If you are interested in using DAPseq for your plant, algal, fungal or microbial genomes, consider applying to one of @jgi.doe.gov's user programs:

jgi.doe.gov/work-with-us...

Final note - all of this data is publicly available! See our GEO ‘superseries’ page here: ​​https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE299028 and please reach out if you have questions or trouble finding anything. We’re excited to see what you uncover!

This project was a huge team effort with five(!) co-first authors, including JGI scientists @leobaumgart.bsky.social, @abmora.bsky.social, Peng Wang, and Yu Zhang each playing crucial roles, and @omalley-regulome.bsky.social at the helm.
An amazing group!

And on a practical level, tracking TF activity instead of expression of individual genes gives us quantitative big-picture way to compare cell types within a species, as well as across both closely and distantly related lineages.

Take-home!
DAP-seq + snRNA-seq
🤜 🤛
Integrating these data types with the framework of comparative genomics is an incredibly powerful way to understand gene regulation, giving us a window into why genes are expressed where they are, and how TF regulons get rewired to enable novelty.

But we also found cases where TFs became active in new cell types, and added hundreds of new target genes along the way. We even found a cool example of evolution in action, where the balance of regulatory power seems to be switching between MYB and NAC TFs in xylem.

First - we saw that plenty does stay the same. We could often recognize a sorghum cell type as the best match to a brassica cell type simply by looking at its TF activity profile.

But we weren’t done yet! Not all functional binding sites are conserved forever, right? Otherwise how do we get novelty? We next jumped across the tree to the bioenergy grass Sorghum and used the same approach to track TF activity, this time with binding sites conserved in its grass relative, rice.

Many TF-celltype relationships reflect long-standing functional knowledge. For example, MYB107 target genes lit up suberized-endodermis, aligning with the TF's known role in suberin synthesis. Others were entirely new. This opens doors for creative ways to describe and even manipulate cell types.

We now had a robust and powerful framework for tracking TF activity. Focusing only on expression of target genes with conserved binding sites, we could infer where each TF was active. Mapping out the active TFs in each cell type gave us the big-picture view of gene expression we’d been waiting for.

Sure enough, the most conserved binding sites showed the highest correlation between TF and target gene expression, and marked genes with the most cell type-specific expression patterns in all 4 species. Makes for a strong case that conservation is a good marker of binding site importance.

So next we generated snRNA-seq atlases for 3 different tissues (seedling, leaf, and flower) of each of the 4 Brassica species. Again, lots of data. But again, multiplexing helps! We found that extracting and profiling nuclei from all species together made for an easier protocol and cleaner data.

Binding sites shared across all 4 species had all the hallmarks of being ‘functional’ - they had lower within-species nucleotide diversity, higher chromatin accessibility, and were near functionally-related genes. But to be sure they impact expression - maybe we should look at expression data?

We tested this theory with 4 related species from the Brassica family. For every TF binding site in A. thaliana, we asked how many of the other species had a binding site for the same TF near an orthologous gene.

Second challenge: while DAP-seq finds all possible binding sites, not all actually impact expression of a nearby gene. Which binding sites matter? For that we turned to comparative genomics. We reasoned that binding sites that stuck around during evolution are probably there for a reason.

If that sounds like a lot of data - it is. The first challenge was profiling all those TF binding sites - our team previously developed multiDAP-seq (www.nature.com/articles/s41...), which applies the in vitro DAP-seq method to a pool of genomes. Here we optimized it for bigger eukaryotic genomes.

Single-nuclei RNAseq reveals which genes are ‘on’ where, but behind the scenes it’s transcription factors directing the show. Here, we profiled binding of 100s of TFs in 10 plant genomes, plus a suite of new snRNA atlases - to ask: does tracking TF activity help simplify and compare snRNA data?