Complex genetic interaction patterns determine the phenotypic trajectory of neurodevelopmental disorders
Open Access
- Author:
- Jensen, Matthew
- Graduate Program:
- Bioinformatics and Genomics (PhD)
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- August 10, 2021
- Committee Members:
- Dajiang Liu, Outside Unit & Field Member
Ross Hardison, Major Field Member
Santhosh Girirajan, Chair & Dissertation Advisor
Reka Albert, Major Field Member
George Perry, Program Head/Chair - Keywords:
- Neurodevelopmental disorders
Copy-number variants
Genomics
Transcriptomics
Drosophila melanogaster
Whole genome sequencing
Network analysis - Abstract:
- Neurodevelopmental disorders, such as autism, schizophrenia, and intellectual, disability, are complex human disorders that have been associated with hundreds of genes and rare and common genomic variants. Rare copy-number variants (CNVs), or duplications and deletions spanning multiple genes, are one of the most common causes of neurodevelopmental disorders, accounting for approximately 10% of all cases. In contrast to CNVs associated with syndromic disorders and invariable sets of phenotypes, many rare CNVs exhibit extensive phenotypic heterogeneity among affected individuals. These variably-expressive CNVs show pleiotropic effects towards multiple comorbid disorders with incomplete penetrance, suggesting that these variants may sensitize the genome for a range of neurodevelopmental outcomes but do not determine the ultimate phenotypic trajectory by themselves. For example, the 520-kbp 16p12.1 deletion is enriched among affected children with developmental delay and other severe early-onset phenotypes. However, the deletion is inherited in >90% of cases from parents with later-onset neuropsychiatric features, and the deletion also contributes to decreased cognition, increased schizophrenia risk, and other clinical traits in the general population. Variability in phenotypic severity and diversity among carriers of rare CNVs has led to challenges in identifying causal genes for the observed developmental phenotypes. Unlike for syndromic CNVs, where individuals with single-nucleotide variants (SNVs) in a single gene have the same phenotypes as those with the entire CNV, a survey of every gene in variably-expressive CNV regions is required to understand the molecular underpinnings of each region. In fact, evidence from functional model studies suggested that several genes in each region are responsible for neurodevelopmental phenotypes, opening the possibility for complex interactions between CNV genes in shared pathways. Furthermore, we previously found that children with 16p12.1 deletion carried an increased number of additional rare CNVs and SNVs elsewhere in the genome, called “second-hit” variants, compared with their carrier parents. We hypothesized that these “second-hit” variants could ultimately be responsible for the early-onset phenotypes observed in the carrier children. However, the contributions of these “second-hit” variants to specific developmental and psychiatric phenotypes has not been explored, and the molecular mechanisms for how CNVs and “second-hit” variants interface with each other remained unclear. In this dissertation, we used multi-disciplinary approaches to assess rare CNVs associated with neurodevelopment for two main genetic sources of phenotypic heterogeneity: interactions among genes within CNV regions, and contributions of “second-hits” elsewhere in the genome. We first assessed neuronal, cellular, and developmental phenotypes for homologs of 26 genes and 879 pairwise interactions within two variably-expressive CNVs, the 16p11.2 and 3q29 deletions, using Drosophila melanogaster and Xenopus laevis functional models along with transcriptome and human brain-specific network analysis. We found that genes within the 16p11.2 deletion exhibit pervasive complex interactions with each other, where many genes (such as MAPK3 and TAOK2) suppress the neuronal phenotypes of other genes within the region. Many of these phenotypes were modulated through cell proliferation pathways, suggesting a shared molecular mechanism for the 16p11.2 deletion. In contrast, we found mostly additive effects or synergistic interactions among genes in the 3q29 deletion region, in particular those between NCBP2 and all other genes in the region. These interactions were mediated through apoptosis pathways, and their developmental phenotypes could be rescued with simultaneous knockdown of apoptosis regulatory genes. Overall, we found two distinct interaction models for CNV pathogenicity: a pervasive complex interaction model involving most genes within a CNV region, and a synergistic model where one gene preferentially interacts with other genes in the region. To better define the roles of “second-hit” variants towards developmental phenotypes, we performed deep clinical phenotyping and whole-genome sequencing to identify a broad spectrum of variants for 416 individuals in 135 families with the 16p12.1 deletion, and further performed RNA-sequencing for lymphoblastoid cell line samples of 32 individuals in five large 16p12.1 families. In line with previous results, we found that children with the 16p12.1 deletion had increased numbers of rare coding and non-coding “second-hit” variants compared with their carrier parents, which were preferentially transmitted from the non-carrier parent. Rare variants inherited from the non-carrier parent were also enriched for gene expression changes, suggesting that they could be more pathogenic than “second-hits” inherited from the carrier parent. However, carrier parents had higher polygenic risk for schizophrenia compared to carrier children, suggesting differential contributions for rare and common variants towards early-onset and late-onset clinical features. We further found that classes of rare and common variants differentially contribute to developmental phenotypes, such as short-tandem repeats (STRs) towards nervous system phenotypes. Except for rare variants and IQ, genotype-phenotype associations differed across individuals carrying rare CNVs and SNVs in other cohorts, suggesting that contributions of “second-hits” to developmental phenotypes are highly dependent on the context of the primary variant. Interestingly, we also found that parents of children with the 16p12.1 deletion had a high degree of assortative mating, which was also observed among carriers of other inherited CNVs but not de novo or syndromic CNVs. This finding could explain why carrier children in families with a stronger history of cognitive features in particular had more rare variants and a more diverse phenotypic presentation. Finally, we found 11 cases of “second-hit” variants that synergistically interacted with the 16p12.1 deletion to alter gene expression in carrier children, providing a putative mechanism for non-additive effects of CNVs and other rare variants towards developmental phenotypes. Overall, our work provides further evidence for a complex multi-genic model of variably-expressive CNVs, which can be broadly applied towards neurodevelopmental and other complex genetic disorders. Further studies are needed to continue exploring complex interactions between primary and “second-hit” genes using more advanced functional models and larger and more diverse cohorts of human patients. These results implicate a need for precision medicine approaches in the clinic to aid individuals who carry variably-expressive pathogenic variants towards diagnosis, management, and potential treatment of these disorders.