What we are building
A program overview — the biology, the construct, the stage we're at, and why this is being done in public.
This is a public scientific program. The entries in this journal — this one included — are a running record of what we know, what we're doing, and what we find.
The problem
Some brain tumors carry a mutation that creates a molecule the body doesn't normally produce in large amounts. That molecule — D-2-hydroxyglutarate, or 2-HG — does two things: it reprograms the tumor's cells into an aggressive, proliferating state, and it suppresses the T cells that would otherwise attack it.
A drug called vorasidenib reduces 2-HG by about 93%. It is approved, it extends progression-free survival, and it has become the standard of care for IDH-mutant Grade 2–3 gliomas. But it leaves a residual — roughly 7% of baseline — and it cannot remove 2-HG that is already present. Whether that residual is enough to maintain partial immune suppression is not a settled question.
What we are building
We are engineering a bacterium to deplete the residual.
The bacterium is a modified strain of E. coli Nissle 1917, a probiotic with decades of clinical use. Our construct carries a three-part genetic circuit: sense D-2-HG, deploy depletion enzymes, kill itself when the substrate is gone. The intended delivery route is stereotactic intratumoral injection — a minimally invasive, MRI-guided procedure placing the construct directly at the tumor site, encapsulated in a hydrogel to keep it local.
Vorasidenib blocks D-2-HG production. Our construct eliminates the product. Together, the two mechanisms could achieve near-complete depletion that neither accomplishes alone. And if vorasidenib resistance mutations emerge — restoring 2-HG production — the bacterial circuit becomes more active, not less. It is designed to improve when the drug fails.
The experiment queue
We don't know yet which depletion enzyme will work best. Three enzymatic strategies are being tested in parallel: a candidate for the human oxidoreductase path (E-02), a bacterial anaerobic dehydratase with promiscuity potential (E-03), and a five-gene pathway from a clostridial organism that natively metabolizes D-2-HG (E-04).
Before any of those, the first experiment builds and verifies the chassis itself: a safety-modified EcN strain with its genotoxin-encoding genomic island deleted. That's E-01. Nothing else proceeds without it.
Every experiment in this program is named, dated, and structured around a testable hypothesis before it runs. The journal records what we find — positive results, kill criteria, and course corrections.
Why public
Science moves faster when it is legible.
This program has a specific patient. The journal is part of the accountability structure: named experiments, stated hypotheses, honest readouts. When something doesn't work, we'll say so.
We are not pretending to certainty we don't have. The experiment entries here are questions we are testing, not results we have in hand. That is the honest description of where preclinical science lives.