The effect of post-weaning social isolation and chronic celecoxib administration on gene expression in the mouse hippocampus

Many environmental factors lead to an increased risk of neurodevelopmental and neuropsychiatric disorders such as schizophrenia (SCZ). One example is early life stress (ELS). A form of ELS, social isolation (SI), can be modelled in animals. Another factor is chronic inflammation. Human evidence implicates this factor, with studies showing an elevated baseline level of pro-inflammatory cytokines in human SCZ compared to healthy controls. In an effort to investigate this factor, chronic celecoxib (CEL) administration can be used to simulate an anti-inflammatory environment in mouse. We can study these environmental stressors to understand the underlying mechanisms that may be responsible for disease risk in humans. Here we measure the functional effect of these stressors on mouse brain by assaying gene expression in the hippocampus and further investigate the biological relevance of these changes. A total of 32 female C57 Bl6 mice were split between chronic celecoxib-treated (CEL) and vehicle-treated controls. Half of the animals were also subject to post-weaning SI from postnatal day (PD) 21 for a total of 40 days for an n = 8 per group. RNA was extracted from the hippocampus and subject to paired-end RNA-seq. Reads were aligned to a transcriptome, quantified and analysed for differentially-expressed genes (DEGs). Genes were considered DEGs at an FDR < 0.05. Sets of DEGs from both factors were subject to functional analysis including enriched gene ontology (GO) terms, cell-type enrichment analysis and enrichments in genes harbouring rare and common risk for neurodevelopmental/neuropsychiatric disorders. SI induced a total of 55 DEGs in the hippocampus, while CEL administration induced 355 DEGs. Five genes (Adamtsl5, Neurod1, Hrasls, Kcnab3, Cdh6) were differentially-expressed by both factors. The most significantly-enriched GO terms for SI DEGs were gamma-aminobutyric acid:sodium symporter activity (GO:0005332; FDR = 0.0012) and calcium ion binding (GO:0005509; FDR = 0.0020). For CEL administration, the top GO terms were neuron part (GO:0097458; FDR = 3.6E-11), neuron projection (GO:0043005; FDR = 1.4E-09) and locomotory behaviour (GO:0007626; FDR = 5.1E-09). Neither set were enriched for common risk genes contributing to neurodevelopmental or neuropsychiatric disorder. Both sets were nominally-enriched (p < 0.05) for rare de novo missense variants contributing to SCZ. When cell type enrichment analysis was performed, SI-induced DEGs were enriched in a population of astroeppendymal cells, while CEL-induced DEGs were enriched in four populations of medium spiny neurons (MSNs). Both factors had a clear impact on neuron function in the mouse brain. SI appeared to have a relatively subtle effect on the hippocampus in this case, however, did highlight the inhibitory GABA neurotransmitter as playing a part in regulating social stress. Furthermore, the cell type analysis implicated eppendymal cells, which may be indicative of a change in metabolic products in the brain. CEL administration had a more pronounced effect, with DEGs showing enriched GO terms related to neurodevelopment and behaviour. In addition, a change in neurotransmitter function, including glutamate and dopamine was implicated in the GO analysis. This links to human SCZ where the function of these neurotransmitters are known to be disrupted. The nominal implication of rare missense variants and MSN cell populations (previously implicated in human SCZ) in the cell type analysis support this link.