Weill Cornell Team Turns Chromosome 8p Deletion Into a Tumor-Selective WNT Drug

Hyperactive WNT signaling is one of the most common driving forces in colorectal cancer, and it appears in substantial fractions of breast, lung, prostate and liver tumors as well. For years, drug developers have tried to shut down the pathway, only to run into a stubborn problem: the molecules that block WNT signaling in a tumor also damage healthy tissue, particularly the gut and bone. That toxicity has limited the usefulness of an entire class of experimental medicines known as tankyrase inhibitors.

Tankyrase inhibitors work by attacking two closely related enzymes, TNKS1 and TNKS2. The two proteins are redundant members of the PARP family, and both promote WNT signaling by PARylating the negative regulator AXIN1, marking it for degradation. Blocking both enzymes can suppress WNT activity, but it also disrupts normal stem cell compartments. The result has been a narrow therapeutic window and dose-limiting side effects in early clinical testing of experimental tankyrase drugs.

A research team at Weill Cornell Medicine has now taken a different route. Rather than inhibiting both enzymes, the group designed a compound that selectively targets TNKS2 and exploits a deletion that cancer cells frequently acquire on chromosome 8p. The deletion removes the TNKS1 gene, leaving tumor cells dependent on TNKS2 while healthy cells retain TNKS1 and remain relatively protected. The work was published in Nature Chemical Biology on November 2, 2026.

The lead molecule, TDI-012804, is a first-in-class TNKS2-selective inhibitor. In preclinical experiments it suppressed WNT signaling and cancer cell growth only in tumors that had lost TNKS1 through chromosome 8p deletion. The finding reframes a common genetic accident in cancer as a therapeutic opportunity, and it offers a possible way around the toxicity that has held back WNT-directed drugs.

Key Facts

Nature Chemical Biology reported on November 2, 2026 that the study was published open access under DOI 10.1038/s41589-026-02358-z, with Jill Zimmerman as first author and Richard K. Hite and Lukas E. Dow as corresponding authors. The work comes from Weill Cornell Medicine's Sandra and Edward Meyer Cancer Center. The paper describes a structure-guided design campaign that began with more than 300 synthesized small molecules. Of those, 88 entered full dose-response assays, and seven inhibited TNKS2 with an IC50 below 50 nM and more than 25-fold selectivity over TNKS1.

Weill Cornell Medicine said on November 2, 2026 that TDI-012804 inhibits TNKS2 about 30 times more than TNKS1. In laboratory dish tests, the compound suppressed growth of cancer cells with partial or full TNKS1 loss but did not affect cells with a normal amount of TNKS1. The institutional release also described how the Dow lab found the vulnerability partly by accident, while knocking down TNKS1 and TNKS2 independently to understand their different biological roles.

Nature Chemical Biology reported on November 2, 2026 that TDI-012804 achieved an EC50 below 50 nM in a TOPFlash WNT reporter and produced up to 87 percent suppression of transcriptional activity, compared with about 94 percent for direct beta-catenin silencing. A 2.2 angstrom cryo-electron microscopy structure showed the ligand spanning the nicotinamide donor and acceptor sites and forcing the TNKS2 B-loop open. A single residue difference, L1136 in TNKS2 versus Y1289 in TNKS1, helps explain the selectivity.

Nature Chemical Biology reported on November 2, 2026 that the team analyzed TCGA and ICGC whole-genome data and found that about half of colon, lung, breast, liver and prostate cancers carry heterozygous or homozygous deletions of chromosome 8p, which contains the TNKS1 locus. Loss of the region correlated with reduced TNKS1 transcript and protein. Even heterozygous TNKS1 loss sensitized cells to TNKS2 suppression, creating what the authors describe as a collateral vulnerability.

In isogenic DLD1 cells, 250 nM TDI-012804 for 48 hours reduced intestinal stem and WNT target genes only in TNKS1 heterozygous or knockout cells. An APC-mutant patient-derived organoid with heterozygous 8p loss, labeled WCM895, was significantly more sensitive than an 8p-diploid organoid, WCM2176, with a P value of 0.02. The authors emphasize that the work is preclinical only and that no human dosing has been performed.

Analysis

The significance of the result lies less in the potency of a single molecule than in the logic used to achieve selectivity. Conventional tankyrase inhibitors were designed to hit a pathway, and they succeeded well enough to cause collateral damage in normal tissue. TDI-012804 is designed around a genetic difference between tumor and host. The compound does not need to be a better general WNT inhibitor than its predecessors. It needs to be a good TNKS2 inhibitor in cells that have already lost TNKS1, and the data suggest that this is enough to produce a measurable effect on WNT target genes and cancer cell growth.

What this really means is that the chromosome 8p deletion, rather than being only a lost piece of DNA, can function as a built-in drug target. The deletion is common: about half of colon, lung, breast, liver and prostate cancers carry it, according to the TCGA and ICGC analysis. That breadth matters because it defines a patient population that could be identified with genomic testing rather than with a functional biomarker. A tumor with heterozygous or homozygous TNKS1 loss is, in principle, a tumor that depends on TNKS2.

The structural work strengthens that argument. The 2.2 angstrom cryo-electron microscopy structure shows how TDI-012804 occupies the nicotinamide donor and acceptor sites and forces the TNKS2 B-loop open, and it identifies a single residue difference between TNKS2 and TNKS1 as a key determinant of selectivity. That kind of mechanistic detail is what separates a screening hit from a credible chemical series. The team's campaign, which moved from more than 300 synthesized molecules to seven compounds with IC50 below 50 nM and more than 25-fold selectivity, suggests the design strategy is reproducible rather than a one-off accident.

There are caveats. The data come from cell lines and patient-derived organoids, not from animals or patients, and the paper explicitly states that no human dosing has occurred. The observed difference between the 8p-deleted organoid WCM895 and the diploid organoid WCM2176 reached P = 0.02, which is statistically significant but based on a limited comparison. The Weill Cornell release also notes that the lab now hopes to develop a next-generation compound with better drug-like properties, a sign that TDI-012804 is a proof of concept rather than a finished drug candidate.

Why It Matters

WNT pathway hyperactivity is a near-ubiquitous feature of colorectal cancer and is implicated in breast, lung, prostate and liver cancers. A tumor-selective inhibitor would change the risk-benefit calculation for a pathway that has been difficult to drug precisely because it matters in normal tissue. The gut and bone toxicity seen with pan-tankyrase inhibitors is not a minor inconvenience. It has been a central reason that experimental tankyrase drugs produced dose-limiting side effects in initial clinical trials, according to Weill Cornell Medicine's November 2, 2026 release.

Chuo Chen wrote in a News and Views commentary in Nature Chemical Biology on November 2, 2026 that loss of tankyrase 1 in some WNT-driven cancers forces cancer cells to depend on tankyrase 2 for survival. Chen described the new TNKS2-specific inhibitor as a way to overcome that compensatory regulation, and framed tumor-restricted TNKS2 blockade as a route to a wider therapeutic window rather than a broad WNT shutdown. The journal's decision to publish dedicated commentary on the day of publication signals that the finding is being treated as more than a routine preclinical report.

For drug developers, the broader lesson is that redundancy can be turned into a selection mechanism. TNKS1 and TNKS2 are redundant, which is why inhibiting both causes toxicity. But that same redundancy means a tumor that loses one enzyme becomes dependent on the other. The value of the strategy is that it uses a genetic loss to define which cells should respond, rather than relying on a general pathway marker. For patients, the immediate implication is more modest: a new route to WNT suppression has been validated in principle, and the next step is to determine whether it can be made into a medicine.

Next Up

The immediate next step, according to Weill Cornell Medicine, is to develop a next-generation compound with better drug-like properties. If that molecule performs well in full preclinical tests, the lab hopes to take it into clinical trials. That sequence is set out in the release, which makes clear that full preclinical tests remain ahead.

The paper's own conclusion is deliberately restrained: TDI-012804 is a preclinical tool compound, and no human dosing has been performed. The key questions are whether TNKS2-selective inhibition can be sustained in vivo without compensatory changes, whether 8p deletion can serve as a reliable enrollment biomarker, and whether the roughly 30-fold selectivity seen in biochemical assays translates into a meaningful therapeutic window. Nature Chemical Biology's publication of the study and an accompanying commentary on the same day gives the work a prominent hearing, but the clinical verdict remains open.