Drosophila is a well established model to investigate
Drosophila is a well-established model to investigate the molecular and cellular defects underlying human neurodevelopmental disorders (Oortveld et al., 2013, van der Voet et al., 2014). Here, we establish Drosophila as a model for understanding the mechanisms linking mutations in KDM5 family proteins and cognitive defects. Specifically, we generated a fly strain harboring a missense allele of kdm5 analogous to a mutation in KDM5C found in ID patients that does not alter protein levels (kdm5). We show that kdm5 mutant flies have learning and/or memory deficits and define the gene Sephin 1 defects associated with this allele. These analyses revealed a striking dysregulation of genes involved in cytoplasmic translation that we propose to be a key cause of the cognitive phenotypes associated with this mutation.
Discussion Here, we present key findings regarding the gene expression and cognitive defects of a Drosophila kdm5 allele (kdm5) that is analogous to a KDM5C mutation found in ID patients (KDM5CA388P). The KDM5A512P mutation did not affect protein levels, distinguishing this model from the previous mouse KDM5C knockout model (Iwase et al., 2016, Scandaglia et al., 2017), and caused dramatic changes to the neuronal transcriptome and to the cognitive abilities of the fly. kdm5 mutants behaved indistinguishably from the demethylase-inactive strain (kdm5), suggesting that loss of enzymatic activity is the primary defect associated with this mutation. Our analyses also showed that the demethylase activity of KDM5 is required for both gene activation and gene repression in a manner that correlates with distinct promoter elements. Together, our data are consistent with an evolutionarily conserved neuronal function for KDM5 family proteins. Based on analyses of kdm5 in flies, human KDM5CA388P likely exerts its effects through a demethylase-dependent mechanism. This suggests that, while previous in vitro data of KDM5CA388P showed that this protein retained 45% of its demethylase activity (Iwase et al., 2007), the in vivo effect of this mutation is more dramatic. Because this mutation lies outside the catalytic JmjC domain, the molecular basis for the effect of KDM5A512P on enzymatic activity is not immediately obvious. Although the existing crystal structure data of KDM5 proteins do not include the residue affected by this missense mutation, it shows that domains in the N-terminal half of KDM5 family proteins interact extensively (Vinogradova et al., 2016). It is therefore likely that the tertiary structure of KDM5 places alanine 512 in close proximity to catalytically critical regions of the protein. The similarity between the phenotypes of kdm5 and kdm5 raises questions regarding the mechanism by which other alleles of KDM5 family genes lead to disease. It is possible that all missense mutations, independent of whether they reside within the JmjC domain, ultimately affect the histone demethylase activity of KDM5. This would be expected to result in the dysregulation of a common set of target genes. In support of this possibility, six of the eight missense mutations in KDM5C that have observable in vitro demethylase activity defects lie outside the JmjC domain (Brookes et al., 2015, Iwase et al., 2007, Rujirabanjerd et al., 2010, Tahiliani et al., 2007). It is, however, noteworthy that one mutation, KDM5CD87G, showed only minimal defects to in vitro demethylase activity (Tahiliani et al., 2007) and the remaining mutations remain untested. It is therefore also possible that there is more than one mechanism by which mutations in KDM5 family proteins disrupt transcription to impair cognition. As with previous transcriptome studies of KDM5 mutants across a number of species, the changes to gene expression observed in kdm5 and kdm5 mutants were mild (Iwase et al., 2016, Liu and Secombe, 2015, Lopez-Bigas et al., 2008, Lussi et al., 2016). KDM5 therefore likely acts by “fine-tuning” the expression of numerous genes within pathways essential to neuronal function. Consistent with this model, the majority of direct KDM5 target genes that were downregulated genes in kdm5 and kdm5 mutants have been previously implicated in ribosome assembly, structure, and function. Demonstrating the significance of these transcriptional changes, kdm5 and kdm5 mutant head, but not thorax, tissue showed attenuated translation. Short-term memory formation relies on the modification of existing proteins, whereas long-term memory is dependent on de novo protein synthesis to stabilize synaptic changes within the brain (Gal-Ben-Ari et al., 2012, Jung et al., 2012, Slomnicki et al., 2016). Thus, defective translation could contribute to the long-term learning and/or memory defects observed in our mutants either alone or in combination with the dysregulation of other genes involved in neuronal function. The link between translational regulation and cognitive impairment is highlighted by the observation that dysregulation of the Akt-mTOR pathway that regulates translation rates is found in patients with fragile X, Down syndrome, and Rett syndrome (Troca-Marín et al., 2012). Impaired ribosomal production or assembly is also linked to the neuronal dysfunction observed in Alzheimer’s disease and other tauopathies (Meier et al., 2016). KDM5-mediated regulation of growth pathways that impact neuronal function may be evolutionarily conserved, as brain tissue of KDM5C knockout mice showed dysregulation of genes required for growth, including ribosomal protein genes (Iwase et al., 2016).