A shout-out to people who made this possible: first and foremost @martaig.bsky.social and @arnausebe.bsky.social, but also @zolotarg.bsky.social @xgrau.bsky.social as well as all the ASP lab members, and of course @lukasmahieu.bsky.social @steinaerts.bsky.social and all the members of LCB in Leuven

We anticipate that applying the same approaches to other species of cnidarians and early-branching animals will enable comparative cell type analyses that will reconstruct evolutionary relationships of the major animal cell types and regulatory processes by which they first evolved.

To wrap up, here we pave the way for moving beyond conventional transcriptome-based cell type characterization in non-model species, by analyzing regulatory traits that define cell type identities in Nematostella, such as CREs sequence motif composition, active TFs, and GRN architecture.

We therefore show that effector gene usage groups functionally similar cell types, but regulatory features also reflect their ontogenetic relationships. E.g. GATA/Islet neurons show regulatory seq. similarities with
EMS and pharyngeal derivatives, and Pou4/FoxL2 neurons with ectodermal derivatives.

Finally, we explored cell type clustering using different features. We highlight transcriptionally similar retractor muscles, which share many access. genes, but have distinct sets of CREs bound by distinct TFs, and each clusters with the derivatives of their precursors (ecto. for TR and EMS for MR)

With invaluable help of @lukasmahieu.bsky.social and @steinaerts.bsky.social lab we trained deep learning sequence models to prioritize motifs predictive of cell type specific accessibility, and to uncover mostly flexible motif syntax in Nematostella, in line with billboard-like model of TF binding.

With that in hand, we characterized each cell type by usage of TF motifs, and then linked active TFs to their target genes in cell type specific gene regulatory networks (GRNs). We showcase cnidocyte GRN as an example and highlight important TFs with central roles in the network (FoxL2, Pou4, Sox2).

The link between ATAC and RNA - from gene regulatory perspective - are TF binding motifs, which are not known for most Nematostella TFs. We devised a correlation-based approach to assign one motif to each TF, selected as best correlated among all motifs inferred by sequence similarity and orthology.

We used the atlas to characterize and quantify candidate CREs, including cell type-specific enhancers, cell type-specific promoters (SP), constitutive promoters (CP) and a smaller number of candidate alternative promoters (AP). We validated muscle and neuron AP of Gabra2 using transgenic reporters.

To start, @martaig.bsky.social produced the first scATAC atlas for a non-model species, profiling 60k cells from adult and gastrula Nematostella vectensis - see it annotated in the app: sebelab.crg.eu/nematostella-cis-regulatory-atlas/ and the genome browser: sebelab.crg.eu/nematostella-cis-reg-jb2

In this project we wanted to extend cell type characterization in early-branching animals from transcriptome-based (scRNA) to regulatory-based definition, by experimentally profiling chromatin accessibility (scATAC) and computationally inferring TF binding to cis-regulatory elements (CREs).

Fixed! Thanks for pointing it out