Phenotypic landscape of schizophrenia-associated genes defines candidates and their shared functions

Background

Psychiatric disorders are a major healthcare concern, accounting for up to one quarter of worldwide disabilities. There are many challenges to understanding the biological basis of these diseases. First, psychiatric disorders are non-Mendelian; they involve many loci and inheritance patterns are therefore complicated. In 2014, a large genome-wide association study implicated over 100 loci in schizophrenia; each had a small effect. A second challenge is to understand the molecular mechanisms underpinning a psychiatric disorder, whether they be changes in neuronal activity or connectivity, changes in glial cell gene expression or countless other potential pathologies. The third, and perhaps most difficult, challenge is to understand how the molecular pathologies manifest in the context of the brain, the human organism, and outside environmental factors to control a biological output as multifaceted and nuanced as behavior.

An important step on the road to meeting these challenges is to define the roles of genes associated with psychiatric disorders in brain function by using animal models. The zebrafish, Danio rerio, is an exquisite system for this purpose. Brain anatomy, neural subtypes and aspects of development are conserved between humans and zebrafish, while behaviors such as anxiety and sociability can be assessed with high throughput. Imaging of brain activity is also possible in the semi-transparent larval stages. Importantly, most human disease genes are conserved in zebrafish and, now with the advent of CRISPR/Cas, the generation of mutant lines is increasingly tractable.

Overview of the work

Summer Thyme, Alexander Schier and colleagues have used all of these advantages to investigate the potential roles of genes located within schizophrenia-associated loci. In a heroic effort described in the current preprint, the team generated 132 mutant zebrafish lines and assessed them for behavioral defects, brain structural malformations, and perturbations to brain activity. While only a small number of mutants exhibited lethality prior to adulthood, more than half showed brain activity and/or behavioral phenotypes.

The researchers searched for behavioral phenotypes in larvae, kept in multi-well plates so that the experiment could be done with high-throughput, across a 2-day experiment (4-6 days post fertilization [dpf]) (Figure 1). Measures included how much the larvae moved and where they preferred to be in their dish, their sleep behavior and their responses to various stimuli (light flash, dark flash, acoustic startle, heat and others). Larvae were also tested for defects in habituation, the diminishing of a response to repeated stimuli, as well as prepulse inhibition, where a weak pre-stimulus reduces the reaction to a subsequent stronger stimulus. Across the library of mutants assessed, phenodeviants, mutants departing significantly from the wild-type, were discovered in all of these behavioral assays, and many mutants were affected in several tests. This revealed that different aspects of behavior can be impacted, and to different extents, by mutations in schizophrenia-associated genes, something which further underlines the complexity of this disease. 

Next, the library of mutants was assessed for changes in brain activity (where the distribution of phosphorylated ERK was taken as a measure of calcium signaling) and alterations in morphology. This data has been made available in an accessible form at stackjoint.com/zbrain. While brain morphology phenotypes were only present in 12% of mutants, changes to brain activity in free swimming larvae occurred in over half of the mutants. These changes were caused by mutations in genes involved in processes previously associated with schizophrenia, such as calcium channels and synaptic pruning, but many were caused by previously ignored genes of unknown function.

One immediate use of all of this data is in the discovery of strong candidate schizophrenia genes amongst the sometimes many genes at disease-associated loci. The work also extends beyond schizophrenia to other neuropsychiatric disorders: autism and epilepsy genes were also mutated, since there is overlap between the genes associated with these diseases. Certainly, many mutants exhibited interesting phenotypes that warrant further attention. 

Figure 1: Summary of behavioral test results across mutant library (from Figure 2E of the preprint)

What I liked about this preprint

Our understanding of gene function in development, homeostasis and disease has mostly been advanced by manuscripts reporting on the phenotypes of one or a small number of genes at a time. One disadvantage of this approach is that phenotyping is often not standardized; different labs perform different assays or instead perform the same assays but in different ways, making comparative analysis of the effects of distinct genetic lesions complicated. Another issue is that labs naturally look at phenotypes that interest them and miss, or do not report, others. Issues like this are well understood in the mouse community, and consortia such as EUMODIC have established large-scale broad and standardized phenotyping pipelines for mouse mutants. The current preprint takes the similar approach of “many mutants/broad phenotyping”, and some of the interpretations they make were only possible because they studied so many mutants at once with standardized analysis pipelines.

This preprint also demonstrates how useful this approach is. Some unstudied genes gave unexpected phenotypes, whilst previously missed phenotypes were found in some genes that had been studied before. Happily, similarities were observed between zebrafish and mammalian phenotypes for many genes, acting as an important positive control. Given the success of this work, it is likely we’ll see others taking this approach in the study of diseases for which association data exist.

Future directions 

With all this data, it is natural to ask: what next? The authors suggest how their data can be followed-up by providing proofs-of-principle for how it can generate specific hypotheses. For instance, one mutant (tcf4, a transcription factor) showed decreased activity in an area of the brain known to ‘light up’ during hunting. Upon further investigation, they indeed found reduced hunting behavior in mutants. In this way, this preprint provides a large number of entry points into detailed studies of various kinds of behavior.

Questions for the authors

For me, it would be nice to hear if the authors have any practical tips for the generation and organization of such a large mutagenesis project, accomplished in a single lab. It would also be interesting to hear their opinion on whether such large-scale approaches could in the future be used to investigate inter-(or intra-)genic regulatory sequences that may also play an important role in disease.

Further reading

Brown SDM, Holes CC, Mallon AM, Meehan TF, Smedley D and Wells S. High-throughput mouse phenomics for characterizing mammalian gene function. Nat. Rev. Genet. 2018 19(6):357-70.

McCammon JM and Sive H. Challenges in understanding psychiatric disorders and developing therapeutics: a role for zebrafish. Dis. Model Mech. 2015 8(7)647-56.

Schizophrenia Working Group of the Psychiatric Genomics, C. Biological insights from 108 schizophrenia-associated genetic loci. Nature 2014 511, 421-427.

(1 votes)