Targeting dysregulated CB1 receptors in a Down syndrome mouse model improves neurological outcomes

Abstract

INTRODUCTION

Down syndrome (DS) is the most common genetic cause of intellectual disability, affecting cognitive function and increasing the risk of early‐onset Alzheimer's disease (AD). The endocannabinoid system may serve as a therapeutic target for cognitive deficits by inhibiting cannabinoid type‐1 receptor (CB1R) function.

METHODS

CB1R expression was analyzed in the hippocampi of aged DS‐associated AD (DSAD) individuals and middle‐aged Ts65Dn mice. Long‐term oral treatment with the CB1R antagonist rimonabant was used to assess its effects on memory and neuroinflammation in the Ts65Dn mouse model of DS.

RESULTS

CB1R expression was significantly increased in both aged DSAD subjects (specifically in the dentate gyrus and CA2 posterior hippocampal subregions) and Ts65Dn mice. Long‐term rimonabant treatment improved memory performance, normalized microglial morphology, and reduced plasma inflammatory markers in trisomic mice without preventing neuron decline.

DISCUSSION

These findings suggest that sustained CB1R inhibition may enhance cognitive function by modulating neuroinflammation, highlighting its therapeutic potential for cognitive impairments in DS.

Highlights

• Cannabinoid type‐1 receptor (CB1R) expression is increased in the posterior hippocampus of aged Down syndrome (DS) subjects and Ts65Dn mice.

• Long‐term rimonabant treatment enhances memory in middle‐aged Ts65Dn mice.

• CB1R inhibition shifts neuroinflammatory features in Ts65Dn mice.

• CB1R inhibition does not halt noradrenergic/cholinergic neurodegeneration in Ts65Dn.

• CB1R inhibition presents potential for memory improvement in DS‐related deficits.

1. BACKGROUND

Down syndrome (DS) is the most significant genetic cause of intellectual disability, affecting ≈ 1 in 700 to 1000 live births worldwide. ¹ It is produced by trisomy of human chromosome 21 (HSA21). ² Intellectual disability, the most prevalent feature of DS, is an important limitation for people with DS especially affecting hippocampus‐related cognitive domains. ³ The life expectancy for people with DS has markedly increased because of medical advances reaching to a median age at death of almost 60 years. ⁴ As a result, nowadays the DS population is more prone to suffer from age‐related comorbidities throughout their lives, such as an early‐onset Alzheimer's disease (AD), which is present in almost all adults with DS by age 65. ⁵ This condition is often referred to as DS‐associated AD (DSAD). At the cellular level, aging subjects with DS show decreases in the density of noradrenergic neurons in the locus coeruleus, the main noradrenergic center in the brain ⁶ and in cholinergic neurons in the basal forebrain ⁷ , ⁸ as well as neuroinflammation, ⁹ all associated with the development of AD pathology. ¹⁰

Several animal models for DS have been established as a tool for investigating this condition based on the fact that HSA21 is orthologous to three distinct regions in murine chromosomes 10, 16, and 17. ¹¹ , ¹² The Ts65Dn mouse model is the most‐used mouse model for DS and it consists of a partial trisomy of murine chromosome 16 (from App to Mx1), covering > 50% of genes in the homologous region of HSA21. ¹³ , ¹⁴ Importantly, Ts65Dn mice at 10 months of age also display common features with AD. Indeed, Ts65Dn mice show increased amyloid precursor protein (APP) expression, and degeneration of locus coeruleus noradrenergic neurons and basal forebrain cholinergic neurons, among others. ¹⁵

The endocannabinoid system is a neuromodulatory system involved in synaptic homeostasis relevant for neuroinflammation and memory functions. ¹⁶ The endocannabinoid system is composed of the cannabinoid receptors, mainly the cannabinoid type‐1 and type‐2 receptors (CB1R and CB2R, respectively), their endogenous ligands known as endocannabinoids, and the enzymes involved in their synthesis and degradation. ¹⁷ In the Ts65Dn model, previous studies showed that CB1R expression and function was enhanced in the hippocampus of young‐adult Ts65Dn mice. Moreover, pharmacological and genetic CB1R inhibition restored memory deficits, synaptic plasticity, and adult neurogenesis in this mouse model at a young age, ¹⁸ although similar results have also been demonstrated after protecting the endocannabinoid 2‐AG from enzymatic degradation. ¹⁹ However, the potential efficacy of long‐term treatments with CB1R inhibitors on behavioral outcomes and in specific neurological phenotypes characteristic of aged Ts65Dn trisomic mice has not yet been assessed.

A deeper understanding of the endocannabinoid system, and specifically CB1R, in DS is essential, as multiple studies have highlighted the dysregulation of this system and its potential role in DS pathology. ¹⁸ , ¹⁹ , ²⁰ , ²¹ However, there is no evidence that CB1R is dysregulated in individuals with DS, and its pharmacological modulation in aged mouse models has not been studied. Moreover, emerging therapies currently undergoing clinical trials target the modulation of the CB1R (NCT05748405), presenting a promising avenue for novel therapeutic strategies aimed at addressing cognitive and neurological deficits in DS.

In this study, we found that CB1R expression is enhanced in hippocampal post mortem tissue from individuals with DS further confirming CB1R as a potential therapeutic target. In addition, long‐term pharmacological inhibition of CB1R with a low dose of rimonabant, enhanced memory in Ts65Dn mice at an age when this mouse model has a noticeable neurodegenerative and neuroinflammatory phenotype, the latter being sensitive to CB1R inhibition.

2. METHODS

2.1. Post mortem human samples

Brain samples and data from patients included in this study were provided by the Hospital Clínic de Barcelona ‐ August Pi i Sunyer Biomedical Research Institute Biobank (HCB‐IDIBAPS Biobank; B.0000575), integrated in the Platform Instituto de Salud Carlos III (ISCIII) Biobanks and Biomodels, and they were processed following standard operating procedures with the appropriate approval of the ethics and scientific committees. Brain samples included in this study comprised five from subjects with DSAD (median age 64) and five from typically developing male subjects (median age 66; Table S1 in supporting information). Neuropathologic examination was performed according to standardized protocols. ²² The ABC score assesses the burden of AD‐related neuropathological changes, including the density and distribution of amyloid plaques and neurofibrillary tangles, based on post mortem brain tissue analysis. ²³

RESEARCH in CONTEXT

1. Systematic Review: In our previous work (Navarro‐Romero et al., 2019), we identified dysregulation of cannabinoid type‐1 receptor (CB1R) in two mouse models of Down syndrome (DS) using young‐adult mice.

2. Interpretation: We reveal for the first time that CB1R is upregulated in the brain of individuals with DS and in middle‐aged DS mice, and long‐term CB1R inhibition normalizes cognitive performance, reduces inflammatory phenotypes, and does not modulate neurodegeneration, establishing the significance of CB1R as a therapeutic target. Our findings contribute to the understanding of how CB1R alterations in DS could influence memory deficits and neurodegenerative processes, highlighting the need for targeted pharmacological interventions.

3. Future Directions: Future research should focus on identifying translational drugs that modulate CB1R effectively, exploring its role across diverse populations and sexes, and gaining a deeper understanding of its involvement in cognitive impairments in DS.

2.2. Human tissue histology for CB1R detection

Half‐brains were fixed in formaldehyde solution for 3 weeks. The middle‐posterior hippocampal region was embedded in paraffin and cut at 4 µm. The sections were placed in a BOND‐MAX Automated Immunohistochemistry Stainer (Leica Biosystems Melbourne Pty Ltd). Tissues were deparaffinized and pre‐treated with the Epitope Retrieval Solution 2 (ethylenediaminetetraacetic acid buffer pH = 8.8) at 98°C for 40 minutes. CB1R primary antibody (rabbit, 1:1000, Immunogenes) was incubated for 60 minutes. Subsequently, tissues were incubated with polymer horseradish peroxidase (HRP) for 8 minutes and developed with 3,3′‐Diaminobenzidine chromogen for 10 minutes. Slides were counterstained with hematoxylin.

2.3. Image acquisition and analysis of human tissue

Images of stained sections were obtained with a slide scanner. Hippocampal subregions were anatomically defined following Allen Brain Atlas guidelines using QuPath software. ²⁴

CB1R immunostaining and hematoxylin counterstaining were performed sequentially on the same tissue sections, allowing precise spatial correspondence between CB1R signal and cell density estimation.

Quantitative analysis was performed using Fiji (ImageJ). First, mean gray value optical density was measured for CB1R immunoreactivity within manually defined regions of interest (ROIs). In parallel, the percentage of area occupied by hematoxylin‐stained cells was calculated to estimate local cellular density. This was done by applying an automatic threshold (Triangle dark), followed by erosion and dilation (“OPEN” command) to remove background signal.

Finally, CB1R optical density was normalized to the hematoxylin‐occupied area within each ROI, and the resulting values were expressed as a percentage relative to the control group. This normalization accounted for potential regional differences in cellular density.

2.4. Amyloid beta plaque assessment in human hippocampus

Amyloid beta (Aβ) pathology was evaluated in hippocampal sections from the same cohort of human brain samples used for CB1R analysis. Sections were incubated with mouse monoclonal anti–Aβ antibody (clone 6F/3D, Dako, 1:100), which targets amino acids 8 to 17 of the Aβ peptide.

Semi‐quantitative analysis of Aβ plaque burden was conducted in five hippocampal subregions: dentate gyrus, CA1, CA2, CA3, and CA4. For each region, the field with the highest plaque density at 10× magnification was selected. Plaque load was scored using a five‐level scale: 0 (no plaques), isolated (single plaque, inconsistently present), mild (1–5 plaques/field), moderate (6–20 plaques/field), and frequent (> 20 plaques/field). Both diffuse and neuritic plaques were considered. Scoring was performed by an investigator blinded to diagnosis. Regional Aβ scores were descriptively compared to CB1R immunoreactivity in corresponding subregions.

2.5. Animals

All animal procedures were conducted following ARRIVE (Animals in Research: Reporting In Vivo Experiments) guidelines ²⁵ and standard ethical guidelines (European Communities Directive 2010/63/EU). Procedures were approved by the local ethical committee (Comité Ètic d'Experimentació Animal‐Parc de Recerca Biomèdica de Barcelona, CEEA‐PRBB) and local authorities (Generalitat de Catalunya).

All experimental mice were bred at the Barcelona Biomedical Research Park (PRBB) Animal Facility. Ts65Dn experimental mice (Jackson Laboratory, strain #005252) were obtained by repeated backcrossing Ts65Dn females (B6EiC3Sn.BLiA‐Ts(1716)65Dn/DnJ) to C57BL/6JEiJ x C3Sn.BLiA Pde6b+/DnJ F1 hybrid males. The parental generation was purchased from The Jackson Laboratory. Euploid littermates of Ts65Dn mice served as wild‐type (WT) controls.

Mice were housed in plexiglass cages with a maximum of four male or five female mice per cage in a temperature‐controlled (21°C ± 1°C; mean ± range) and humidity‐controlled (55% ± 10%) environment. Lighting was maintained at 12 hour cycles (light on at 8:00 am; light off at 8:00 pm). All experiments were conducted during the light phase in an experimental room at the animal facility. Food and water were available ad libitum. All behavioral experiments were performed by an observer blind to the genotype and treatment.

2.6. Drug treatments

Rimonabant was purchased from Axon Medchem. Rimonabant was first prepared as a 40 mM stock solution in ethanol. Then, rimonabant stock solution was diluted to a final concentration of 2.16 µM (for 0.1 mg/kg/day) or 10.8 µM (for 0.5 mg/kg/day) in 0.3% 2‐hydroxypropyl‐β‐cyclodextrin in water. The compound was administered through the drinking bottle, and the same solution without rimonabant was used in littermate mice as control/placebo/vehicle condition. Animals with different treatment (vehicle or rimonabant) were maintained in different home cages, while home cages could hold both genotypes (Ts65Dn or WT). Mice were included in the study at the age of weaning (postnatal day 21 [PND21]) and randomly associated to one of the treatments. During the first 2.5 months of treatment mice received rimonabant 0.1 mg/kg/day. This dose was increased to the final concentration of 0.5 mg/kg/day by 3.5 months of age. The dosage was then maintained until mice were euthanized at 10 months of age. The concentration of rimonabant in the drinking water was calculated based on the target dose and an estimated average daily water intake equivalent to ≈ 10% of the mouse's body weight (e.g., 3.5 ml/day for a 35 g mouse), consistent with published data. ²⁶ Water consumption was regularly monitored throughout the treatment period and remained within expected ranges, validating the dosing assumption.

2.7. Behavioral tests

All behavioral tests were performed in a sound‐attenuated room with dim illumination. A digital camera on top of the maze was used to record the sessions.

The novel object‐recognition memory test was performed following a previously described protocol. ²⁷ Briefly, novel object‐recognition memory was assessed in a V‐shape maze with dim illumination (3–5 lux). This task consists of three different phases (habituation, familiarization/training, and test) performed on 3 consecutive days for 9 minutes. On day 1, mice were habituated to the empty V‐maze. Next day, mice were introduced in the V‐maze where two identical objects were presented in the familiarization/training phase. Finally, the test was performed 24 hours later, where one of the familiar objects was replaced with a novel object and the exploration time for both objects was recorded. Object exploration was defined as orientation of the nose toward the object at a distance < 2 cm. A discrimination index (DI) was calculated as the difference between the time spent exploring either the novel (Tn) or familiar (Tf) object divided by the total time spent exploring both objects: DI = (Tn − Tf) /(Tn + Tf).

Locomotor activity was assessed for 120 minutes. Individual locomotor activity boxes (9 cm × 20 cm × 11 cm; Imetronic) were used in a low luminosity environment (5 lux). The total activity was detected by infrared sensors.

The elevated plus maze test was performed in a black plexiglass apparatus with four arms (29 cm long x 5 cm wide), two open and two closed, set in a cross from a neutral central square (5 cm x 5 cm) elevated 30 cm above the floor and indirectly illuminated from the top (40–50 lux in the open arms/4–6 lux in the closed arms). Five minute test sessions were performed, and the total number of entries and the percentage of time spent in the open arms were used as measures of anxiety‐like behavior.

2.8. Preparation of histological mice brain samples

Mice were deeply anesthetized by intraperitoneal injection (0.2 mL per 10 g body weight) of a mixture of ketamine (100 mg/kg) and xylazine (20 mg/kg). Then, mice were perfused intracardially with 4% paraformaldehyde in a 0.1 M Na2HPO4/NaH2PO4 (pH 7.5) phosphate buffer (PB) using a peristaltic pump. The brains were removed from the skull and postfixed overnight at 4°C in the same fixative solution. The next day, brains were moved to 30% sucrose in PB solution. Brain sections (30 µm) were obtained with a sliding microtome and kept in a 5% sucrose in PB solution at 4°C until they were used for immunodetection.

2.9. Immunofluorescence

First, free‐floating sections were washed three times (5 minutes each) in PB. Afterward, the tissue was incubated in blocking buffer for 2 hours. The blocking buffer was made of 0.3% Triton X‐100 and 3% normal donkey or goat serum diluted in 0.1 M PB, for neurodegeneration assessment and inflammation assessment, respectively. Then, sections were incubated with one of the primary antibodies (mouse anti‐TH [1:1000, Sigma]; rabbit anti‐p75NTR [1:1000, Millipore]; mouse anti‐Iba1 [1:500, Wako]) for 24 hours at 4°C. Then, sections were washed three times (10 minutes each) and subsequently incubated with the corresponding secondary fluorescent antibodies for 2 hours at room temperature. The secondary antibodies used were: goat anti‐mouse Alexa Fluor 488 (1:1000), donkey anti‐rabbit Alexa Fluor 555 (1:500), and donkey anti‐mouse Alexa Fluor 555 (1:500). Finally, the sections were washed three times in PB (10 minutes each) and were mounted onto gelatin‐coated slides with Fluoromont/DAPI mounting medium.

2.10. Image acquisition and analysis

For tyrosine hydroxylase (TH)+ cell counting, four coronal sections of the locus coeruleus were selected (from 5.34 to 5.52 posterior to Bregma). ²⁸ Images of stained sections were obtained with a confocal microscope TCS SP8 LEICA (Leica Biosystems) using a dry objective (20× objective, 0.75 zoom) with a sequential line scan at 1024 × 1024‐pixel resolution. The number of TH+ cells was manually quantified using Fiji software (Image J). The number of positive cells was calculated as the mean of total number of cells counted referred to the area of the locus coeruleus (µm2).

For p75NTR+ cell counting, systematic series of coronal sections (one every six sections) per animal were selected, covering the rostral to caudal extension of the medial septum (from 1.18 and 0.38 mm posterior to Bregma). ²⁸ Images were acquired using a Leica DM6000B microscope (10x objective). To obtain the macro used for the quantification of p75NTR+ cells, first the “Yen threshold” was applied to images. Then an erosion and dilatation operation was applied with the “OPEN” command followed by “remove outliers” command (radius = 2 and threshold = 50). Finally, the “watershed” command was executed and cells > 70µm² were counted as positive. The number of positive cells was calculated as the mean of total number of cells counted referred to the area of the medial septum (µm²).

Iba1 staining was used to evaluate microglial morphology. Confocal microscopy images of whole Z stack from the slice were acquired in the stratum radiatum. Images of stained sections were obtained with a confocal microscope TCS SP5 STED LEICA (Leica Biosystems) using an immersion‐oil objective (40× objective, 1.5 zoom) with a sequential line scan at 1024 × 1024 pixel resolution and with 0.3 µm depth intervals. The perimeter of the microglial soma was quantified in 20 cells per animal using Fiji software (Image J).

2.11. Western blotting

Mouse brain tissue was rapidly dissected, immediately frozen on dry ice, and stored at −80°C until use. Samples were processed following a protocol previously described ²⁹ to obtain the total solubilized fraction, separated on sodium dodecyl sulfide polyacrylamide gel electrophoresis gels, and transferred into nitrocellulose membranes as previously described. ²⁹ The primary antibodies used were rabbit anti‐CB1R (1:1000, Immunogenes) and mouse anti‐GAPDH (1:50,000, sc‐32233, Santa Cruz). Primary antibodies were detected with horseradish peroxidase conjugated anti‐rabbit and anti‐mouse antibodies and visualized by enhanced chemiluminescence detection (Luminata Forte Western HRP substrate, Merck Millipore). Digital images were acquired on a ChemiDoc XRS System (Bio‐Rad) and quantified using The Quantity One software v4.6.3 (Bio‐Rad). Optical density values for CB1R were normalized to actin optical density values as loading control in the same sample and expressed as a percentage of control group (WT).

2.12. Inflammatory profiling using the Olink Target 48 Mouse Cytokine panel

Plasma cytokine levels were measured using the Olink Target 48 Mouse Cytokine Proximity Extension Assay (PEA, Olink Bioscience), which quantifies 43 inflammation‐related proteins. PEA technology relies on the simultaneous binding of each target protein by a pair of DNA‐tagged antibodies, generating a unique DNA reporter sequence upon proximity. This sequence is subsequently amplified and quantified by real‐time polymerase chain reaction, as previously described. ³⁰ Protein abundance was expressed as Normalized Protein eXpression (NPX) values on a log2 scale. Data normalization was performed according to the manufacturer's instructions. The Olink assay was carried out at the Immune Response and Biomarkers Core Facility at ISGlobal (Barcelona, Spain).

2.13. Rimonabant detection in mice brain

For the analysis of rimonabant levels in the mice brain, tissues (40–80 mg) were weighted and homogenized with 1 mL grinder Dounce (Wheaton) in two steps: first by adding 400 µL of HCOOH 0.1% (30 movements “loose” pestle, followed by 30 movements “tight” pestle were used for homogenization) followed by protein precipitation with 800 µL of ethanol. The mixture was centrifuged 10 minutes at 15,700 g, 4°C, and supernatant was recovered and stored at −20°C until use. Then, 100 µL of supernatant was mixed with 25 µL of the internal standard (rimonabant‐d10, 0.004 µg/mL in ethanol, Bertin Technologies), and 4 µL of the final mixture were used for the high‐performance liquid chromatography tandem mass spectrometry analysis.

Samples were analyzed in an Acquity UPLC System (Waters Associates) coupled to a mass spectrometer (Quattro Premier, Watters Associates). Chromatographic separation was carried out in an Acquity BEH C18 (100 mm x 2.1 mm i.d., 1.7 µm; Waters Associates) at a flow rate of 0.4 mL/minute. Ammonium formate (1 mM)‐HCOOH 0.01% (A) and MeOH with ammonium formate (1 mM)‐HCOOH 0.01% (B) were used as mobile phases. After keeping 40% B for 0.5 minutes, the gradient was increased to 95% B in 3 minutes and maintained at 95% B for 1 minute after going back to initial conditions. Detection of analytes was done by the selected reaction monitoring (SRM) method, being the transitions used for identification and quantification (in bold) for each compound as follows: 465→84, 99, 365 (rimonabant); 475→94, 365 (rimonabant‐d10).

2.14. Experimental design and statistical analysis

Sample size choice was based on previous studies with similar experimental approaches ³¹ , ³² and it is indicated in figure legends for each experiment. Data were analyzed with GraphPad Software using unpaired Student t test or two‐way analysis of variance for multiple group comparisons. Subsequent post hoc analysis (Bonferroni) was used when significance in interaction between factors was found. Comparisons were considered statistically significant when P < 0.05. Outlier measures were not considered for analysis when such measures were higher or lower than two times the standard deviation from the mean.

3. RESULTS

3.1. CB1R expression is enhanced in the hippocampal tissue of subjects with DSAD and middle‐aged trisomic Ts65Dn mice

Previous studies in our laboratory reported that the expression of CB1R was increased in the hippocampus of young‐adult Ts65Dn male mice. ¹⁸ We hypothesized that hippocampal CB1R expression might also be enhanced in human subjects with DSAD. Therefore, we immunodetected and quantified the expression of CB1R in brain slices corresponding to the hippocampal formation of aged subjects with DSAD and sex, age, and post mortem interval–matched controls (Table S1).

We observed an almost significant increase in CB1R immunodetection normalized by hematoxylin detection, when whole dorsal hippocampus was analyzed (Figure 1A; see Figure S1 and Figure S2 in supporting information for representative images of CB1R and hematoxylin detection). This result was not associated with changes in hematoxylin detection because no differences were found in this parameter (Figure S3B in supporting information). Further sub‐region analysis of the normalized CB1R immunoreactivity in the dentate gyrus and CA2 showed a significant increase in samples from DSAD subjects (Figure 1B and 1C) that was not associated with changes in hematoxylin detection because no differences were found in this parameter (Figure S4 in supporting information). In contrast, the CA1, CA3, and CA4 subregions exhibited no differences in CB1R immunoreactivity (Figure 1D–F). To determine whether this regional overexpression extended along the anterior–posterior axis, we also examined CB1R immunoreactivity in the anterior hippocampus from a subset of available cases (Figure S3C–E). No significant differences were observed in this region, supporting a posterior‐specific pattern of CB1R upregulation.

FIGURE 1.

CB1R expression is enhanced in human post mortem samples of aged subjects with DS and in dorsal hippocampus of middle‐aged Ts65Dn mice. A, Representative images and average intensity of CB1R expression in the posterior hippocampus of control and subjects with DS (control, n = 5, DS = 5; scale bar = 1 mm). B–F, Average intensity of CB1R expression relative to cellular content in the different subregion of the posterior hippocampus. G, Representative immunoblots and quantification of CB1R in hippocampus from WT and Ts65Dn mice of 10 months of age (WT, n = 7, Ts65Dn, n = 5). Distribution of individual data with mean ± standard error of the mean. * P ˂ 0.05; ** P ˂ 0.01 (genotype or condition effect) by Student t test. CB1R, cannabinoid type‐1 receptor; DS, Down syndrome; GAPDH, glyceraldehyde 3‐phosphate dehydrogenase; WT, wild type.

To investigate the potential relationship between CB1R expression and Aβ pathology, we performed additional immunohistochemical staining for Aβ in hippocampal sections from the same individuals. A semi‐quantitative scoring system was applied to assess plaque burden across hippocampal subregions. When CB1R immunoreactivity was compared across subregions with different levels of Aβ pathology, no consistent association was observed (Figure S5 in supporting information). Subregions exhibiting higher plaque load did not systematically present with increased or decreased CB1R signal. These observations, although exploratory and limited by the small sample size, do not suggest a clear correlation between Aβ plaque density and CB1R expression in this cohort.

To determine whether this CB1R dysregulation is conserved in DS models, we analyzed hippocampal tissue from Ts65Dn mice at middle age, a time point at which neurodegenerative features are already present. Protein expression analysis of CB1R in dorsal hippocampus of trisomic Ts65Dn middle‐aged mice also showed a significant increase compared to WT littermate mice (Figure 1G).

Together, these data indicate that CB1R overexpression is observed in both the posterior hippocampus of human subjects with DS and in the dorsal hippocampus of middle‐aged trisomic Ts65Dn mice.

3.2. Sustained oral administration of a low dose of rimonabant improves memory performance in young‐adult and middle‐aged Ts65Dn mice

We then tested whether a long‐term sustained pharmacological intervention with a low dose of the selective CB1R antagonist rimonabant ³³ would be suitable to improve memory performance in Ts65Dn trisomic mice. Both male and female mice (Ts65Dn and WT littermates) received the treatment (rimonabant or vehicle) through the drinking water from PND21 until 10 months of age, when brain samples were extracted (Figure 2A).

FIGURE 2.

Long‐term CB1R inhibition improves memory performance in young and in middle‐aged male and female Ts65Dn mice. A, Schematic representation of the experimental protocol. B, Discrimination index in NORT of WT and Ts65Dn male of 4 months of age with VEH or RIM (0.5 mg/kg/day; WT VEH, n = 14; WT RIM, n = 9; Ts65Dn VEH n = 9; Ts65Dn RIM n = 11). C–D, Discrimination index in NORT of WT and Ts65Dn male (C) and female (D) mice treated of 10 months of age with VEH or RIM (0.5 mg/kg/day; males: WT VEH, n = 15; WT RIM, n = 9; Ts65Dn VEH n = 6; Ts65Dn RIM n = 12; females: WT VEH, n = 13; WT RIM, n = 14; Ts65Dn VEH n = 10; Ts65Dn RIM n = 9). E, Representative immunoblots and quantification of CB1R in hippocampus from WT and Ts65Dn mice (10 months of age; WT, n = 6, Ts65Dn, n = 4). F, Representative immunoblots and quantification of CB1R in hippocampus from Ts65Dn treated with vehicle and Ts65Dn mice treated with RIM (0.5 mg/kg/day; 10 months of age; Ts65Dn VEH, n = 4, Ts65Dn RIM, n = 6). Distribution of individual data with mean ± standard error of the mean. * P ˂ 0.05, ** P ˂ 0.01 (genotype effect); # P ˂ 0.05, ## P ˂ 0.01, ### P ˂ 0.001 (treatment effect) by Bonferroni post hoc test after two‐way analysis of variance. CB1R, cannabinoid type‐1 receptor; DS, Down syndrome; EPMT, elevated plus maze test; GAPDH, glyceraldehyde 3‐phosphate dehydrogenase; LAT, locomotor activity test; NORT, novel object recognition test; PND21, postnatal day 21; RIM, rimonabant; VEH, vehicle; WT, wild type.

During treatment, mice were analyzed in their behavioral response to assess different aspects of general activity which could be of relevance to identify possible side effects associated with CB1R inhibition as well as for evaluating the efficacy of the treatment at the cognitive level (Figure 2A). At 4 months of age, mice were tested using the novel object recognition (NOR) task. Ts65Dn male mice treated with vehicle exhibited the expected memory deficits in this task. ¹⁴ , ³⁴ In contrast, Ts65Dn male mice treated with rimonabant demonstrated significantly improved cognitive performance, comparable to that of vehicle‐treated controls (Figure 2B). This observation confirmed that sustained inhibition of CB1R does not lead to tolerance to the mnemonic effects observed in trisomic mice.

Given that Ts65Dn mice display a noticeable neurodegenerative phenotype reminiscent of that of middle‐aged subjects with DS, we reevaluated NOR memory in the same cohort of mice at 10 months of age. Remarkably, no tolerance developed over the course of treatment, as memory improvement was again observed in rimonabant‐treated male and female Ts65Dn mice at this later time point (Figure 2C and  2D). Importantly, the NOR test did not reveal any effects of the treatment on exploration time (Figure S6 in supporting information), and it also did not show an effect on memory in WT mice (Figure 2C and 2D), demonstrating that inhibiting CB1R specifically enhances hippocampal‐dependent memory in Ts65Dn mice.

To evaluate any potential behavioral disruptions related to CB1R inhibition, we first evaluated anxiety‐like behavior using the elevated plus maze task (EPMT) within the same cohort of mice. Ts65Dn trisomic male mice showed a low anxiety‐like phenotype, and rimonabant treatment did not modify such responses (Figure S7A in supporting information). In contrast, rimonabant treatment increased anxiety‐like behavior in WT female mice at this time point (Figure S7B). Interestingly, we observed a significant increase in the total number of entries in Ts65Dn trisomic mice compared to WT that was not altered by rimonabant treatment (Figure S7C and D). Moreover, locomotor activity (LAT) assessment of the same cohorts 2 months later revealed a hyperlocomotor phenotype in Ts65Dn trisomic mice that was not modified by rimonabant exposure (Figure S8A and B in supporting information). Furthermore, the presence of rimonabant was measured in a representative sample of brain tissue from treated mice at 10 months of age (mean ± standard error of the mean, rimonabant treated groups = 12.6 ± 6.0 pg/mg of wet tissue).

To investigate whether the behavioral effects of rimonabant were associated with changes in CB1R protein levels, we performed western blot analysis in the dorsal hippocampus of mice at 10 months of age. Consistent with previous findings, Ts65Dn mice treated with vehicle showed significantly increased CB1R expression compared to WT controls (Figure 2E). Moreover, CB1R protein levels did not differ between vehicle‐ and rimonabant‐treated Ts65Dn mice, indicating that chronic CB1R inhibition did not alter total receptor abundance (Figure 2F). In WT mice, rimonabant also did not significantly change CB1R protein levels, although a trend toward reduction was observed (Figure S9 in supporting information). These findings support the interpretation that the cognitive improvements observed after long‐term rimonabant treatment are mediated by functional modulation of CB1R signaling rather than by downregulation of CB1R expression.

Together, these findings show that chronic long‐term rimonabant exposure is effective in restoring NOR memory in young‐adult Ts65Dn trisomic mice without compromising other behavioral features in these trisomic animals.

3.3. Noradrenergic and cholinergic alterations are not prevented by long‐term rimonabant treatment

Noradrenergic neurons in the locus coeruleus (LC‐NE neurons) suffer degeneration in DS. ⁶ We assessed whether long‐term rimonabant treatment could alter LC‐NE neurodegeneration in Ts65Dn trisomic mice where we revealed that this treatment improved memory performance. We used TH, the limiting enzyme in the synthesis of dopamine and norepinephrine, ³⁵ as a specific marker of LC‐NE neurons. Analysis of LC‐NE neurons revealed a significant decrease in the density of TH+ neurons in this region of Ts65Dn male mice, replicating similar results of previous studies. ³⁶ Notably, LC‐NE neuron loss was not prevented by long‐term rimonabant treatment (Figure 3A).

FIGURE 3.

Long‐term CB1R pharmacological inhibition did not modify the degeneration of cholinergic and adrenergic neurons in male Ts65Dn mice. A, Representative grayscale confocal images and average density of TH+ cells in the locus coeruleus of WT and Ts65Dn mice of 10 months of age treated with VEH or RIM (0.5 mg/kg/day; WT VEH, n = 6; WT RIM, n = 4; Ts65Dn VEH n = 4; Ts65Dn RIM n = 7; scale bar = 100 µm). B, Representative grayscale images and average density of p75NTR+ cells in the medial septum of the basal forebrain of WT and Ts65Dn mice of 10 months of age with VEH or RIM (0.5 mg/kg/day; WT VEH, n = 6; WT RIM, n = 5; Ts65Dn VEH n = 4; Ts65Dn RIM n = 7; scale bar = 200 µm). Distribution of individual data with mean ± standard error of the mean. * P ˂ 0.05 (genotype effect) by two‐way analysis of variance. CB1R, cannabinoid type‐1 receptor; RIM, rimonabant; TH, tyrosine hydroxylase; VEH, vehicle; WT, wild type.

Similarly, basal forebrain cholinergic neurons, which are the primary source of acetylcholine to the hippocampus, also exhibit age‐related degeneration in DS. ⁷ , ⁸ In our experiments, Ts65Dn trisomic male mice displayed a non‐significant trend toward decreased cholinergic neuron density in the medial septum, as measured by immunostaining for the p75 neurotrophin receptor (p75NTR). ¹⁵ This trend aligns with previous reports. ³⁷ Nevertheless, rimonabant treatment maintains this same trend in cholinergic degeneration (Figure 3B).

These findings indicate that the improvement in memory performance observed with rimonabant treatment in aged Ts65Dn mice occurs independently of any neuroprotective effects on noradrenergic and cholinergic neurons.

3.4. Anomalous microglial morphology and inflammatory signatures in Ts65Dn trisomic mice are normalized by long‐term treatment with rimonabant

In addition to early neurodegeneration, brain inflammation is a significant neurological feature of AD associated with DS. ⁹ , ³⁸ Microglial cells are the major mediators of the neuroinflammatory response in the brain and different studies have observed an increase in microglial reactivity in adult Ts65Dn mice ¹⁵ , ³⁹ , ⁴⁰ as well as in subjects with DSAD. ⁴¹ In the same previously studied cohort of mice, we observed an increase in microglial soma size, measured as cell body area, in the hippocampus of trisomic mice. Notably, long‐term treatment with rimonabant resulted in a normalization of microglial morphology to control values (Figure 4A). Together, these results demonstrate that sustained CB1R pharmacological inhibition normalized microglia morphology in the hippocampus of trisomic mice.

FIGURE 4.

Long‐term CB1R pharmacological inhibition reduced microglia reactivity in the hippocampus and markers of inflammation in plasma of male Ts65Dn mice. A, Representative confocal images and average microglial somatic area of Iba1+ cells in the hippocampus of WT and Ts65Dn mice of 10 months of age with VEH or RIM (0.5 mg/kg/day; WT VEH, n = 6; WT RIM, n = 5; Ts65Dn VEH n = 4; Ts65Dn RIM n = 6; scale bar = 10 µm). B–E, Expression levels of inflammatory markers in plasma of WT and Ts65Dn mice of 10 months of age treated with VEH or RIM (0.5 mg/kg/day; WT VEH, n = 7; Ts65Dn VEH n = 8; Ts65Dn RIM n = 9). F, Inflammation score calculated from the plasmatic inflammatory markers in WT and Ts65Dn mice of 10 months of age treated with VEH or RIM (0.5 mg/kg/day; WT VEH, n = 7; Ts65Dn VEH n = 8; Ts65Dn RIM n = 9). G, Correlation between inflammatory score and performance in the novel object recognition (NOR) test at 10 months of age (WT VEH, n = 7; Ts65Dn VEH n = 8; Ts65Dn RIM n = 9). Distribution of individual data with mean ± standard error of the mean. ** P ˂ 0.01 (genotype effect); # P ˂ 0.05, ### P ˂ 0.001 (treatment effect) by Bonferroni post hoc test after two‐way analysis of variance for microglial somatic area analysis and by Student t test for plasma inflammation analysis. Correlation analyzed with Spearman test; $P ˂ 0.05. CB1R, cannabinoid type‐1 receptor; CCL4, chemokine ligands 4; IL, interleukin; RIM, rimonabant; VEH, vehicle; WT, wild type.

To functionally assess the impact of long‐term CB1R blockade on microglial activity, plasma levels of inflammation‐related proteins were analyzed using the Olink Target Mouse Cytokine panel. Among the analytes measured, four inflammatory markers—interleukin (IL)‐1β, chemokine ligands 4 (CCL4), IL‐4, and IL‐16—showed a consistent trend toward increased levels in vehicle‐treated Ts65Dn mice compared to euploid controls and were significantly reduced in trisomic mice treated with rimonabant to values comparable to those observed in the euploid group (Figure 4B–E). Importantly, all four inflammatory markers have been previously implicated in microglial activation and regulation and are known to participate in neuroinflammatory processes associated with neurodegeneration. ³⁷ , ⁴² , ⁴³ , ⁴⁴

To capture coordinated changes across these proinflammatory markers, a composite inflammation z score was calculated per animal by averaging the z transformed values of the four inflammatory markers. This inflammation score showed a trend to increase in vehicle‐treated Ts65Dn mice compared to controls and was significantly decreased by rimonabant treatment (Figure 4F). Importantly, the inflammation score negatively correlated with cognitive performance in the novel object recognition test (r  =  –0.48, P  =  0.03; Figure 4G), indicating a potential association between these two parameters.

These findings provide functional evidence supporting the anti‐inflammatory effect of rimonabant in Ts65Dn mice.

4. DISCUSSION

Intellectual disability in DS is characterized by learning and memory impairments together with age‐related comorbidities, such as AD. Currently, there is no effective treatment to circumvent the genetic alterations in DS. Therefore, there is an urgent need to investigate novel therapeutic targets. In this study, we describe, for the first time to the best of our knowledge, a dysregulation in the expression of CB1R in the hippocampus of adult human subjects with DSAD and Ts65Dn middle‐aged trisomic mice, revealing CB1R as a promising target for long‐term treatments to enhance hippocampal‐dependent memory in this disorder.

We previously described that CB1R expression and function were upregulated in young‐adult Ts65Dn trisomic mice, ¹⁸ but whether this feature is observed in adult human subjects with DSAD had not been previously addressed. Therefore, we analyzed CB1R immunodetection in the hippocampus of aged subjects with DSAD compared to age‐matched healthy controls. A significant enhancement of CB1R immunodetection was found specifically in the dentate gyrus and CA2 of the hippocampus, whereas an almost significant increase was observed when the posterior hippocampus was analyzed as a whole normalizing by cell density. However, analysis of raw (unnormalized) optical density values revealed that the increase in CB1R expression becomes statistically significant at the whole‐region level as well, supporting the robustness of our findings. These findings are congruent with previous studies, which reported regional redistribution of CB1R in human fetal brains with DS, ²¹ suggesting that alterations in CB1R expression may begin early in neurodevelopment and follow a region‐specific pattern in adulthood.

We also analyzed CB1R immunoreactivity in anterior hippocampal sections from a subset of cases and found no significant differences between DSAD and control subjects in this region. This observation supports the anatomical specificity of CB1R upregulation to the posterior hippocampus in DSAD. Based on these findings, all subsequent subregional analyses in human samples were focused on the posterior hippocampus, ensuring anatomical consistency across cases. Similarly, the analysis in Ts65Dn mice was conducted in the dorsal hippocampus, which is considered functionally and topographically equivalent to the posterior hippocampus in humans, ⁴⁵ as both are primarily involved in spatial navigation and episodic memory, domains frequently affected in DS.

This cross‐regional design is further supported by previous findings in young adult Ts65Dn mice, ²⁰ in which CB1R overexpression was observed specifically in the dorsal—but not ventral—hippocampus. The region‐dependent pattern of CB1R dysregulation observed in our study aligns with those results, reinforces the importance of anatomical precision in assessing CB1R alterations, and highlights the relevance of CB1R upregulation in these areas as a potential contributor to the cognitive phenotype.

Together, these findings further underline the potential interest of diminishing CB1R function as a suitable pharmacological approach for the treatment of memory impairment in DS, as we previously suggested. ¹⁸

Importantly, regarding the dementia status of our cohort, four out of the five DSAD subjects had a clinical diagnosis of dementia. The fifth individual did not undergo formal neuropsychological evaluation prior to death due to logistical constraints. However, post mortem neuropathological analysis revealed a Consortium to Establish a Registry for Alzheimer's Disease (CERAD) score of A3B2C3 (indicative of extensive amyloid plaques and neurofibrillary tangles) and consistent with a high likelihood of AD according to National Institute on Aging–Alzheimer's Association criteria. ⁴⁶ , ⁴⁷ Furthermore, it is well established that virtually all individuals with DS exhibit AD‐related neuropathological features by the age of 40, and the prevalence of clinical dementia increases markedly with age. ⁴⁸ Given the subject's age and pathological findings, it is highly probable that they would have fulfilled clinical criteria for AD had an assessment been performed.

Interestingly, prior studies have suggested that endocannabinoid signaling is reorganized during AD progression. ⁴⁹ , ⁵⁰ If similar mechanisms operate in DSAD, CB1R upregulation could be influenced not only by neurodevelopmental alterations but also by the onset of AD‐related pathology.

To further explore whether CB1R expression in DSAD is influenced by AD‐related pathology, we semi‐quantitatively assessed Aβ plaque burden in the same hippocampal subregions where CB1R levels were measured. Notably, this subregional approach allowed us to account for the well‐documented heterogeneity in hippocampal vulnerability to AD pathology, such as the early degeneration of CA1. ⁵¹ Despite this anatomically resolved analysis, no consistent association emerged between Aβ plaque density and CB1R expression. These findings suggest that CB1R upregulation in DSAD is not directly explained by local amyloid pathology. Future studies using larger cohorts may be required to clarify whether CB1R alterations in DSAD reflect or diverge from typical AD‐related processes.

In line with our anatomically resolved findings, recent multiomic studies in DSAD support a region‐ and cell‐type–specific pattern of CB1R dysregulation. A significant increase in CNR1 transcript levels was reported in excitatory neurons of the frontal cortex, despite no differences at the bulk tissue level. ⁵² Similarly, transcriptomic trends in specific excitatory neuron subtypes across DSAD, sporadic AD, and controls ⁵³ align with previous observations from our group in hippocampal excitatory neurons in young‐adult Ts65Dn mice. ¹⁸ At the proteomic level, increased CB1R was detected in the frontal cortex, but not the hippocampus, of DS individuals without confirmed AD. ⁵⁴ Notably, these studies did not distinguish anterior from posterior hippocampal regions or subfields. Altogether, these findings suggest that CB1R alterations may be anatomically restricted and disease stage dependent, and could be missed by non‐resolved bulk approaches. Our targeted analysis revealed CB1R upregulation confined to the posterior hippocampus, further highlighting the value of spatially resolved methods. Future work using single‐cell or spatial omics, along with quantitative strategies capable of resolving subregional expression in post mortem tissue, would be highly informative.

Indeed, leaving aside DSAD, previous studies of CB1R expression in the brains of AD patients have been inconsistent. ⁵⁵ , ⁵⁶ , ⁵⁷ Therefore, the enhanced CB1R expression we observed in specific hippocampal subregions could represent a previously undescribed feature of DS and supports the potential interest of therapeutic approaches attenuating CB1R function.

A key limitation of our study is the small sample size of human brain tissues, which impacts the generalizability of our findings. Additionally, as our study included only male subjects, future research with larger, sex‐diverse samples is crucial to clarify the relationship between CB1R expression and hippocampal function in DSAD and to explore CB1R's role in neurodevelopmental and cognitive alterations across brain regions. Our normalization approach was designed to account for possible regional differences in tissue architecture and cell density, which may be influenced by the altered neurodevelopmental trajectory characteristic of DS.

To complement our human findings, we investigated whether similar CB1R alterations occur in the Ts65Dn mouse model of DS. To our knowledge, this is the first study to report CB1R overexpression in Ts65Dn mice at middle age when neurodegenerative features are already present. Our analysis of dorsal hippocampal CB1R expression in middle‐aged Ts65Dn mice revealed a significant increase compared to WT littermates. At this age, these mice exhibit neurodegenerative phenotypes, ¹⁵ suggesting that elevated CB1R expression may persist over time and contribute to cognitive deficits.

In line with this idea, previous work from our group in the Ts65Dn model has already addressed CB1R expression at the cellular level from a functional perspective. ¹⁸ Specifically, using whole‐cell patch‐clamp recordings in CA1 pyramidal neurons, it was demonstrated that the CB1R agonist produced an enhanced inhibitory effect on excitatory (but not inhibitory) synaptic transmission in young‐adult Ts656Dn mice. These findings support a cell type–specific functional upregulation of CB1R, selectively at glutamatergic synapses, and not at GABAergic terminals. This physiological approach provides synapse‐specific resolution and suggests that CB1R dysregulation in the Ts65Dn hippocampus is not uniform, but rather restricted to defined neuronal populations. Our present western blot results complement these observations by confirming an overall increase in CB1R protein levels in the dorsal hippocampus of middle‐aged Ts65Dn mice. Taken together, these complementary approaches indicate that CB1R dysregulation in Ts65Dn hippocampus is not homogeneous and likely involves specific neuronal populations, an aspect that deserves further investigation in middle‐aged animals.

In our investigation, recognition memory performance was preserved in both male and female young‐adult and middle‐aged Ts65Dn trisomic treated mice discarding tolerance processes, which have been observed to be a key adverse feature to reduce the effectiveness of other treatments envisioned to improve cognition in other intellectual disability disorders. ⁵⁸ Our results of long‐term exposure further support the relevance of CB1R to tackle cognitive impairment in DS as observed in sub‐chronic treatments previously assessed, ¹⁸ and open the possibility to evaluate this approach in other intellectual disability disorders in which low doses of CB1R antagonists have also been found effective, such as in fragile X syndrome. ³¹ , ⁵⁹ Importantly, the cognitive improvement was not accompanied by a reduction in CB1R expression, suggesting that functional modulation rather than receptor downregulation underlies the therapeutic effects of rimonabant.

Middle‐aged human adults with DS present neuropathological alterations common to AD, involving the degeneration of noradrenergic neurons in the locus coeruleus and basal forebrain cholinergic neurons, ¹⁰ , ⁶⁰ features that are reproduced by middle‐aged Ts65Dn mice. ³⁶ , ⁶¹ Therefore, we have analyzed the neurodegenerative features of Ts65Dn mice after long‐term rimonabant exposure. Interestingly, we observed that Ts65Dn treated with rimonabant presented evident noradrenergic cell loss and an emerging cholinergic impairment to a similar extent as Ts65Dn mice treated with vehicle. These results discard an effect of rimonabant in these neuronal populations but open the possibility that rimonabant is enhancing some type of cognitive reserve that prevents the memory deficit from becoming evident.

Previous research has revealed that endocannabinoids are dysregulated in AD and contribute to the disorder's progression. ⁶² Because research shows conflicting results, it is difficult to determine whether an increase or reduction in cannabinoid tone is associated with an improvement in pathology. ⁶³ For instance, CB1R and/or CB2R agonists improved memory and/or cognitive deficits in Tg2576 mice, APP/PS1 mice, ⁶⁴ and in rodents receiving intracerebral injections of Aβ. ⁶⁵ , ⁶⁶ Conversely, CB1R antagonism protected against Aβ‐induced memory impairment in mice, ⁶⁷ suggesting that activation of CB1R by endocannabinoids inhibits neurotoxicity, but may worsen its long‐term consequences (such as reduced acetylcholine signaling) that lead to cognitive impairment.

Given that long‐term rimonabant treatment prevented memory deficits in the Ts65Dn mice, we explored other neurological parameters relevant for cognitive performance, such as neuroinflammation. Neuroinflammation is a major contributor to neurodegenerative disorders ⁶⁸ , ⁶⁹ including AD in DS. ³⁸ , ⁴¹ , ⁷⁰ , ⁷¹ Interestingly, recent studies also demonstrated that a reduction in microglial reactivity was associated with improvements in learning and memory in a mouse model of DS. ⁴¹ We found that Ts65Dn trisomic mice displayed an increase in microglial soma size, which is associated to a reactive type of microglia, ⁷² in agreement with previous studies in microglial populations in the Ts65Dn model. ⁷³ Interestingly, rimonabant‐treated Ts65Dn trisomic mice that showed similar recognition memory performance than control mice, exhibited similar reduced microglial somatic areas as control mice. These findings suggest a link between improved memory function and reduced microglial reactivity after rimonabant treatment.

To complement this structural observation with functional evidence, we profiled inflammation‐related proteins in plasma using a high‐sensitivity proximity extension assay increasingly used in biomarker studies of neurodegeneration. ⁷⁴ , ⁷⁵ Although plasma represents a peripheral matrix, accumulating evidence indicates that cytokine levels in blood can reflect neuroimmune status and correlate with central nervous system pathology in both AD and DS. ⁷⁶ , ⁷⁷ , ⁷⁸

Four inflammatory mediators (IL‐1β, CCL4, IL‐4, and IL‐16) exhibited consistent elevation in vehicle‐treated Ts65Dn mice and were significantly reduced by rimonabant, returning to near‐control levels. All four are strongly implicated in microglial function and neurodegenerative processes. Notably, IL‐1β levels have been reported to be up to 10‐fold higher in individuals with DS compared to sporadic AD, ⁷⁹ reinforcing its relevance as a marker of DS‐associated neuroinflammation. IL‐1β and CCL4 are canonical M1‐associated factors involved in amplifying neuroinflammation and promoting cell recruitment, ³⁷ , ⁸⁰ while IL‐4 supports M2 polarization, and IL‐16 acts as a chemoattractant for CD4+ T cells during neuroinflammatory progression. ⁴² , ⁴⁴ The coordinated normalization of these mediators supports a functional reprogramming of microglia toward a less reactive state.

To integrate these changes, we computed a composite inflammation z score, a strategy used in clinical cohorts to capture multidimensional immune dysregulation. ⁸¹ , ⁸² This score showed a trend toward elevation in Ts65Dn mice and was significantly reduced by rimonabant. Importantly, it negatively correlated with novel object recognition performance, reinforcing the link between neuroinflammation and cognitive decline. These findings are in line with preclinical studies in DS models showing that modulation of IL‐1β or other inflammatory pathways can rescue memory deficits. ³⁷

Altogether, these results invite the hypothesis that CB1R blockade reduces microglial reactivity not only morphologically but also functionally. The use of a peripheral, translationally aligned approach provides additional value, particularly given the distinct neuroimmune profile observed in DSAD compared to sporadic AD. ³⁸ This highlights the relevance of targeting CB1R as a strategy to modulate neuroinflammation in DS‐related neurodegeneration.

Furthermore, it would be interesting to extend studies as the present one to new DS models with improved construct validity. ⁸³ While the Ts65Dn mouse model has been instrumental in advancing our understanding of DS‐associated neurodegeneration and cognitive decline, it presents some genetic limitations. In particular, Ts65Dn mice harbor a segmental trisomy that includes not only orthologs of human chromosome 21 on mouse chromosome 16, but also genes from a non‐syntenic region on mouse chromosome 17. ⁸⁴ To address these limitations, genetically refined models such as the Dp(16)1Yey mouse line have been developed, in which triplication is restricted to the Hsa21‐orthologous region on mouse chromosome 16. ⁸⁵ Incorporating such models into future research could provide deeper insights into the mechanisms underlying cognitive impairment and the potential benefits of CB1R inhibition.

Our results need to be interpreted as a mechanistic exploration of CB1R inhibition. Although rimonabant is known as a specific CB1R antagonist, ³³ is not a suitable candidate for clinical translation. Its historical use as Acomplia for obesity treatment led to market withdrawal due to significant psychiatric side effects in certain patients. ⁸⁶ , ⁸⁷ These adverse effects are primarily attributed to the high doses required for its anti‐obesity properties, which convey inverse agonist characteristics on CB1R and may contribute to such negative outcomes. ⁸⁸ , ⁸⁹

In this context, there is a clear need for new pharmacological agents that modulate the endocannabinoid system safely, avoiding the inverse agonist profile that led to the adverse effects of drugs like rimonabant. AEF0217, developed by Aelis Farma, represents a promising example of this approach. It is a novel CB1R modulator, targeting cognitive deficits in conditions such as DS. Phase I clinical trials (NCT05748405) have shown that the compound is safe and well tolerated in healthy subjects.

Interestingly, our findings indicate that CB1R remains upregulated in older individuals, suggesting that targeting this receptor may still offer cognitive benefits for populations beyond the current trial age range. This opens the door to potential future studies involving older patients who experience neurodegenerative changes.

At present, there are no approved treatments for intellectual disability in DS, despite the extensive preclinical studies that have been conducted. ⁹⁰ In our current study, we maximized the potential translational value of our experimental approach by taking numerous variables into consideration. First, we studied the status of CB1R in DSAD human hippocampus, and we found interesting parallelisms with the preclinical models. Second, we performed our study in the most used preclinical model for DS, the Ts65Dn, for which predictive validity was recently proven for novel experimental approaches to enhancing memory performance in subjects with DS. ⁹¹ , ⁹² In addition, the Ts65Dn mouse model is the only model in which the neurodegenerative phenotype is widely described, necessary for the assessment of long‐term effects of our treatment. ¹¹ , ¹² Third, we assessed rimonabant efficacy in male and female mice, presenting comparable sex results. Altogether, our results reinforce the potential interest of decreasing CB1R activity to maintain cognitive function and prevent specific neurological deficits associated with neuroinflammation in DS and this expands our understanding of the potential use of CB1R inhibition for long‐term periods in the treatment of this disorder.

AUTHOR CONTRIBUTIONS

Anna Vázquez‐Oliver participated in experimental design; conducted and analyzed behavioral, biochemical, immunofluorescence and immunohistochemical experiments; and wrote the manuscript. Silvia Pérez‐García analyzed immunofluorescence experiments and revised the manuscript. Rafael Romero‐Pérez conducted and analyzed biochemical experiments and revised the manuscript. Nieves Pizarro conducted and analyzed biochemical experiments and revised the manuscript. Diana Galarraga‐Shinin analyzed immunohistochemical experiments and revised the manuscript. Laura Molina‐Porcel conducted and supervised analysis of immunohistochemical experiments and revised the manuscript. Rafael Maldonado participated in the supervision and experimental design, funded the project, and revised the manuscript. Andrés Ozaita conceptualized, participated in experimental design, supervised, funded the project, and wrote the manuscript. All authors reviewed and approved the final version of the manuscript

CONFLICT OF INTEREST STATEMENT

R.M. and A.O. declare intellectual property of the patent PCT/EP2013/055728. The remaining authors declare no conflicts of interest. Author disclosures are available in the supporting information.

CONSENT STATEMENT

All brain tissue samples were obtained from patients after they or their legal representatives gave written informed consent for the use of their brain tissue and medical records for research purposes, as approved by the ethics committee of our institution, in accordance with the Declaration of Helsinki.

Supporting information

Supporting Information

Supporting Information

Supporting Information

Vázquez‐Oliver A, Pérez‐García S, Romero‐Pérez R, et al. Targeting dysregulated CB1 receptors in a Down syndrome mouse model improves neurological outcomes. Alzheimer's Dement. 2025;21:e70874. 10.1002/alz.70874