Chapter 3. Functional screening of autism-associated missense variants in C. elegans
3.5 Discussion
0.0375. N2 of the second batch = 169.9 ± 8.2; efr-3(D497G) = 126.9 ± 8.2, p = 0.0138. All strains were compared to the wild-type controls of the same experimental condition.) Missense mutants in the gene expression and regulation clusters did not show any difference in speed. As for the measurement of reversal rate, which we defined as the frequency of omega turns within the tracking duration, we detected higher reversal rates in two missense mutant strains, tbx-8(K31E) and dpy-18(Y413C) (N2 of the first batch = 14.4 ± 0.3, tbx-8(K31E) = 19.4 ± 1.1, p = 0.001; N2 of the second batch = 8.9 ± 0.5, dpy-18(Y413C) = 15.8 ± 2.1, p=0.001). Overall, these data suggested five ASD-associated missense mutant strains with movement phenotypes, most of whom belonged to the synaptic function gene cluster.
converts inosine monophosphate (IMP) to adenosine monophosphate (AMP), and missense variant in the adenosine monophosphate deaminase 1 (AMPD1) gene, which recycles AMP to IMP, displayed the opposite effect on fecundity (Camici et al., 2018). Overall, we identified 18 functional changing missense loci, especially 8 of them caused multi-system functional changes.
We prioritized these phenotypic changing missense variants for future studies on functional validation and disease mechanisms.
ASD is a heterogeneous disease with functionally diverse genes and various clinical descriptions. The relationship between genotype and functional outcomes can be defined on the basis of phenotype profiles of multiple genes across a standard battery of behavioral assays (Iakoucheva et al., 2019). Indeed, the genotype-phenotype correlation was used to distinguish the clinical subtype in common diseases (Luo et al., 2019), and factoring the analysis of phenotype data was proposed to further categorize clinical subtypes of ASD (Georgiades et al., 2013). With this in mind, recent studies applied a standard battery of assays to characterize ASD-associated genetic variants in various model systems in a comprehensive manner (Chen et al., 2018, 2020; Deneault et al., 2018; McDiarmid et al., 2020; Wong et al., 2019). Taking a similar approach, we conducted a set of behavioral assays to missense mutants in two gene clusters. We discovered that missense mutations in different gene clusters contributed to distinct phenotypic profiles. Moreover, we identified a proportion of missense variants causing phenotypic changes in multiple assays, indicating physiological changes at the system level.
According to the genomic analysis in humans, the most highly constrained variants showed higher expression levels and broader tissue expression (Ardlie et al., 2015; Lek et al., 2016).
Our results suggested these missense loci may locate in the constraint region that were less tolerant to sequence changes.
We used two standard functional assays to screen the phenotypic changing missense variants. In the fecundity assay, most phenotypic changing missense variants belonged to the gene expression and regulation cluster. This result indicated that changes in genes involved in chromatin remodeling and nucleotide cycle impacted developmental functions significantly. In fact, some of the genes have been documented to affect organism survival. For instance, histone demethylase activity of Utx was shown to impact viability in Drosophila (Copur & Müller, 2018). Similar to our finding of reduced fecundity in the exc-7 missense mutant strain, the Drosophila fnenull mutant, one of the paralogs in ELAV family, also displayed reduced fecundity (Zanini et al., 2012). In addition, we examined the contribution of missense variants in the synaptic function cluster. We discovered that genes in the synaptic function cluster have more significant impacts on morphology and locomotor patterns. The effect in size was consistent with previous large-scale analyses of inactivating mutations in constrained genes in the humans genome and ASD-associated models in C. elegans (Ganna et al., 2018; McDiarmid et al., 2020).
We detected a change in one of the constraint genes, unc-73, which is the C. elegans ortholog of human TRIO. Consistently, loss-of-function of TRIO impaired motor coordination and synaptic function in rodents and individuals with mild intellectual disability and (Ba et al., 2016;
Katrancha et al., 2019). It is proposed that a single genetic variant can be widespread through inter-connecting gene networks and have emergent effects on specific phenotypes (Iakoucheva et al., 2019). Our findings proved this concept and highlight the importance of phenotypic specificity in different gene networks and fine-grained dimensional phenotyping.
The problem of a large amount of genetic variants with unknown significance presents a massive challenge for ASD studies. We attempted to solve this problem with a multi-cellular screening platform in C. elegans. We introduced the disease-associated missense variants in the endogenous C. elegans protein. This approach allows us to inspect the functional consequence of missense variants in its original genomic and cellular content, eliminating the effect caused by
cell-type-specific chromatin states (Ernst & Kellis, 2015; Starita et al., 2017). Besides, using the endogenous C. elegans protein preserves the original content, such as protein-protein interactions, intronic regulation, and isoform balance (Reble et al., 2018; Robison, 2014). The endogenous protein approach can avoid possible confounding effects such as mismatched heteromeric complexes presented in the “humanized” models as well (Bend et al., 2016; Prior et al., 2017) (Baruah et al., 2017; McDiarmid et al., 2018; Walsh et al., 2017). Furthermore, the short lifespan and well-developed genetic tools allow rapid generation of targeted mutation and double mutants before embarking on less efficient and more costly animal models (Markaki &
Tavernarakis, 2010). Overall, our approach and model organism facilitates the dissection of the mechanisms of pathological conditions and drug target identification (Schmeisser et al., 2017).
Using C. elegans as a model to study psychiatric disorders has some limitations. Unlike rodent models, the C. elegans model lacks complicated behaviors. However, C. elegans and humans share essential physiological pathways (e.g., insulin signaling, Ras/Notch signaling, p53, and many miRNAs), neurotransmitter systems, and receptor pharmacology (Engleman et al., 2016; Markaki & Tavernarakis, 2010). The disruption of protein functions can be reflected in multiple well-developed behavioral assays, such as locomotion and habituation learning.
Another limitation is that the evolutionary distance between humans and C. elegans restricts the subject to C. elegans orthologous genes. Fortunately, C. elegans and humans share 60% of gene orthologs (Kaletta & Hengartner, 2006; Shaye & Greenwald, 2011) and disease-associated genes are particularly conserved throughout evolution (López-Bigas & Ouzounis, 2004; Shpigler et al., 2017). Furthermore, for genetic candidates that show correlated expression, C. elegans models facilitate the study of interaction between missense variants by rapidly generating double or multiple missense mutation models.
With the accelerated identification of disease-associated variants, the challenge of interpreting their functional outcomes persists. To date, most preclinical research is biased toward experimentally well-accessible genes due to our limited knowledge of disease mechanisms (Stoeger et al., 2018). With the advance in data science, broad-scale phonemic analysis will be the future trend. Our study proved that missense mutants in different gene clusters displayed distinct phenotypic profiles and laid the foundation for future systematic characterization of disease-associated variants.