This work is a proof of concept towards fully programmable chromatin, something I think will become very common. By combining evolutionary insights, high-throughput assays and predictive/generative modeling, we should be able to uncover some true “superpowers” of chromatin (more on this soon!) 7/9

Finally, we set out to achieve what I’d always wanted. Using screen data and sequence embeddings, we trained a classifier to predict heterochromatin repression, and used it to design totally synthetic histones we called synH4s that demonstrate optimized heterochromatin-repressing activity. 6/9

This process was able to alter cellular plasticity to transcription factor inputs: T cells expressing the tail mutant G4D displayed increased accessibility of heterochromatin-associated transcription factors and were able to respond to overexpression of those factors with proliferation! 5/9

We were surprised to see that some mutations in the N-terminal tail of H4 caused outsized effects on chromatin structure- digging in deeper, we found that these mutations alter nucleosomes in cis, ablating the heterochromatin mark H3K9me3 and shaping spatial compartmentalization in the nucleus. 4/9

The first step was defining sequence-to-function relationships. To do this, we turned to high-throughput screening of 1000s of variants for effects on chromatin accessibility and histone marks. We found that some single-residue changes have global effects on chromatin structure! 3/9

Organisms across the tree of life, including humans, have evolved histone sequences that remodel chromatin in diverse ways. As I learned about these examples, I grew mildly obsessed with the idea of designing histones to do what *I* wanted them to do. 2/9

Excited to share my postdoc work, out on bioRxiv today! Histones package DNA into nucleosomes to form the building blocks of chromatin, but how modular and programmable is this system? 1/9