Can a conserved transcriptomic axis predict state modulation of cortical interneurons?

Inhibitory interneurons in cortex consist of many subtypes that differ in, e.g., connectivity and physiology. Recently, it was shown that —in layers 1-3 of mouse primary visual cortex— interneuron subtypes differ systematically in their modulation with an animal’s behavioural state, and that this state modulation can be predicted from the first principal component (PC) of the gene expression matrix (Bugeon et al. 2022). Here, we ask if this link between transcriptome and state-dependent processing is a general principle that extends across layers, areas, and species.
To this end, we analysed 6 publicly available single-cell RNA sequencing (scRNA-seq) datasets collected from mouse, human, songbird, and turtle forebrains. All data sets contain a similar set of interneurons. In spite of this conservation at the level of cell types, we found clear differences between transcriptomic PCs, with larger differences between evolutionarily distant species. Further analyses revealed two factors underlying these differences: Divergence in gene expression within conserved cell types, and —to a lesser extent— divergence in cell type abundance. We confirm that cross-species differences are much larger than within-species differences, and show that within-species differences are mainly due to the presence or absence of deep-layer subtypes in the dataset.
Finally, we study the cross-species expression of cholinergic receptors, thought to causally link transcriptome and state modulation (e.g. Muñoz et al. 2017, Bugeon et al. 2022). We find that many cholinergic receptors correlating with state modulation in mouse interneurons are not expressed in interneurons of other species, suggesting two hypotheses for future work: (1) Interneurons are similarly state-modulated across species but via different pathways, indicating adaptive significance; (2) Interneurons are differentially state-modulated across species, indicating evolutionarily divergent control of information flow. Overall, our work leverages genomic data to test the universality of findings from the mouse as an accessible model organism.