Enjoyed spring in Tokyo last month for a much needed respite from the chaos here in the US. Glad to get a second dose of spring now that I'm back here in Colorado. Refueled to get out there and protest some more!

#StandUpForScience #FortCollins

Getting my poster ready...

Found them on a brisk, clear day with blue skies!

Hmmm... Might just be that cell. Here's another example of same experiment (from stasevichlab.colostate.edu)

Here’s a cool movie showing histone H3K27ac increase upon TSA treatment (inhibits deacetylation). Here the intrabody (green; anti-H3K27ac scFv-sfGFP) translocates into the nucleus as acetylation levels there increase.

Here’s an example for histone H4 Lysine 16 acetylation (H4K16ac). The original does not colocalize with the nucleus at all (dashed yellow line). After passing through the pipeline and quantification, design 3 is the best, having nice localization, signal to noise, and expression.

After these successes, we went a little crazy and made 17 new intrabodies against residue-specific histone H3 and H4 modifications:

We next tried an antibody against a linear epitope from the Sars-CoV-2 nucleocapsid protein (EpNP; Addgene antibody #211756 from @geisslaboratory.bsky.social ). As a standard scFv it fails, but after going through the pipeline it works, colocalizing with EpNP-tagged H2B in the nucleus:

So we ran the antibody sequence through our pipeline (which you can think of as AI-based loop grafting). Wow… All four designs we tested now functioned, colocalizing with FLAG-tagged histone H2B in the nucleus. Here’s one of those designs:

For example, the popular M2 antibody against the FLAG tag fails to function in living cells when converted to an intrabody (single chain variable fragment), even after loop grafting onto more stable antibody scaffolds that have high sequence similarity.

To solve this problem, we created an AI-driven pipeline combining antibody domain annotation, AlphaFold2, ProteinMPNN, and live-cell screening to rapidly convert an antibody sequence into a functional intrabody. Code at github.com/jbderoo/scFv... (thanks to Jacob DeRoo @jderoo.bsky.social).

The main issue – aside from getting *often proprietary* antibody sequences (another LONG story) – is antibodies don’t fold and function well inside cells (they are meant for the bloodstream, which is less reducing).

Ever wonder why there are so few intrabodies given the huge number of antibodies out there? Us too!

Intrabodies are engineered antibodies that function inside living cells, enabling therapeutic, diagnostic, and imaging applications.

Making intrabodies from antibodies just got easier! Learn how we made 𝟭𝟵 intrabodies to bind and light up peptides and histone modifications in live cells. And thanks to Academia, all sequences are freely available. (video credit: Yuko Sato @YukoSatoT2) (1/15)
www.biorxiv.org/content/10.1...