Blocking the Brain’s Cannabis Receptor Repairs Faulty RNA Splicing in Fragile X Mice

Fragile X syndrome, the most common inherited cause of intellectual disability and the leading single-gene form of autism spectrum disorder, has long been traced to a single missing protein. Yet a new study published in the journal iScience suggests that the damage it inflicts on the brain runs far deeper than previously appreciated, reaching all the way down to how RNA molecules are cut and pasted at the synapse. Researchers in Spain report that blocking the brain’s cannabinoid type-1 receptor with a very low dose of the drug rimonabant can reverse a striking array of these molecular defects in mice, offering a previously unknown explanation for how the treatment improves cognition in fragile X.

The syndrome arises from an expansion of CGG repeats in the fragile X messenger ribonucleoprotein 1 gene, or FMR1, on the X chromosome. This expansion silences the gene through hypermethylation, eliminating production of its protein, FMRP. That protein is an RNA-binding molecule with a critical job: it shuttles to both pre- and postsynaptic compartments of neurons, where it fine-tunes local mRNA translation to maintain synaptic plasticity and connectivity. It also influences transcription factors, chromatin-modifying enzymes, and other RNA-binding proteins. When FMRP is absent, the consequences ripple outward, producing the mild cognitive deficits, hyperactivity, and autistic-like behaviors seen in both patients and the widely used Fmr1 knockout mouse model.

Previous work from the same group had already shown that inhibiting CB1, the main receptor of the endocannabinoid system, improves cognitive performance in fragile X mice. Low doses of rimonabant, a selective CB1 antagonist, resolved the aberrant mGluR5-dependent long-term depression and the abnormal hippocampal spine density characteristic of the model. What remained unknown was the molecular machinery underlying those improvements. The new study, led by Lucía de los Reyes-Ramírez and Andrés Ozaita of the Universitat Pompeu Fabra in Barcelona, together with bioinformatician Eduardo Eyras, set out to find that machinery by looking somewhere no one had looked before: inside the synaptoneurosomes, the tiny sealed vesicles that capture both sides of a neuronal connection.

Synaptoneurosomes are essentially pinched-off synaptic terminals, preserving the pre- and postsynaptic machinery in a single preparation. Because FMRP is highly concentrated at synapses, the team reasoned that these structures would offer a far more sensitive window into fragile X pathology than bulk brain tissue. To test this, they split single hippocampi from wild-type mice into two fractions, one processed as bulk homogenate and one enriched for synaptoneurosomes, and sequenced the RNA from both. The fractionation method, not the individual mouse, emerged as the dominant factor shaping the transcriptome. Compared with bulk tissue, synaptoneurosomes were enriched for 749 genes and depleted of 1,019, with the enriched set dominated by Gene Ontology terms related to synapses, transmembrane transport, and mitochondrial activity. More than half of the enriched biological processes were tied to synaptic organization, synaptic plasticity, and mitochondrial bioenergetics, confirming that the preparation genuinely concentrates the synaptic transcriptome.

With that foundation in place, the researchers turned to fragile X mice. Comparing synaptoneurosomes from knockout animals with those of wild-type littermates revealed 29 upregulated and 19 downregulated genes, including changes in genes linked to dendritic arborization, ERK signaling, synaptic transmission, and mTORC1 signaling. The team then treated fragile X mice with rimonabant at 0.1 milligrams per kilogram for seven days, a dose and schedule previously shown to improve cognitive performance and synaptic plasticity. The drug modified the expression of 14 genes in fragile X synaptoneurosomes, and crucially, three of them, Mff, Hnrnpd, and Slain2, moved in the opposite direction to the fragile X signature, effectively reverting toward normal. Quantitative PCR in a larger independent cohort confirmed the reversal for Mff and the fragile X downregulation of Hnrnpd, an RNA-binding protein involved in synapse formation and plasticity. Strikingly, none of these changes appeared when the same animals’ bulk hippocampal tissue was analyzed, underscoring how easily synaptic signals are diluted in whole-tissue preparations.

Because Hnrnpd encodes a direct regulator of mRNA, the team next examined the data at the level of individual transcripts rather than genes, a finer-grained approach that captures the different versions a single gene can produce. This differential transcript expression analysis revealed 254 transcripts altered in fragile X synaptoneurosomes, with enrichment in synapse structure and organization and, notably, in mRNA processing and RNA splicing. Splicing factors such as Celf2 and members of the Hnrnp family were among the downregulated transcripts. When rimonabant was given to fragile X mice, 211 transcripts changed, and the upregulated set was enriched for synapse organization, mRNA splicing, and neuron differentiation, including transcripts from spliceosome-associated genes such as Samd4b, Acin1, and several Hnrnp family members. In wild-type mice, rimonabant produced a completely different and functionally unremarkable pattern, demonstrating that the drug’s effects on the synaptic transcriptome are genotype dependent.

The overlap between the two comparisons was the study’s most striking quantitative result. Of the 254 transcripts altered in fragile X synaptoneurosomes, 34 percent shifted in the opposite direction after rimonabant treatment. More than half of the transcripts downregulated in fragile X animals, 59 out of 109, were restored by the drug, a statistically robust rescue confirmed by hypergeometric testing. Every Gene Ontology term associated with those rescued transcripts related to mRNA processing and splicing, reinforcing the idea that rimonabant ameliorates fragile X synaptic defects by acting on the splicing machinery itself.

That hypothesis was put to a direct test by mapping the alternative splicing landscape, the pattern by which exons are included, skipped, or reconnected to generate different protein isoforms. The absence of FMR1 dysregulated the splicing of 206 genes through 259 alternative splicing events, with alternative first exons and skipped exons the most prevalent categories. Among the affected genes were Cpeb2, FosB, Arf4, Cacng5, and Vdac3, all implicated in synaptic transmission or morphology. Rimonabant modified 178 splicing events in 130 genes in fragile X synaptoneurosomes, and 77 events across 56 genes were common to both the genotype and treatment comparisons, moving in opposite directions. Cross-referencing those 56 genes with a single-cell hippocampal atlas showed that roughly two-thirds are predominantly expressed in neurons, and the affected genes spanned RNA processing, cytoskeleton modulation, ion channels, neuronal development, and mitochondrial function. Most of these events were not significantly altered in rimonabant-treated wild-type mice, indicating a genetic context-specific rescue.

To validate these findings independently, the team designed custom TaqMan probes for three RNA-binding proteins, Ddx6, Celf1, and Cpeb2, and measured both the inclusion and exclusion isoforms of each splicing event in a new, larger set of samples. Ddx6 exon inclusion was downregulated in fragile X synaptoneurosomes and restored by rimonabant, with matching results for exon exclusion. Celf1 transcripts were significantly more abundant after treatment, and Cpeb2 exon inclusion, reduced in fragile X animals, was restored by the drug. The Cpeb2 event was particularly compelling: inclusion or exclusion of a 90-base-pair exon yields two isoforms that differentially regulate RNA translation, and the inclusion isoform is linked to translation activation while the exclusion isoform represses it. Because Cpeb2 controls the translation of neuron-specific mRNAs essential for synaptic plasticity, the researchers propose that its splicing rescue could help normalize the excessive local protein synthesis thought to underlie fragile X synaptic deficits. None of these isoform changes appeared in bulk hippocampal homogenates.

The authors are careful to frame the mechanism as indirect. Given the genes bearing splicing events, FMRP is unlikely to be a direct splicing regulator; instead, it appears to mediate RNA processing by controlling other downstream RNA-binding proteins, many of them brain-specific, that in turn govern synaptic transmission. By restoring the splicing patterns of proteins like Celf1 and Cpeb2, CB1 inhibition may trigger a cascade that normalizes the synaptic microenvironment, potentially including the excitatory-inhibitory balance that is a common feature of autism and fragile X. The study has limitations: only male mice were used, the enriched synaptic fraction may contain other cell types, rimonabant was withdrawn from clinical use and remains an experimental tool, and the splicing changes detected by high-throughput sequencing were subtler when validated by qPCR. Even so, the work establishes that the endocannabinoid system actively modulates the synaptic splicing machinery, and that a previously invisible layer of RNA regulation at the synapse may be a critical step in rescuing fragile X phenotypes. Further studies will determine whether these RNA changes translate into specific protein isoforms capable of restoring fragile X synaptic physiology in humans.

Subject of Research: Restoration of synaptic alternative splicing by CB1 receptor inhibition in a fragile X syndrome mouse model

Article Title: CB1 receptor inhibition restores synaptoneurosomal alternative splicing landscape in fragile X syndrome mice

Article References: Reyes-Ramírez, L. D. L., Ciaran-Alfano, L., Reixachs-Solé, M., Bergadà-Martínez, A., Galera-López, L., Martínez-Torres, S., Navarro-Romero, A., Maldonado, R., Eyras, E., & Ozaita, A. (2026). CB1 receptor inhibition restores synaptoneurosomal alternative splicing landscape in fragile X syndrome mice. iScience, 29(11), Article 117793. https://doi.org/10.1016/j.isci.2026.117793

Image Credits: AI Generated

DOI: Not provided

Keywords: fragile X syndrome, CB1 receptor, rimonabant, alternative splicing, synaptoneurosomes, FMRP, endocannabinoid system, RNA-binding proteins, hippocampus, autism spectrum disorder, transcriptomics, synaptic plasticity