Prenatal Treatment of Down Syndrome to Improve Brain Development and Neurocognition

National Human Genome Research Institute

Investigators

Linked publications, trials & patents

Abstract

During the past year, we achieved the following objectives for each of our goals: 1) Molecular and cellular phenotyping of iPSCs and iPSC-derived neural progenitor cells (NPCs) from individuals with T21 and age and sex matched euploid individuals. We have generated a novel, genetically diverse panel of age- and sex-matched T21 and Eup induced pluripotent stem cells (iPSCs) and NPCs for in vitro modeling and have performed cellular and transcriptomic analyses to characterize these lines. Our panel of cell lines is larger than those used in previous studies, thus providing the ability to ascertain phenotypic and transcriptomic variability across a diverse dataset. NPCs were used to generate stable cell lines expressing a fluorescent nuclear marker (NucLight Red) for use in live-cell imaging and drug screening. To gain better insights into the molecular mechanisms underlying atypical brain development in fetuses with DS, we performed transcriptome analyses on fibroblasts, and both transcriptome and proteome analyses on iPSCs and NPCs. Pathway analyses showed dysregulation across the cell cycle, DNA damage/repair, inflammation, oxidative stress, mitochondrial dysfunction, and oxidative phosphorylation. We also performed single cell RNA-Seq on NPCs to extend our knowledge of the transcriptomic effects of T21. Assays were performed to measure mitochondrial function, oxidative stress and antioxidant capacity. 2) Screening of NPCs for therapeutic responses to drugs identified using the Connectivity Map (CMap) and Library of Integrated Network-Based Cellular Signatures (LINCS) databases. We propose that NPCs from individuals with T21 will replicate brain differences in DS, and that T21 gene expression signatures paired with bioinformatics tools, such as the Connectivity map (Cmap) and the Library of Integrated Network-Based Cellular Signatures (LINCS), can be used to repurpose compounds for prenatal therapies to improve neurocognition. The Cmap and LINCS databases contain small molecule-induced gene expression signatures, and these signatures can be queried with dysregulated transcriptomic signatures to predict drugs that would rescue altered gene expression. We used transcriptome data from T21 and Eup NPCs to query the CMap and LINCS databases for safe drug candidates that can rescue T21-associated transcriptomic changes. During the past year, we have screened 35 drug candidates at a range of concentrations across our diverse panel of NPCs, evaluating cytotoxicity and efficacy in improving the reduced proliferation rate that we have observed in T21 NPCs. Several candidates show promise for improving NPC growth and will be further evaluated in vitro and in mouse models (4, below), while others show consistent toxicity, and thus have been eliminated from consideration. 3) Molecular, cellular and behavioral phenotyping of mouse models of DS to discover phenotypes in each model that mimic those present in DS, with an emphasis on the embryo and placenta because of our interests in prenatal treatment. We have completed deep phenotyping across the lifespan of four mouse models of DSDp(16)1/Yey, Ts65Dn, Ts1Cje, and Ts66Yah. These models are trisomic for overlapping mouse chromosome 16 (Mmu16) regions that are orthologous to Hsa21 but are cytogenetically distinct from each other. Interestingly, the Dp(16)1/Yey, Ts65Dn, Ts1Cje, and Ts66Yah mouse models each exhibit distinct prenatal gene expression and postnatal phenotypes, however, they share several dysregulated pathways that could be targeted in future prenatal therapies, including neuroinflammation, interferon signaling, and oxidative stress response. This past year we published a study describing brain transcriptomic and behavioral analyses comparing Ts65Dn with a new mouse model of DS, Ts66Yah. Ts65Dn mice are trisomic for genes from Mmu17 that are not syntenic to Hsa21 and thus not relevant to DS, while Ts66Yah mice have had the extraneous Mmu17 region removed by CRISPR/Cas9 engineering. Therefore, Ts66Yah more precisely models the genomics of T21 in comparison to Ts65Dn. We showed that Ts66Yah and Ts65Dn exhibit distinct gene expression patterns and behavioral phenotypes, suggesting that the extraneous triplicated genetic material in Ts65Dn contributes to this models more severe phenotypes. This study raised fundamental questions about the suitability of Ts65Dn as a mouse model of DS, which is highly significant because most preclinical DS therapy studies have used this model. Further, our results may explain why preclinical trials that have primarily used Ts65Dn have unsuccessfully translated to human therapies. Overall, our studies have provided crucial quantitative data for the DS research community that will guide future experimental design and mouse model selection. We also published a study this year identifying quantitative differences in embryonic body weight, placenta, and brain phenotypes across four mouse models of DS (Dp(16)1/Yey, Ts65Dn, Ts1Cje, and Ts66Yah), and used a novel statistical approach to show that trisomic mice could be separated into mild and severe phenotypic classes across these four cytogenetically distinct models. Our correlation of placental and embryonic measurements with distinct classes of trisomic phenotypic severity in mouse models of DS suggests that these quantitative measurements could be extrapolated to human ultrasound imaging studies, providing a potential new approach to study fetal development, predict phenotypic severity, and guide future prenatal treatments for DS. During the past year we published our pilot touchscreen studies on Ts1Cje, Ts65Dn, and Dp(16)1/Yey mice. Non-verbal touchscreen-based tasks are widely used in cognitive and behavioral tests to assess intellectual disabilities in humans with DS, and humans with DS exhibit learning deficits in the Cambridge Neuropsychological Test Automated Battery (CANTAB). Therefore, we translated the CANTAB Visual Distinction (VD) and Extinction tasks using rodent touchscreen behavioral testing to investigate visual discrimination learning and inhibitory control in three DS models. We found significant differences in each model relative to euploid, but also found that hyperactive behavior and strain genetic background influenced successful learning in touchscreen behavioral testing. This study provides important information to guide appropriate DS mouse model selection in future tests of novel therapeutics. (4) Administration of promising candidate drugs to the best mouse model of DS and evaluation of their safety and efficacy. Our comprehensive analyses of four mouse models of DS highlight Ts66Yah as the most promising model for placental, embryonic, and candidate drug response studies. Studies on promising drug candidates from our NPC screens will begin once the previous goals are accomplished.