CYFIP1 and the Open Niche in 15q11.2

Most of the noise in the 15q11.2–15q13 region comes from two genes. UBE3A drives Angelman and Dup15q syndromes; SHANK3, just past the region, drives Phelan-McDermid. Both have active clinical programs: Ionis’s ION440 (ATTUNE trial) for UBE3A, Stoke and others for SHANK3. The proximal BP1–BP2 interval — four genes including CYFIP1, NIPA1, NIPA2, and TUBGCP5 — has stayed in a different category. It is well-described in the human genetics literature, but the translational pipeline is empty.

There is no public Phase 1 program targeting CYFIP1 in microduplication. There is no published mouse model dedicated to the BP1–BP2 dup as a primary indication; what exists are larger Angelman/Dup15q lines where the BP1–BP2 region is incidentally covered. There is no commercial iPSC line for the duplication, no characterized organoid platform, no FDA-recognized natural history study.

This is unusual for a CNV with the prevalence and phenotypic load that BP1–BP2 carries. Estimates put the duplication around 1 in 1,000–2,000 live births in some series, with a phenotypic spectrum that includes intellectual disability, autism, speech delay, and seizure risk. The numbers are not trivial. The molecular tractability is also not trivial — CYFIP1 is a gene where partial reduction is plausibly therapeutic without crossing into the loss-of-function range that causes the reciprocal microdeletion syndrome.

The likely reason the niche stayed open is incentive structure rather than biology. CNV indications without a “famous” single gene attract fewer academic labs. Pharma sees a phenotype-heterogeneous patient population and walks. Foundation funding follows established names. The same dynamics that make Angelman and Phelan-McDermid attractive — clean monogenic story, organized advocacy — work against BP1–BP2.

I think this is a mistake. The U-shaped dosage relationship of CYFIP1 (too little is bad, too much is bad) means a partial knockdown therapy has a window that does not exist for most genes. That is unusually favorable for an antisense approach. It is also favorable for a clinical development plan: the therapeutic goal is not to silence the gene but to nudge it back inside the normal range. That is an easier conversation with the FDA than “we will knock out 90%.”

What is missing is not the molecular case. What is missing is the unglamorous infrastructure that other indications already have. Patient registries with confirmed BP1–BP2 trios. A mouse line trimmed down to the BP1–BP2 interval specifically. iPSC lines from affected probands. Natural history with quantitative endpoints. Anyone who builds those pieces — even outside a drug-development context — accelerates whatever the eventual therapy turns out to be.

The therapy is what I am working on. The infrastructure deficit is what I think anyone in the rare disease space should be looking at. There are several intervals like this one, sitting in the gap between “well-known indication” and “completely unknown variant.” The economics of working on them are not great. The patients are still there.