Effects of the sigma-1 receptor agonist blarcamesine in a murine model of fragile X syndrome: neurobehavioral phenotypes and receptor occupancy

Fragile X syndrome (FXS), a disorder of synaptic development and function, is the most prevalent genetic form of intellectual disability and autism spectrum disorder. FXS mouse models display clinically-relevant phenotypes, such as increased anxiety and hyperactivity. Despite their availability, so far advances in drug development have not yielded new treatments. Therefore, testing novel drugs that can ameliorate FXS’ cognitive and behavioral impairments is imperative. ANAVEX2-73 (blarcamesine) is a sigma-1 receptor (S1R) agonist with a strong safety record and preliminary efficacy evidence in patients with Alzheimer’s disease and Rett syndrome, other synaptic neurodegenerative and neurodevelopmental disorders. S1R’s role in calcium homeostasis and mitochondrial function, cellular functions related to synaptic function, makes blarcamesine a potential drug candidate for FXS. Administration of blarcamesine in 2-month-old FXS and wild type mice for 2 weeks led to normalization in two key neurobehavioral phenotypes: open field test (hyperactivity) and contextual fear conditioning (associative learning). Furthermore, there was improvement in marble-burying (anxiety, perseverative behavior). It also restored levels of BDNF, a converging point of many synaptic regulators, in the hippocampus. Positron emission tomography (PET) and ex vivo autoradiographic studies, using the highly selective S1R PET ligand [18F]FTC-146, demonstrated the drug’s dose-dependent receptor occupancy. Subsequent analyses also showed a wide but variable brain regional distribution of S1Rs, which was preserved in FXS mice. Altogether, these neurobehavioral, biochemical, and imaging data demonstrates doses that yield measurable receptor occupancy are effective for improving the synaptic and behavioral phenotype in FXS mice. The present findings support the viability of S1R as a therapeutic target in FXS, and the clinical potential of blarcamesine in FXS and other neurodevelopmental disorders.

[ 18 F]FTC-146 47 . The incorporation of [ 18 F]FTC-146 PET imaging into this study provides insight as to how blarcamesine and the S1R receptor interact in vivo. Taken together, the behavioral and biochemical results, supported by the S1R occupancy profile, demonstrate that blarcamesine is a potentially valuable therapeutic approach for patients with FXS.

Results
Blarcamesine substantially improved the behavioral phenotype of Fmr1 KO2 mice. Comparisons between Fmr1 KO2 groups demonstrated a significant reduction in total distance traveled (number of squares crossed in 3 min) by the blarcamesine-treated animals with respect to vehicle-treated mice (Student's t-test p = 0.0006). The relevance of these changes was confirmed by comparing vehicle-treated Fmr1 KO2 mice to their wild type (WT) counterparts, since the former displayed an increase in the abovementioned measure of hyperactivity (p < 0.001). When all 4 mouse groups were contrasted, chronic treatment with blarcamesine significantly reduced the behavior in Fmr1 KO2 mice to levels indistinguishable from those observed in vehicletreated WT mice (Fig. 1a). Since tests for equality of variances showed trend-level p-values (i.e., borderline equal variance among groups), nonparametric Kruskal-Wallis analysis of variance (ANOVA) and Wilcoxon rank sum posthoc tests were also performed. These confirmed the parametric ANOVA and posthoc test results.
In the contextual fear conditioning paradigm, similar differences between Fmr1 KO2 groups were also found. There was a significant increase in percentage of freezing behavior in drug-treated Fmr1 KO2 mice when compared with their vehicle-treated counterparts (Student's t-test p < 0.0001). Again, the relevance of this effect was demonstrated by comparing vehicle-treated Fmr1 KO2 mice to their WT counterparts, since the mutant mice displayed a significant decrease in freezing response (p < 0.001). The four-group analyses showed that the  www.nature.com/scientificreports/ improvements in the blarcamesine-treated Fmr1 KO2 mice were at the phenotype rescue level (i.e., no differences between drug-treated Fmr1 KO2 mice and vehicle-treated WT animals) (Fig. 1b). A species-specific behavior, which represents anxiety and perseverative behavior 48,49 , marble-burying activity was mildly increased (improved) by drug administration when vehicle-and blarcamesine-treated Fmr1 KO2 groups were compared (Welch's t-test p = 0.025). The 4-group comparison confirmed that vehicle-treated Fmr1 KO2 mice buried significantly fewer marbles than similarly treated WT mice (p < 0.001). As in the paired t-test, the ANOVA's posthoc test demonstrated that blarcamesine partially rescued this behavior in Fmr1 KO2 animals (p ≤ 0.05) (Fig. 1c). Data for the WT were not normally distributed (leptokurtic), compatible with a ceiling effect. In line with this, tests for equality of variances showed borderline (trend level) comparable variances. Therefore, nonparametric tests were also performed. They confirmed the abovementioned parametric test results.
Blarcamesine restores BDNF levels in the hippocampus of Fmr1 KO2 mice. In order to yield insight into the mechanisms underlying the behavioral improvements described above, we examined the levels of several key cell signaling markers in the hippocampus. As for the behavioral paradigms, we first analyzed the effects of blarcamesine by comparing vehicle-and drug-treated Fmr1 KO2 mice.
As previously mentioned, BDNF constitutes a convergence point of PI3K/Akt/mTOR and MAPK/ERK pathways and other regulatory pathways 31,50,51 . BDNF levels in hippocampal homogenates in vehicle-treated Fmr1 KO2 mice were significantly lower than in vehicle-treated WT animals (Welch's t-test p < 0.05); comparisons between vehicle-and drug-treated Fmr1 KO2 mice showed significantly higher BDNF levels in hippocampi after blarcamesine administration (Welch's t-test p = 0.0008). ANOVA analyses based on 6 animals per group confirmed these findings, by demonstrating that vehicle-treated KO2 mice had markedly lower levels of BDNF than vehicle-treated WT mice and they returned to WT levels after drug treatment (Fig. 2c). www.nature.com/scientificreports/ Blarcamesine yielded dose-dependent increases in S1R occupancy. Additional insight into the mechanisms by which blarcamesine improved key behavioral phenotypes and signaling pathway markers was obtained by analyzing the drug's effects on S1R occupancy.
PET/CT scanning. A two-tissue compartment model (2TCM) was successfully fitted to the 60-min dynamic mouse brain data in order to calculate the k3/k4 macro parameter. Data points were weighted by frame duration and blood volume was calculated for each animal. The whole blood image-derived arterial input function (IDIF) was described by a 3-exponential model and the chi value was used to determine best fit. The analyses of [ 18 F]FTC-146 metabolism in 7-week-old WT mice showed that 16% of counts in plasma could be attributed to the parent radioligand 5 min post-dose (Fig. S1); the data also suggested an average fixed correction of 1:1.14 for whole blood counts to plasma. Model parameters and receptor occupancy calculations are summarized in Table 1. Specifically, Table 1 depicts comparable 2-tissue model parameters, receptor occupancy (RO) and percent of injected dose per gram of tissue (%ID/g) calculations for WT and Fmr1 KO mice. In these studies, blarcamesine oral (PO) dosing was 1 mg per kilogram of weight (mg/kg), 10 mg/kg, and 30 mg/kg, while S1R blocking with PRE-084 S1R was at a 1 mg/kg PO dose. No significant differences were found between blarcamesine's RO in WT mice and KO mice. In Table 1, values are expressed as mean ± standard deviation, with the exception of % RO. Calculations of % RO are described in "Materials and methods" section. S1R occupancy increased proportionally to the blarcamesine dose, with a plateau of 64% ± 9% at 30 mg/kg PO in WT mice and 64% ± 12% in Fmr1 KO mice for the whole brain (Fig. 3, Table 1). When comparing k3/k4 parameter values in the absence of blarcamesine, no significant differences were observed between Fmr1 KO and WT mice (0.88 ± 0.06 vs. 1.02 ± 0.17), suggesting that there were no discernable differences in the number of S1Rs present in the whole brain. For all outcome measures in the blarcamesine and control groups (k3/k4, total volume, volume of specific binding, % receptor occupancy and %ID/g), there was a significant decreasing trend across dose levels (all p < 0.001), but no effect of genotype (%ID/g p = 0.47, all other parameters p = 0.23). This data supports the notion that blarcamesine binds to the S1R receptor and that its occupancy can be imaged using the highly selecting S1R radiotracer, [ 18 F]FTC-146. Moreover, at each dose, there was no effect of genotype on any calculated measure. This indicates that the amount of S1R present in Fmr1 KO mice does not differ from WT when measured with [ 18 F]FTC-146 PET imaging. PRE-084, an established reference S1R agonist 36 , was also administered PO at 1 mg/kg showing no significant difference to blarcamesine on any measured outcome (all p = 0.77) except with %ID/g, measured from 30 to 40 min, which showed significantly better blocking by blarcamesine (p = 0.029) ( Table 1, Fig. 4).

Ex vivo autoradiography.
In order to further examine the uptake of [ 18 F]FTC-146 in various brain regions at higher resolution, ex vivo autoradiography (ARG) was performed immediately after PET imaging in the frontal cortex, caudate, hippocampus, thalamus, amygdala, pons, and cerebellum. [ 18 F]FTC-146 binding levels via ex vivo ARG varied significantly among brain structures (p < 0.001), particularly in the control and 1 mg/kg dose groups. There was a significant decreasing dose effect (p < 0.001), but no significant genotype effect (p = 0.15) (Fig. 5, Supplementary Fig. S2). Dosing at 1 mg/kg PO of either blarcamesine or PRE-084, a reference S1R agonist, exhibited no significant difference in S1R blocking in WT mice (p = 0.13) (Fig. 6).

Discussion
Fragile X syndrome (FXS) is the most prevalent genetic form of intellectual disability and autism spectrum disorder 1,2 . As most neurodevelopmental disorders, FXS is considered a disorder of synaptic development and function 1,9 . Mouse models of FXS display a variety of cognitive and behavioral impairments of clinical www.nature.com/scientificreports/   www.nature.com/scientificreports/ relevance 9,40 . Despite advances in drug development using these experimental models, no pivotal trial in FXS has thus far been successful; therefore, this neurodevelopment disorder continues to have an unmet therapeutic need. Administration of blarcamesine to Fmr1 KO2 mice for two weeks led to correction of two key neurobehavioral phenotypes and marked improvement of a third one. Moreover, a major neuronal signaling abnormality in mouse models of FXS, namely decreased BDNF levels, was restored to WT levels in the hippocampus of Fmr1 KO2 mice. Since BDNF is a converging point of several synaptic regulators disrupted in FXS, these findings suggest that blarcamesine corrects Fmr1 KO2 mouse behavioral phenotypes through multiple synaptic signaling mechanisms known to be affected by FMRP deficiency 9,31 . These dramatic effects of blarcamesine upon the Fmr1 KO mouse phenotype can be explained by the drug's dose-dependent level of S1R occupancy and the wide but variable brain regional distribution of these receptors 19 , which are not affected by Fmr1 mutation, as demonstrated by the imaging component of the present study. Fmr1 mouse models of FXS are very informative since they replicate a wide range of molecular, anatomical, physiological, cognitive, and behavioral features of the disorder 9,40,52 . The tested behavioral paradigms covered three key aspects of FXS with implications for other neurodevelopmental disorders: cognitive impairment, ADHD features, and anxious and perseverative behaviors 1,2 . The fact that our investigation involved chronic administration, as opposed to acute dosing as in previous studies 53,54 , provides additional evidence in favor of the clinical use of blarcamesine. Since the behavioral paradigms reflect the involvement of multiple cortical and subcortical regions, their marked improvement by blarcamesine suggest widespread activation of S1Rs by the drug and modulation of multiple neural pathways. Indeed, the observation of normalization of BDNF levels after blarcamesine administration, in a brain region critical for cognition and behavior, is also a finding with important implications for FXS and other synaptic disorders 30 . BDNF is point of bidirectional convergence of several signaling pathways implicated in FXS pathogenesis, including PKA, PI3K/Akt/mTOR and MAPK/ERK, and plays key roles in neuronal and synaptic development as well as in the maintenance of circuitry 31,51 . As BDNF levels normalize without changes in pGSK-3β and Rac1 at the tested blarcamesine dose, these results suggest relative specificity in drug effects since the latter pathways are not directly linked to BDNF 31 . Experimental data suggests that FMRP and BDNF regulate each other and levels of BDNF modulate the FXS phenotype 31 . The recent demonstration of blarcamesine's effect of enhancing autophagy in both animal and cellular models 16 suggests that the drug may also play a role in correcting impaired autophagy and protein homeostasis in FXS 9 . Fmr1 KO mice show multiple abnormalities in synaptic plasticity 9 , including compensatory homeostatic synaptic scaling 8 . The latter is a fundamental synaptic plasticity process that has been corrected, by administration of S1R agonists, in mouse models of other neurologic disorders 18 and likely one of the mechanisms of blarcamesine's action, in addition to stabilization of cellular bioenergetic and fostering neuronal homeostasis, all representing the wide range of functions of S1Rs [13][14][15] .
In the evaluation of novel compounds presumed to be centrally active, such as blarcamesine, it is necessary to confirm that the drug engages the targeted receptor within the central nervous system (CNS). PET imaging studies are useful tools in this effort because they provide a means to (1) demonstrate that the drug crosses the blood-brain barrier and (2) calculate the percentage of receptors that are occupied in vivo over time. Based on the results of the PET and ex vivo ARG analyses in the present study, S1R occupancy by blarcamesine was observed to be dose-dependent. It is shown that target engagement of S1Rs with blarcamesine is achieved already at relatively low doses with similar receptor occupancy profile to widely-used agonist PRE-084 at 1 mg/kg (Figs. 4, 6), thus providing a broader therapeutic window for S1R activation by this drug. Also, comparable S1R levels in Fmr1 KO and WT mice support the notion of S1R preservation and the therapeutic potential of modulators of this receptor in FXS. The observed saturation of target occupancy before reaching 100% is likely due to the nature of the modulatory binding to the S1R by blarcamesine. A clinically-relevant component to this evaluation of receptor occupancy was evaluating blarcamesine's oral administration used in the imaging studies, which demonstrated target engagement when administered via the route intended for human use. All PET scans were 60-min dynamic scans except for one scan which terminated at 55-min due to scanner error, but attenuation was not affected. In addition to demonstrating the adequacy of blarcamesine as a S1R agonist in the CNS through [ 18 F]FTC-146 imaging, ex vivo ARG of multiple brain regions with the radioligand confirmed the wide distribution and regionspecific binding of S1Rs. It also showed that there is no difference, at any blarcamesine dose, in regional S1R binding between WT and Fmr1 KO mice. Since there is no reference region for the [ 18 F]FTC-146, the overall mean of all structures was used as the reference region. In the post-scan ex vivo ARG, we observed that while no significant differences were observed between genotype groups from 0 to 10 mg doses, the WT group appears to begin to exhibit higher blocking than the Fmr1 KO group at the highest blarcamesine dose. Perhaps due to limit of detection within our sample size, no significance was found; however, this possible genotype difference in the highest dose group is worth investigating in a larger sample size in the future. Another noticeable trend is that as the dose of blarcamesine increases, the differences between [ 18 F]FTC-146 uptake in each structure decreases. This observation can be attributed to decreased S1R availability for radioligand binding due to an increase of receptor occupancy by blarcamesine in each structure. In Table 1, it is also notable that the 10 mg/kg Fmr1 KO group has the highest standard deviation for %RO calculations. This may be due to the Fmr1 KO mice requiring a higher dose to reach saturation compared to the WT mice, which are near saturation at 10 mg/kg. Receptor binding as a function of ligand concentration can be described by the Hill-Langmuir equation 55 , which reflects the degree of cooperativity between ligands. This equation describes the receptor occupancy range from 0% to saturation and includes an exponential phase within that range. The Fmr1 KO mice may be in the exponential phase at 10 mg/kg, therefore, small variations in effective ligand concentration will thus lead to large variations in occupancy, as reflected in the data. At 30 mg/kg, once saturation is reached, the datapoints for both groups of mice are very similar. As mentioned in our methods section, for the blarcamesine-dosed groups, one animal from each dose-group were scanned at a time, for a total of 4 animals per scan, to uphold the scientific rigor of our scanning. This design allows the measurement of animals from each dosing group to be taken on multiple www.nature.com/scientificreports/ scan days over the course of several weeks accounting for variations to be included between scans, radiotracer productions and different animal cages. Thus, the implemented design resulted in higher standard deviations due to accumulated data for each cohort being collected over multiple scans, rather than a single scan. Similarly, in Table 1 and Fig. 4, a larger standard deviation for the %ID/g values was observed in the 1 mg/kg blarcamesine WT dose group (animals were separately imaged over several weeks) when compared to the 1 mg/kg PRE-084 WT dosed group (animals were imaged in the same scan). Overall, through all parameters examined, our findings further support the notion that, while the measured amount of [ 18 F]FTC-146 bound to S1Rs changes throughout the WT brain, there were no observed changes between WT and Fmr1 KO mice. Altogether, these neurobehavioral, biochemical, and imaging data demonstrate that corresponding doses of blarcamesine that yield measurable receptor occupancy are effective for substantially correcting key synaptic and behavioral phenotypes in Fmr1 KO mice. Our data also suggest that these positive effects are mediated by S1R activation in multiple brain regions, where blarcamesine binds to the receptor in a dose-dependent and genotype-independent manner. Limitations of the present study include the relatively small number of behavioral paradigms examined, some with limited range of responses; the range and duration of drug dosing; evaluation of BDNF and other signaling molecules in a single brain region; and the use of two different Fmr1 KO models in the neurobehavioral and imaging studies. Follow up dose-response studies can be performed to determine the point of saturation in the Fmr1 KO mice and examine the effects of different doses on positive responses and their variability in multiple brain regions relevant to the FXS phenotype (e.g., neocortex). Endpoints in these investigations ought to include indices of activation of BDNF-related signaling pathways known to be affected in FXS and Fmr1 mouse models. A shortcoming of using mice in the imaging studies was the need of a separate cohort of mice to the one subjected to PET imaging to determine both plasma:whole blood ratio and metabolism of [ 18 F]FTC-146 over time. This was done because, in order to obtain metabolic profiles from the same mice being scanned, a large amount of blood (~ 25-35% of total blood volume) would have been needed with the resulting perturbation in animal physiology. For these studies, WT mice were used because there was no expected change in metabolism between WT and Fmr1 KO mice. Furthermore, a separate group of WT and Fmr1 KO mice was used as the control for both the PET and autoradiography studies, rather than imaging a baseline in each mouse.
In conclusion, the present findings confirm the dose-dependent receptor occupancy of the S1R with blarcamesine and, combined with the therapeutic response observed at low doses in the tested preclinical model, emphasize the viability of S1R as a therapeutic target in FXS and the clinical potential of blarcamesine in FXS and other neurological disorders. Indeed, pre-clinical studies in a mouse model of Rett syndrome showed similar positive effects on multiple clinically relevant neurobehavioral phenotypes 21 . Furthermore, clinical efficacy was demonstrated in a placebo-controlled Phase 2 study in Rett syndrome (NCT03758924) and previously in a smaller PK cohort of patients with this neurodevelopmental disorder 39 , as well as significant cognitive improvements in a Phase 2 trial in Parkinson's disease dementia (NCT03774459). Late-stage clinical studies of blarcamesine in adult and pediatric patients with Rett syndrome (NCT03941444, NCT04304482) and Alzheimer's disease (NCT02756858, NCT03790709) are currently ongoing. Continued findings from these clinical studies with blarcamesine, combined with the presented data strengthens the rationale for potentially a dependable and effective treatment strategy for FXS and other neurological disorders targeting the S1R with blarcamesine.

Materials and methods
The study was divided into two parts, the first of which involved behavioral and biochemical assays to characterize the effect of blarcamesine in reversing the murine FXS phenotype. The second component, also carried out in mice, focused on determining the drug's S1R receptor occupancy by PET and S1R distribution by ex vivo autoradiography. All animal studies were performed in accordance to Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines. The behavioral and biochemical assay studies were carried out in accordance with the guidelines and regulations of the United Kingdom Animals (Scientific Procedures) Act of 1986. The PET imaging and receptor occupancy studies were carried out in accordance with the guidelines and regulations of Stanford University's Institutional Animal Care and Use Committee (IACUC).
Behavioral and cell signaling analyses. Animals. For both the behavioral and biochemical assessments, experiments were conducted in accordance with the United Kingdom Animals (Scientific Procedures) Act of 1986. Fmr1 KO2 mice 40 and wild type (WT) littermates, which were generated on a C57BL/6J background and repeatedly backcrossed onto a C57BL/6J background for more than eight generations, were provided by Professor David Nelson (Baylor College of Medicine, Houston, TX, USA) and the FRAXA Research Foundation. Mice were housed in commercial plastic cages on a ventilated rack system without enrichment, in groups (4-6 per cage). All animals were provided with ad libitum food and water and maintained on a 12 h light/dark cycle in a temperature-controlled environment (21 ± 1 °C). All studies were conducted on male mice. In contrast with the Fmr1 KO mouse 41 , the first murine model of FXS, the more recently developed Fmr1 KO2 mouse 40 , is characterized by no expression of Fmr1 mRNA (the Fmr1 KO mouse expresses up to 27% of WT brain Fmr1 mRNA levels 40 ). Both mouse models do not express FMRP and show no substantial phenotypical differences 56 .
Drug treatment. Blarcamesine was administered to 2-month-old animals twice daily at a dose of 1 mg/kg IP for a total of 14 days. Saline served as vehicle and control. For behavioral studies, four dose groups (N = 10 mice per group, 40 mice total) were included: 2 groups of WT mice given either blarcamesine or saline and 2 groups of Fmr1 KO2 mice given either blarcamesine or saline. For the cellular assays assay, four dose groups (pGSK-3β: N=10 mice per group, Rac1: N=10 mice per group, BDNF: N=6 mice per group, BDNF:N=6 mice per group pERK: N = 7, 28 mice total, BDNF: N = 6 mice per group, 24 total) were included: 2 groups of WT mice given either blarcamesine or saline and 2 groups of Fmr1 KO2 mice given either blarcamesine or saline. Animals www.nature.com/scientificreports/ were inspected for changes in general appearance that might occur following a single dose, prior to the onset of chronic dosing. Items monitored in these tolerability assessments included coat appearance, piloerection, eye conditions (runny eyes or porphyria, ptosis), gait, tremor, tail tone, and reactivity to handling.
Behavior. Behavioral testing was conducted during the light phase at 2 months of age, with experimenter's blind to genotype and drug treatment. Mice were tested with one of three behavioral tasks (open field, contextual fear conditioning, or marble burying) on each experimental day; each behavioral test was separated by 3 days. Prior to behavioral testing, mice were randomly assigned to treatment groups. Apparatuses were cleaned with moist and dry tissues before testing each mouse, in order to create a low but constant background mouse odor for all experimental subjects. Behavioral tests served to characterize efficacy-related key endpoints of relevance to the FXS phenotype and performed as previously published 1,2,57,58 (see Supplementary Materials for further details): Open field test 58 (anxiety, hyperactivity, habituation to a novel environment), contextual fear conditioning 58 (associative learning), and marble-burying 48 (anxiety, perseverative behavior).
Cell signaling analyses. Western blot and ELISA assays for measuring (1) activated glycogen synthase kinase 3 beta (pGSK-3β), (2) Ras-related C3 botulinum toxin substrate 1 (Rac1) expression, and (3) brain-derived neurotrophic factor (BDNF) expression was conducted in hippocampal homogenates. This brain region was selected because of its role in the abovementioned behavioral paradigms and other key phenotypes in FXS mouse models 59 . Hippocampi were collected from mice sacrificed by CO 2 followed by cervical dislocation. Samples were frozen on dry ice and stored at − 70 °C until use. The aforementioned assays were performed as previously published 50  General. Unless stated otherwise, all compounds and chemicals were purchased from commercial sources and used without modification. PET imaging was performed using a micro-PET/computed tomography (CT) or D-PET equipped with cobalt point source (Inveon; Siemens Medical Solutions Inc, Tarrytown, NY, USA). Attenuation correction was applied to each dataset from the CT or cobalt transmission images. Frames were reconstructed using three-dimension ordered-subset expectation maximization (3DOSEM Data analysis. PET images were analyzed by drawing 3-dimensional regions around the whole brain. A twotissue compartment (2TCM) model was used to fit the measured time activity curve (TAC) for brain using PMOD software version 3.7 (PMOD Technologies LLC, Zurich, Switzerland) and to calculate k3/k4 also known as the binding potential for the specific binding at equilibrium in relation to non-displaceable binding (BP ND ) 61,62 , which was used as binding potential (BP) for the subsequent receptor occupancy calculations, as previously described 62 . To obtain the arterial whole blood input function, an imaged-derived input function (IDIF) was determined by drawing the volume of interest over the left heart ventricle representing the highest pool of radiotracer in the blood. Both the plasma:whole blood ratio and the % intact parent [ 18 F]FTC-146 over time were incorporated into the PMOD software to generate the 2TCM model to estimate BP, represented as k3/k4, and was used for receptor occupancy calculations. Calculation of the percent of injected dose per gram (%ID/g) was also used to assess target engagement of blarcamesine. For %ID/g calculations, a time period of 30-40 min was examined. To analyze ex vivo autoradiography images, ImageJ 1.48v 63 was used to define regions of interest and all structures were normalized to muscle. Three samples for each region of interest was collected from each mouse and averaged. These studies were not blinded as one person performed all dosing, scanning, PET and ex vivo ARG image analysis.
Statistical analysis. WT (N = 18) and Fmr1 KO (N = 17) mice in groups at each of four blarcamesine dosage levels (0, 1, 10, 30 mg/kg) were scanned via PET/CT or D-PET. Using 2TCM as described above, the following parameters were calculated: binding potential (k3/k4), specific volume (V s ) bound, total volume bound (V t ). An additional four WT mice were given 1 mg/kg PRE-084 were analyzed separately. Due to the between-subjects design, receptor occupancy at dose d, defined as (BP(0) − BP(d))/BP(0) × 100 could not be calculated per-individual. As an approximation, the median value of BP(0) for type of animal was used and any occupancy value ≤ 0 was replaced with the lowest observed positive value.
The effect of dose was tested with a nonparametric linear-by-linear association test of trend stratified by genotype; the effect of genotype was tested by a Wilcoxon rank sum (Mann-Whitney) test stratified by dose. Comparison of drugs was also done by the Wilcoxon rank sum test.
Binding of [ 18 F]FTC-146 was evaluated via %ID/g from 30 to 40 min post-injection for each dose of blarcamesine or PRE-084. The effects of genotype and blarcamesine dose on %ID/g were tested with a regression of %ID/g on genotype and dose. The effect of drug type when comparing blarcamesine and PRE-084 on %ID/g (in 4 WT animals per each drug at dose 1 mg/kg) was tested with an exact Wilcoxon test.
To assess binding of [ 18 F]FTC-146 varying concentrations of blarcamesine or PRE-084 in post-scan ex vivo autoradiography (ARG), the regions of interest were hand-drawn in Image J, in triplicate for each region (3 slices per region). The mean pixel intensity values for each brain region (frontal cortex, caudate, hippocampus, thalamus, amygdala cortex, pons and cerebellum) were averaged. The effects of genotype, blarcamesine dose and brain structure on ARG were tested with a generalized linear regression with log link of ARG on genotype, dose and location, adjusted for clustering within animal. Since there is no established reference brain region for [ 18 F]FTC-146, the overall mean across the structures of interest (adjusted for dose and type) was used as the reference value to compare tracer uptake among these structures.
The effect of drug type when comparing blarcamesine and PRE-084 via ARG (in 4 WT animals per each drug at dose 1 mg/kg) was tested with van Elteren's test (stratified Wilcoxon rank sum test) using neural structures as strata.
Statistical analyses were done using R version 3.6.3 and package "coin" version 1.3-1. For all tests, a p-value less than 0.05 was considered statistically significant. Given the small sample and exploratory nature of this study, no correction for multiple testing was done.