Abstract
Montani C, Iovino L, Di Vetta F, Rene’ Pasquin Mariani J-C, De Guzman AE, Gini S, Galbusera A, D'Epifanio B, Ghirardini E, Cornuti S, Dadà L, Putignano E, Alessandrì MG, Vasirani G, Bertozzi SM, Armirotti A, Baroncelli L, Gozzi A. Brain. 2025:awaf275. doi: 10.1093/brain/awaf275. Online ahead of print. Creatine Transporter Deficiency (CTD) is an X-linked disorder due to the loss of SLC6A8 gene and presenting with low brain creatine, intellectual disability, autistic-like behavior and seizures. No treatments are available yet for CTD, and little is known about the brain circuit alterations underlying its pathological endophenotypes. Here, we tracked brain network and behavioral dysfunction in a murine model of CTD at two stages of disease progression. fMRI mapping revealed widespread disruption of brain connectivity in Slc6a8-KO mice, with robust somatomotor hypoconnectivity in juvenile animals, and weaker and more focal alterations of cortical and subcortical connectivity in adulthood. Notably, perinatal AAV-mediated expression of human SLC6A8 in Slc6a8-KO mice robustly prevented juvenile fMRI hypoconnectivity, an effect accompanied by the regression of multiple translationally relevant phenotypes, including reduced stereotyped movements, improved declarative memory and increased body weight, all of which persisted into adulthood. However, early cognitive deficits, impairments in working memory and residual fMRI hypoconnectivity in adult mice were not ameliorated by gene therapy. Furthermore, significant cognitive impairments were observed in WT mice receiving gene therapy, highlighting a potential detrimental effect of ectopic expression of SLC6A8 in healthy brain circuits. Finally, multivariate modeling in adult mice revealed a basal forebrain network whose activity was associated with behavioral performance and modulated by brain creatine levels. This brain-behavior relationship was disrupted in Slc6a8-KO mice. Our results document robust network disruption in CTD and demonstrate that CTD pathology can be partially alleviated by perinatal genetic expression of SLC6A8, providing a foundation for the future development of experimental therapies for this genetic disorder.
Commentary
Recent developments in gene therapies have made real the possibility of developing highly targeted cures for a wide range of genetic disorders. There are numerous well-studied genetic epilepsies, involving mutations to a single gene, that are particularly attractive candidates for gene therapy. For example, heterozygous sodium channel gene mutations produce a mixture of wild-type (WT) sodium channels and toxic variants, which lead to severe developmental epileptic encephalopathies in patients. Antisense oligonucleotides are showing promise both in animal models 1 and recent human trials 2 for targeting these toxic exons, reducing the disease-causing mutant channels without affecting the WT gene expression. CRISPR-based therapies are somewhat behind in the translational pipeline, perhaps due to perceived risk associated with the permanence of their gene-editing mechanism, but also hold the potential to produce something close to a true cure, wherein a one-time treatment fixes the mutant DNA. 3
Many genetic epilepsies are caused by loss-of-function mutations, which, rather than producing toxic variants of functional proteins, result in nonfunctional proteins. Adeno-associated virus (AAV)-based therapies seek to compensate for such mutations by using virally delivered genes to (over)express the deficient protein. As with CRISPR-based approaches, AAV-mediated overexpression therapies are typically administered 1 time and have a permanent effect on the patient.
While it seems intuitive that monogenic epileptic disorders would be “low-hanging fruit” for gene therapies, much remains unknown about how to optimally implement this relatively immature technology. The ultimate impact of gene therapies is not always straightforward to predict. Fixing the gene does not necessarily fix the disease phenotype, particularly for genetic disorders with developmental implications. Furthermore, the impact of the treatment depends on timing, dosing, and targeting. Given the irreversible nature of some of these therapies, it is necessary to use biofidelic animal models to maximally optimize putative treatments for safety and efficacy.
In the highlighted study, 4 Montani et al perform such foundational work toward the development of a treatment for creatine transporter deficiency (CTD), a rare X-linked disorder involving loss of the Slc6a8 gene, which encodes the protein responsible for cellular creatine uptake. CTD leads to low brain creatine followed by intellectual disability, autistic-like behaviors, and severe seizures. In human CTD patients, the electroencephalogram is characterized by decreased theta/alpha power and increased gamma activity, 5 while neuroimaging reveals white matter demyelination and corpus callosum thinning. This indirect evidence leads the authors to hypothesize that long-range functional connectivity may underlie aspects of the CTD neurological phenotype.6,7
Slc6a8 knockout (KO) mice reproduce key phenotypes of CTD, including low brain creatine, decline in performance of cognitive tests, progressive relative bodyweight decline, and emergence of autistic-like behaviors. To test whether the emergence of behavioral phenotypes was correlated with the hypothesized changes in long-range connectivity, longitudinal measurements of functional connectivity were taken using resting-state functional magnetic resonance imaging (fMRI) under light anesthesia. Indeed, KO mice exhibited progressive fMRI hypoconnectivity at postnatal day 40 (P40) and P140. One caveat with such measurements is that anesthesia affects functional connectivity, and anesthesia sensitivity may be different in a model of neurological disease. The authors partially control for this using minimal alveolar concentration testing to evaluate halothane sensitivity in the 2 populations, but ultimately, a differential effect of isoflurane on functional connectivity cannot be ruled out. Any such artifact could impact connectivity comparisons between WT and KO mice, but would likely not affect within-genotype evaluation of therapeutic interventions.
In the gene therapeutic arm of this study, the deficit in cellular creatine uptake was countered with AAV-based overexpression of the hSlc6a8 gene, using the nonspecific, high-level promoter JeT. Following intracerebroventricular injection of AAV on postnatal day 1, a battery of assays was performed at juvenile and adult timepoints, including measurements of brain creatine levels, body weight, behavior, and functional connectivity. Ultimately, the intervention had a mixture of partial effects on the disease phenotype in Slc6a8-KO mice. The gene therapy partially restored brain creatine and modestly increased body weight in KO mice at all ages measured. AAV-treated animals had reduced hypoconnectivity at P40, but by P140, treated KO animals had similar functional connectivity to untreated KO animals. Conversely, the effect of treatment on performance of the behavioral tasks was not apparent until the later time point, when there was a reduction in autistic-like behaviors (elevated grooming frequency) by P140 and an improvement in performance of the object recognition test by P100.
Taken together, it is not immediately obvious from the results of these experiments how best to optimize gene therapy for CTD. One interpretation might be that postnatal day 1 is too late to treat a neurodevelopmental disorder such as CTD, since significant pathological brain development will have already taken place by the time of birth. Elevating the brain creatine beginning at P1 may not undo the pathological circuitry that has already been established. In this case, in utero introduction of gene therapy may improve the efficacy. Alternatively, perhaps the partial restoration of brain creatine levels (to roughly 50% of WT controls) is insufficient to completely ameliorate the CTD neurologic phenotypes, suggesting that the therapy may be improved by increasing the level of overexpression of hSlc6a8. However, the finding from the current study that overexpression of hSlc6a8 in WT animals decreases performance in the Y-maze tasks suggests that caution should be exercised in maximizing expression levels. One finding that may help to guide the optimization efforts is that functional connectivity maps correlate with both brain creatine levels and behavioral performance. Thus, functional connectivity measurement may represent a biomarker for efficacy, reducing the dependence on laborious behavioral assays or postmortem brain creatine measurements. Functional connectivity can also be safely measured in human patients and thus may be useful in clinical translational efforts.
One allure of gene therapies is that we will soon be able to treat the underlying cause of disease, rather than simply developing or identifying treatments that improve symptoms. The highlighted study illustrates nicely the fact that fixing or replacing a gene defect may not be sufficient to cure a genetic disorder. Rather, these rapidly developing therapeutic interventions represent new tools that must be carefully evaluated and optimized in preclinical models. Ultimately, for a developmental disorder such as CTD, the optimal therapeutic window may not be clinically feasible, and adjuvant therapies will need to be developed to ameliorate pathologies that have already been established by the time of treatment. In their work, Montani et al provide critical fundamental knowledge toward developing a novel gene therapy for CTD.
