←
Constantine Tzouanas198 posts
Constantine Tzouanas avatar

Constantine Tzouanas

@constantine_sci

Making every day in lab as unique as every single cell | @shaleklab | Harvard-MIT @mit_hst | @HertzFoundation Fellow | @NSFGRFP Fellow

◷ Joined April 2020▧ 67 Photos and videos
Constantine Tzouanas avatar
Constantine Tzouanas@constantine_sci·Oct 26

What can we learn from all cytokines in the human genome? For most, we have limited info on what they do and when to target them Meet TAPAS & EPICURE: tools to synthesize all human cytokines➡️scRNA data of primary human immune responses➡️🖥️methods to use them in your work (1/n)

Post attachment
Constantine Tzouanas avatar
Constantine Tzouanas@constantine_sci·Oct 26

We’re very excited for integrated wet lab and computational tools to help decipher and engineer the signaling language of the human body – many thanks to the study participants and to collaborators who made this possible, in particular co-advisors @shaleklab @snbhatia (12/n)

Constantine Tzouanas avatar
Constantine Tzouanas@constantine_sci·Oct 26

Lots more in text: signaling proteins as a protein class primed for therapeutic impact; EPICURE to prioritize signals key in your disease/system of interest; TAPAS to advance AI/ML virtual tissues and empower your cells, your measurements, your phenotype of interest (11/n)

Constantine Tzouanas avatar
Constantine Tzouanas@constantine_sci·Oct 26

EPICURE prioritized ovary-produced cytokine GDF9 as mimic of IL-2 (one of the 1st immunotherapies), despite no sequence or structural similarity. We validated this at protein level and in tumor killing capacity of GDF9. Power of exploring all cytokines in the human genome! (10/n)

Post attachment
Constantine Tzouanas avatar
Constantine Tzouanas@constantine_sci·Oct 26

We aggregated >1000 signatures across different diseases/tissues/contexts➡️unified them via EPICURE around shared upstream signaling drivers. More in main text, like chemokine drivers of T stem cell memory states that we supported via cross-cancer human spatial -omics (9/n)

Post attachment
Constantine Tzouanas avatar
Constantine Tzouanas@constantine_sci·Oct 26

Many studies defined sets of genes that predict disease outcomes or stratify patient survival, but hard to know what’s upstream and causal. We made EPICURE to prioritize cytokines that drive or reverse these signatures, based on TAPAS’ primary human immune perturbation data (8/n)

Post attachment
Constantine Tzouanas avatar
Constantine Tzouanas@constantine_sci·Oct 26

We studied signal transduction pleiotropy, systematically across cytokines & cell types. Cytokines drive strong or variable responses, but only rarely both. Impt: not predicted from receptor expression alone (as assumed in many🖥️tools)➡️value in testing cytokine responses!(7/n)

Post attachment
Constantine Tzouanas avatar
Constantine Tzouanas@constantine_sci·Oct 26

Here’s another demo, for progranulin: normally made as a single large protein, but can be proteolytically cleaved into individual domains. We synthesized them all with TAPAS and showed one polypeptide can drive many shared vs. divergent responses in downstream cells (6/n)

Post attachment
Constantine Tzouanas avatar
Constantine Tzouanas@constantine_sci·Oct 26

Lots of interesting demos of flexible protein synthesis➡️immune responses: here’s TAPAS making previously reported IL-2 structurally engineered variants, nanobody surrogates, & de novo designs. Very different sequences and structures, but shared downstream cell responses (5/n)

Post attachment
Constantine Tzouanas avatar
Constantine Tzouanas@constantine_sci·Oct 26

For broad disease relevance, we made all cytokines in the human genome➡️stimulate primary human PBMCs➡️scRNA perturbation datasets. Proteins group together around the shared downstream responses they drive in primary human immune cells (4/n)

Post attachment
Constantine Tzouanas avatar
Constantine Tzouanas@constantine_sci·Oct 26

We created TAPAS to synthesize genome-scale protein signals, then measure how they regulate primary human cells’ healthy functions and disease responses. Human-based cell-free protein synthesis lets us do this in high-throughput, even by a single scientist in just 48 hours! (3/n)

Post attachment
Constantine Tzouanas avatar
Constantine Tzouanas@constantine_sci·Oct 26

90% of data focuses on 10% of cytokines. Which cytokines are most studied today is best predicted by their historical discovery order, back to the 1970s. Millions of patients are on therapies targeting already-well-studied signals – what can the rest of signals do? (2/n)

Post attachment