens, Knoxville, TN). All animal experiments were carried out with the approval of the Institutional Animal Care and Use Committee (IACUC) of Columbia University Medical Center, New York State Psychiatric Institute and Wake Forest University Medical Center.
Chemistry and Radiochemistry CIMBI-5 and the corresponding radiolabeling precursor were synthesized by reported procedures (24)(25)(26)(27). The radiosyntheses of [ 11 C]CIMBI-5 and [ 11 C]MDL100907 were performed by minor modifications of published methods [24][25][26][27]. Briefly, [ 11 C]CIMBI-5 was synthesized by transfer of [ 11 C]MeOTf, to a vial containing ~0.5 mg of desmethyl-N-boc protected precursor in 400 L of acetonitrile containing 5 L of 2M NaOH at room temperature. To the resulting solution, 250 L of trifluoroacetic acid: acetonitrile (1:1) was added and the mixture was heated at 80 o C for 4 min. After neutralization with 750 L of 2M NaOH, the reaction mixture was purified through a semi HPLC (Phenomenex Prodigy™ ODS-prep, 250 x 10 mm, 10 µ; 35% acetonitrile: 65 % 0.1 M ammonium formate solution in water containing 0.5% acetic acid, 10 mL/min). The product fraction based on detector was collected, diluted with 100 mL water and passed through a C-18 Sep-Pak cartridge, washed with 5 mL of 12.5 mM NaOH solution, 10 mL of water and eluted with 1 mL of ethanol. A portion of radioproduct was used for quality control studies using analytical HPLC for purity and specific activity measurements (Phenomenex Prodigy™ ODS3 250 x 4.6 mm, 5 ; 40% acetonitrile: 60% 0.1 M ammonium formate solution in water containing 0.5% acetic acid; 2 mL/min, wavelength: 254 nm). The remaining ethanol solution was diluted with 9 mL of normal saline and filtered through a 0.22 μm sterile filter into a sterile vial for further studies.
[ 11 C]MDL100907 was synthesized by trapping [ 11 C]MeOTf into a solution of desmethyl-MDL100907 (~ 0.5 mg) in acetone (400 µL) containing 5 L of 2N NaOH at room temperature. At the end of the trapping, the reaction mixture was directly injected into a semi preparative HPLC column (Phenomenex Prodigy™ ODS-prep, 250 x 10 mm, 10 µ, 25% acetonitrile: 75% 0.1 N ammonium acetate in water and 0.5% acetic acid; 10 mL/min). The product fraction based on -detector was collected, diluted with 100 mL deionized water and passed through a classic C-18 Sep-Pak cartridge. The Sep-Pak was washed with 10 mL of deionized water and the product was then eluted with 1 mL of ethanol. A small portion of the ethanol solution was analyzed by analytical HPLC (Phenomenex Prodigy™ ODS-3, 250 x 4.6 mm, 5 µ, 30% acetonitrile: 70% 0.1 N ammonium acetate in water containing 0.5% acetic acid; 2 mL/min, wavelength: 254 nm) to determine the molar activity and radiochemical purity. The remaining ethanol solution was diluted with 9 mL of normal saline and filtered through a 0.22 µm sterile filter into a sterile vial.

PET Imaging studies in monkeys and baboons
Magnetic resonance imaging (MRI) brain scans were acquired for each animal on a GE 1.5-T Signa Advantage system. Regions of interests (ROIs) were drawn on the MRI using MEDX software (Sensor Systems, Inc., Sterling, VA). PET scans were performed in two male baboons and two male vervet monkeys with an ECAT EXACT HR+ scanner (CPS/Knoxville, TN) and GE 64-slice PET/CT Discovery VCT Scanner (General Electric Medical Systems, Milwaukee, WI, USA), respectively. The fasted animals were anesthesia-inducted with ketamine (10 mg/kg i.m.) and subsequently anesthetized with 1.5-2.0% isoflurane via an endotracheal tube. Core temperature was kept constant at 37C with a heated water blanket. An intravenous infusion line with 0.9% NaCl was maintained during the experiment and used for hydration and radiotracer injection. In the case of the baboons, an arterial line was placed to collect arterial blood samples for determination of a metabolitecorrected input function. The head was positioned at the center of the field of view, and a 10 min transmission scan was performed before the tracer injection. For each scan, [ 11 C]CIMBI-5 (185 ± 18 MBq, molar activity of 74 ± 18 GBq/μmol, n=6) or [ 11 C]M100,907 (148 ± 18 MBq, specific activity of 111 ± 18 GBq/ μmol, n=2) were injected as an i.v. bolus and PET data were collected for 120 min in 3-D mode. In the baboons, arterial blood samples were taken every 10 s for the first 2 min, using an automatic system, and manually thereafter for a total of 27 samples over 120 minutes. Blocking studies were performed in vervet monkeys by pretreatment with MDL100907 (0.5 mg/kg i.v.) 60 minutes prior to the PET scan. PET images were each co-registered with the MRI using FLIRT. ROIs drawn on the animal's MRI scan were transferred to co-registered automated image registration (AIR) frames of PET data. Time activity curves (TACs) in the right and left regions were averaged into one TAC per region.

Protein binding and metabolite analyses
The methods for protein binding of [ 11 C]MDL100907 and [ 11 C]CIMBI-5 in baboon blood samples used in the experiments reported here are described elsewhere [25][26][27]38]. The percentages of unchanged radiotracers radioactivity in plasma were determined by HPLC (25)(26)(27)39). Blood samples (2 mL per time point) taken at 2, 12, 30, 60, and 90 min after radioactivity injection were considered for metabolites analyses. Briefly, the supernatant liquid obtained after centrifugation of the blood sample at 3,400 rpm for 10 min was transferred (0.5 mL) into a tube and mixed with acetonitrile (0.7 mL). The resulting mixture was vortexed for 10 s, and centrifuged at 14,000 rpm for 4 min. The supernatant liquid (~1 mL) was removed, the radioactivity was measured in a well-counter, and the majority (~0.8 mL) was subsequently injected onto the HPLC column (Phenomenex Prodigy™ 5 µm ODS-3, 250 x 4.6 mm; mobile phase: acetonitrile/ 25 mM Na2HPO4 in water, 40:60 (v/v), flow rate: 2 ml/min, retention time: 7 min) equipped with a series of radioactivity detectors. The metabolite and parent fractions collected from HPLC were analyzed using a Bioscan gamma detector. All the acquired data were then subjected to correction for background radioactivity and physical decay to calculate the percentage of the parent compound in the plasma at different time points.

Image Analysis
PET data were reconstructed with attenuation correction using the transmission data, and scatter correction was performed using model-based scatter correction (40). The reconstruction filter and estimated image filter were Shepp 0.5, the axial (Z) filter was all pass 0.4, and the zoom factor was 4.0. Final image resolution at center of field of view was 5.1 mm FWHM (41,42). For experiments where arterial blood samples were available, distribution volumes (VT), and corresponding binding potentials BPP and BPND, were calculated for each ROI using the likelihood estimation in graphical analysis (LEGA), and the Logan plot; for experiments without arterial blood samples, LEGA and Logan plot with a reference region as input, and simplified reference tissue model (SRTM), were used to calculate the binding potential BPND (43,44). Cerebellum was used as the reference region for both tracers. For experiments where arterial blood samples were available, brain TACs were corrected for vascular contribution by assuming a 5% blood volume (VB) in the ROIs before applying LEGA or Logan plot (44). VT (ml of plasma/ml of tissue) is defined as the ratio of the tracer concentration in the ROI to the metabolite-corrected plasma concentration of the tracer at equilibrium and represents the sum of the specific and nondisplaceable distribution volumes (VND). BPP refers to the ratio at equilibrium of specifically bound radioligand to that of total parent radioligand in plasma (i.e., free plus protein bound). BPP in each ROI was obtained from the VT values as BPP = VT -VND, with VND estimated using the distribution volume in the cerebellum. BPP relates to Bavail as fP*Bavail/KD, where fP is the plasma free fraction, Bavail is the density of 5-HT2AR available to bind to the radiotracer, and KD the affinity for the target of the radioligand in question. BPND (= BPP/VND) refers to the ratio at equilibrium of specifically bound radioligand to that of non-displaceable radioligand in tissue and compares the concentration of radioligand in receptor-rich to receptor-free regions (50).

PET studies of [ 11 C]CIMBI-5 in monkey
PET imaging experiments in anesthetized vervet monkeys show that [ 11 C]CIMBI-5 penetrates the blood brain barrier (BBB) and accumulates in brain ( Figure 2). Specific binding of [ 11 C]CIMBI-5 to 5-HT2AR was determined by performing blocking studies with MDL100907. The TACs for [ 11 C]CIMBI-5 in a representative monkey brain are reported in Figure 3 for both baseline and blocking scan.
Cortical regions exhibited the most uptake of the radiotracer, whereas, hippocampus showed moderate binding; caudate, putamen and cerebellum showed low uptake of [ 11 C]CIMBI-5. Slow washout of radioactivity was found in cortical regions, whereas, the TACs showed that the clearance of radiotracer was relatively faster from caudate, putamen and cerebellum. Cortex to cerebellum ratio was in the ranges of 1.7 to 1.4 at 120 minutes post injection, with and highest binding ratios were found in anterior cingulate (ACN) and medial prefrontal cortex (MED) (1.7). Hippocampus has moderate binding (HIP: CER = 1.3), whereas, putamen (1.15) and caudate (1.04) showed least binding ratios to cerebellum. Blocking studies with 5-HT2AR antagonist MDL100907 indicate a displacement of radioactivity across the brain regions ( Figure 3) where 5-HT2AR is present. Cortical regions show moderate displacement of [ 11 C]CIMBI-5 activity (~30-50%), hippocampus (25%), caudate (-5%), putamen (3.3%) and cerebellum (12%) after pretreatment with MDL100907. Hence, caudate, putamen and cerebellum showed the least displacement of radioactivity during blocking experiments.    (Figure 4 A, B and C). A high correlation of BPND was obtained for [ 11 C]CIMBI-5 baseline and MDL100907 blocking studies between Logan plot and LEGA methods (Figure 4 D). High variations of BPND were found with the SRTM method ( Figure 3D). BPNDs obtained for blocking studies for cortical regions such as DOR, OCC, ORB, PFC, PAR and hippocampus were not in agreement with the distribution of radiotracer based on TAC method. PET studies of [ 11 C]CIMBI-5 in baboon PET imaging experiments in baboons confirm that [ 11 C]CIMBI-5 penetrates the BBB and accumulates in brain ( Figure 5). TACs show that the radiotracer is preferentially retained in 5-HT2AR rich brain regions ( Figure 6). Cortical regions show the highest radioligand binding, whereas putamen and cerebellum show the lowest binding. The radioactivity level peaked around 40 min post injection, and target to cerebellar radioactivity ratios at 120 minutes were ~1.5 for most cortical regions ( Figure 6). Insular and occipital cortex showed radioactivity ratios of 1.85 and 1.65 with respect to cerebellum at 120 minutes. Slow washout of [ 11 C]CIMBI-5 was observed in cortical regions, where higher density of 5-HT2AR is present. Relatively faster washout was observed in cerebellum, caudate and putamen, which are regions with comparatively low density of 5-HT2AR. The free fraction of radioligand in plasma was 1-2% (N=2) as determined using the ultracentrifuge method. Fast metabolism and polar metabolites were observed for [ 11 C]CIMBI-5 in baboon plasma. Percentages of unmetabolized parent radioligand were 81% at 2 min, 45% at 12 min, 20% at 30 min, 9% at 60 min and 3% at 90 minute post injection, respectively (Figure 7). The free fraction and percentage of unmetabolized [ 11 C]CIMBI-5 in baboon plasma determined here are in agreement with the values reported for pigs [24,25]. Subsequently, we compared the binding parameters of [ 11 C]CIMBI-5 and [ 11 C]MDL100907 in baboon brain ( Figure 8) and the results indicate that VT, BPP, and BPND of [ 11 C]MDL100907 are higher than that of [ 11 C]CIMBI-5, presumably because the antagonist ligand [ 11 C]MDL100907 binds to both high and low affinity states of 5-HT2AR, whereas [ 11 C]CIMBI-5 binds predominantly to the high agonist affinity state of the receptor (Figure 8). A markedly lower BPP (25%) and BPND (33%) were observed for [ 11 C]CIMBI-5 scans in comparison with the corresponding [ 11 C]MDL100907 data for 5-HT2AR rich regions in baboon brain. Highest changes of BPND were found for prefrontal cortex (56.1%), parahippocampal gyrus (PIP) (59.7%) and parietal cortex (61.7%) with corresponding BPP differences of 66.6%, 69.3% and 70.8% respectively. Caudate, putamen and thalamus showed low BPP or BPND with [ 11 C]CIMBI-5 binding vs [ 11 C]MDL100907 binding, may due to low HA 5-HT2AR population in these regions (Figure 8).

DISCUSSION
Here we report the evaluation of [ 11 C]CIMBI-5, the first 5-HT2AR agonist PET ligand, in ververt monkey, as well as the comparison of its binding potential to that of the 5-HT2AR antagonist PET ligand [ 11 C]M100907 in baboons. The highest and lowest uptake of [ 11 C]CIMBI-5 was observed in cortical regions and in the cerebellum, respectively ( Figure 2). The pattern of [ 11 C]CIMBI-5 retention in monkey brain matches the expected distribution of 5-HT2AR in brain as well as the distribution reported for the same ligand in pig brain (25). However, washout of the radiotracer in monkey brain was slower than that reported in pig. Although the blocking agents used are different, [ 11 C]CIMBI-5 exhibits similar blockade effect in monkey and pig brain, with the highest specificity in cortex and the lowest in cerebellum and striatum (25). Metabolite analyses of [ 11 C]CIMBI-5 in pig and baboon indicate less lipophilic (polar than the parent CIMBI-5) metabolite. However, ex vivo brain homogenate assays in pigs reveal the absence of radioactive metabolite entering the brain, indicating that the sole brain radioactivity in brain is due to parent radioligand. Both Logan plot and LEGA methods show high test retest and test-block reliability for CIMBI-5 in brain in comparison to SRTM. It appears that there is only partial blockade of the 5-HT2ARs with M100907 ( Figure 2) and the rate of radiotracer washout from the brain was not significantly affected by the blocking agent even though the absolute uptake was lower (Figure 3). The washout of the radiotracer from the assumed reference region (cerebellum) is also very slow in monkey scans, suggesting the potential presence of off target binding of the radiotracer in the brain. Since 5-HT2BRs are less abundant in brain, the most likely off target for CIMBI-5 is 5-HT2CR (Ki= 7 nM) (24).  The observed distribution of [ 11 C]CIMBI-5 in baboon brain is in agreement with that in vervet monkey and in danish landrace pigs [24,25]. The washout out of the radiotracer in baboon is similar to the one in pig and is faster than the one in monkey (24,25). This is probably due to the different metabolic stability of radiotracer across species. The lower binding potential of [ 11 (34). Our studies suggest that [ 11 C]CIMBI-5 binding in brain is comparable to that of [ 11 C]CIMBI-36, and therefore [ 11 C]CIMBI-5 can be useful for occupancy measurement of 5-HT2AR. However, both ligands exhibit slow kinetics, limited blocking effect and presence of brain penetrating radiometabolites. Different tracers have different characteristics, such as fP, free fraction of radioligand in the non-displaceable compartment (fND), KD, completion with endogenous 5-HT, brain uptake and washout, interactions with specific or non-specific sites, and so on. Therefore, the observed difference in BPP and BPND values of [ 11 C]CIMBI-5 and [ 11 C]MDL100907 could in part arise from the above differences. Although the results reported here show, as a proof of concept, that in vivo high affinity agonist site can be imaged using [ 11 C]CIMBI-5 and [ 11 C]CIMBI-36, the development of 5-HT2AR agonist PET tracers with improved binding characteristics remains an important goal.

CONCLUSION
We have shown lower binding of 5-HT2AR agonist tracer compared to 5-HT2A antagonist tracer in nonhuman primates as measured in vivo by PET. We observed target specific distribution of [ 11 C]CIMBI-5 to 5-HT2AR in vervet monkey and baboon brains, consistent with the known distribution of 5-HT2AR by autoradiography. In general, a markedly lower binding potential is observed for the agonist [ 11 C]CIMBI-5 in comparison with the antagonist [ 11 C]MDL100907 in 5-HT2AR rich regions in baboons. The PET imaging data reported here indicate that [ 11 C]CIMBI-5 behaves as a high affinity 5-HT2AR agonist tracer in baboon and monkey. The lower binding of the agonist [ 11 C]CIMBI-5 binding in baboon compared to the antagonist [ 11 C]M100907 in this study is consistent with the reported ratio of high affinity site binding of agonists and antagonists by in vitro studies, but could also be ascribed to the difference in the free fractions, endogenous completion with 5-HT, and in vivo KD of the radiotracers. Studies comparing in vivo Bmax via Scatchard analyses or endogenous changes of 5-HT2AR after pharmacological stimulation would provide further validation of the percentage of HA binding of [ 11 C]CIMBI-5.