Totus Medicines

A cancer cell that should have won is losing

In a clinic somewhere, a 72-year-old woman with two PI3Ka mutations and a long history of failed treatments is responding to a pill. Her tumor has shrunk by roughly 68%. By the textbook, she should have run out of options. She did not.

That pill is TOS-358, and the company behind it is Totus Medicines - a clinical-stage outfit in Emeryville, California, with around 35 people and an unusually stubborn belief: that the targets everyone else gave up on are exactly the ones worth chasing. Most of biotech politely avoids the word "undruggable." Totus treats it as a to-do list.

Today Totus is not a slide deck or a someday-promise. It is patients on therapy, data on conference posters, and a platform that does something most drug hunters only daydream about - screen drug candidates inside living cells, at a scale measured in billions.

The proteome is enormous. The "druggable" part is tiny

FHere is the uncomfortable truth that shapes everything Totus does. The human body runs on thousands of proteins. Only a small fraction of them have ever been successfully targeted by a drug. The rest get filed under "difficult" or "undruggable" - which is the pharmaceutical industry's polite way of saying "we tried, it was hard, we moved on."

Traditional drug discovery screens molecules a few thousand at a time, in tidy test tubes, against a single purified protein floating outside its natural home. It is efficient. It is also a little like auditioning actors in an empty parking lot and being surprised when the play falls apart on opening night. Proteins behave differently inside a cell. Context matters.

So the targets that are flat, slippery, or shape-shifting - the ones responsible for some of the hardest cancers and immune diseases - stayed out of reach. Not because they didn't matter. Because the tools weren't good enough.

One scientist, eight years, a contrarian wager

Totus Medicines was founded in 2019 by Neil Dhawan, a scientist who spent his career stitching together disciplines that usually keep to themselves - small-molecule design, structural biology, genetics, biochemistry, and cell biology. The bet was simple to state and hard to pull off: combine DNA-encoded chemistry with machine learning, screen inside real cells, and go after the targets nobody else would touch.

Dhawan built the company from the ground up over eight years. Then, in December 2023, he did something founders rarely do gracefully - he handed over the wheel. Nassim Usman, Ph.D., a chemistry veteran with a Ph.D. from McGill, a postdoc at MIT, and a track record of pushing drugs into the clinic and leading biotechs, stepped in as President and CEO. Dhawan moved to Chief Scientific Officer and Head of R&D, then onto the Board - staying close to the science while letting a clinical operator run the play.

It is the kind of succession that looks obvious in hindsight and is almost never executed cleanly. The platform needed a builder. The pipeline needed someone who had taken molecules through trials before. Totus arranged to have both.

Meet OmniDEL, the proteome's least favorite machine

OOmniDEL is the engine room. It is an AI-powered DNA-encoded covalent library platform - a phrase that does a lot of quiet work. In plain terms: each candidate molecule carries a DNA "barcode," so millions of them can be tested at once and then read back out. Totus adds covalent chemistry (molecules that bond to their target rather than just brushing past it) and, crucially, runs the whole thing inside living cells instead of purified protein soup.

The result is scale that is hard to overstate - billions of candidates screened against thousands of targets, with large-scale machine learning mapping the relationships between molecular shapes and the proteins they hit. It is less a microscope and more a cartographer, charting which chemistry sticks to which target across a broad swath of the proteome.

Out of that machine comes the pipeline. TOS-358, the lead program, is the first and only covalent PI3Ka selective inhibitor in clinical development - an oral drug for select solid tumors. Behind it sits a first-in-class covalent IRF5 inhibitor aimed at immunology and inflammation. The platform finds the molecules; the clinic proves them.

The data is doing the talking

Platforms make promises. Patients keep score. In its Phase 1b doublet cohort - TOS-358 paired with fulvestrant in HR+/HER2- breast cancer - Totus reported a 100% disease control rate and an 89% clinical benefit rate, including in patients who had already failed prior PI3K/AKT/mTOR pathway therapy. Sixty percent of patients stayed on therapy past 24 weeks.

The number that quietly matters more is on the safety side. Covalent drugs have a reputation for being aggressive. Across 56 patients, TOS-358 showed 0% bone marrow, hepatic, renal, or ocular toxicity, no rash or grade 3 diarrhea, and zero discontinuations due to intolerance. Only 3.6% needed insulin for manageable hyperglycemia. A potent drug that patients can actually stay on - that is the rarer achievement.

"Bring hope to every diagnosis"

Totus phrases its mission plainly: save lives and end untreatable disease. It is the kind of line that could read as boilerplate, except the company keeps backing it with the harder, less glamorous work - going after targets with no precedent, screening in conditions closer to real biology, and obsessing over tolerability as much as potency.

The culture matches the method: cross-disciplinary by design, with chemists, biologists, geneticists, and machine-learning researchers working the same problem from different angles. It is a small team for an ambition this large. That is rather the point - you don't map the whole proteome by playing it safe.

If the platform holds, the to-do list gets longer

The interesting thing about a discovery platform is that one win is never the whole story. TOS-358 is the proof of concept. If OmniDEL can reliably turn "undruggable" targets into clinical candidates, the same machine points at a long catalog of underserved targets across oncology, immunology, and beyond - which is precisely why a company like Lilly came knocking.

Back in that clinic, the 72-year-old woman with two mutations and a shrinking tumor is not a metaphor. She is what happens when a target everyone abandoned finally meets a molecule built to fit it. Totus Medicines is betting there are thousands more targets like that, and a machine patient enough to find them.

The cancer cell that should have won is still losing. That is the whole idea.