Many sequencing workflows capture one layer of biology at a time: RNA on one platform, protein on another, morphology somewhere else entirely. Researchers end up stitching together data sets generated under different conditions, on different instruments, hoping the pieces connect.
Jens Durruthy Durruthy, Director Product Management at Element Biosciences, opened the webinar by laying out what a different approach looks like: generating RNA, protein, and morphology measurements simultaneously from the same cells, on the same instrument, without library prep.
Then, Tilmann Buerckstuemmer, CSO at Myllia Biotechnology, a pioneering company in high-content CRISPR screening, walked through exactly that kind of experiment: an optical pooled CRISPR screen probing NF-κB signaling in human cells, using cell morphology and NF-κB nuclear translocation as a phenotypic readout. It's a well-characterized pathway, which made it a useful benchmark—if the screen recovers known biology cleanly, you can trust the platform with biology you don't know.
Want the full story? Watch a recording of the webinar available on demand.
Using Cas9-expressing HeLa cells, the team at Myllia infected cells with a pooled guide RNA library, selected the cells, and seeded them directly onto AVITI24 flow cells. Prior to the run, the cells were treated with TNF-α, then imaged on AVITI24.
Each cell yielded three layers of data: direct guide RNA sequencing, NF-κB localization via oligo-conjugated p65 antibodies, and global morphology through cell painting. The underlying technology is Direct In Sample Sequencing (DISS). Sequencing happens within intact cells on the flow cell surface with no library prep required. RNA, protein, and morphology all tie back to the same cell without cross-platform data integration.
Guide RNA assignment reached ~82%, with ~363K of 440K cells receiving a unique guide identity. Neighboring cells on the image frequently shared guide identities which is consistent with cell division during the 24-hour pre-run incubation period and an indication of mapping reliability.
Looking at the location of NF-κB protein expression, they determined which gene knockouts affected the nuclear translocation of NF-κB from the cytosol, both those perturbations that blocked and enhanced translocation. Top hits from the screen included TNFRSF1A (a TNF-α receptor), RIPK1, MAP3K7, and multiple IKK complex members—all established NF-κB regulators. The screen recovered known pathway biology cleanly, a powerful validation of the platform.
A downsampling analysis defined practical coverage thresholds for these screens. At ~1,000-1,500 cells per gene knockout, 13–14 significant hits were recovered. At ~400-500 cells per gene, the strongest hits remained detectable with reduced significance, and weaker hits dropped out.
A unique feature of AVITI24, is that every cell in the experiment generates a full cell painting image, and the team at Myllia used those morphological profiles to cluster cells according to only morphology features, looking for relationships with the ~200 gene knockouts using no prior biological assumptions.
While most knockouts clustered together and did not cause major morphological changes, the Myllia team identified a set of 31 genes that clearly affected cell morphology. Interestingly, one cluster contained genes belonging to the SAGA complex—a large protein complex involved in chromatin modification.
The broader implication Dr. Buerckstuemmer raised: optical pooled screening on AVITI24 can uniquely match morphological profiles across genetic knockouts or small molecule treatments which can help connect compounds to their target biology or identify perturbations that phenocopy a drug of interest.
We also flagged a few capabilities on AVITI24 to be launched later this year as part of our innovation roadmap:
• 3’ transcriptome readout alongside guide RNA and protein
• Immunofluorescence-style protein readouts for intensity-based measurements, moving beyond barcoded antibody dot counting
• Tissue compatibility (FFPE and fresh frozen)