Gluing Cancer Proteins Together Shrinks Prostate Tumors in Mice

MANY cancer drugs have a single job: jam one protein and hope the tumor can’t find a way around it. In prostate cancer cells and in mice, a molecule that glues two cancer-driving proteins together suppressed cancer growth more than a pair of inhibitors blocking those same proteins. Arul Chinnaiyan, Shaomeng Wang and colleagues at the University of Michigan built the molecule and report in Nature how it fared in cultured cells, lab-grown tumor models called organoids, and mice. Laboratory work for now, with no human trial behind it.

The idea comes from a young field called chemically induced proximity, in which small molecules are designed to bring proteins together and change what a cell does with them (some tag a protein for destruction, others hijack a gene-switching helper). This one joins two proteins that already work as partners inside prostate cancer cells.

Two Partners, Forced Into an Awkward Hold

The androgen receptor is a protein that responds to male hormones, and prostate cancers that keep growing despite hormone-lowering treatment stay addicted to it. It works with p300, an enzyme that tags histones (the spools DNA winds around) so that nearby genes can be read. The team’s molecule has two ends joined by a short linker, one gripping the receptor and one gripping p300, so it hitches the partners together in an awkward position. In test tubes with purified proteins, the molecule pulled the two together, whereas the inhibitor pair and look-alike molecules missing one end did not.

Why should hitching two proteins beat blocking them? In prostate cancer cell lines, the molecule curbed growth more than the two inhibitors used together, and versions with either end disabled lost much of their activity. The authors suggest the gap reflects a new activity created by the forced miswiring, which two separate blockades cannot supply.

Two cancer cell lines that lack the androgen receptor but depend on p300 were hit far less by the molecule than by a p300-destroying drug used for comparison. A four-hour dose followed by a wash kept suppressing growth about as well as continuous dosing, while the comparison treatments lost much of their effect. The selectivity matters, p300 does honest work in healthy tissue too, and shutting it down across the body has been linked to disrupted blood-cell production and low platelet counts.

Chinnaiyan, who directs the Michigan Center for Translational Pathology at Michigan Medicine, reached for a workshop image. “Our molecule acts like a molecular brace, connecting these two cancer-driving proteins in an abnormal configuration that traps them in an inactive state,” he said.

Both proteins survive the treatment intact, and what changes is the work p300 does. Normally p300 tags histones with chemical acetyl marks that keep nearby genes readable. In treated cultured cells, histone tagging fell sharply, while the androgen receptor and its helper proteins picked up more acetyl marks than before. So where does the tagging go? The authors suggest the molecule steers p300 toward the wrong targets. Both proteins stayed on DNA, and at some sites more of them gathered than usual. But the machinery that reads genes arrived in smaller numbers there, and cancer-driving gene activity fell.

Human Tumors Grafted Into Mice Began to Shrink

In the first mouse experiment, human prostate cancer cells grew as tumors in castrated mice, a model of cancer that keeps growing despite hormone-lowering treatment. Injected into the abdomen, the molecule made 12 of 16 tumors in these mice regress, a 75% response, and doses given by mouth also suppressed growth and outperformed the inhibitor pair. Tissue analysis showed the receptor and p300 sitting closer together in treated tumors, with lower levels of cancer-linked proteins. No drop in the receptor or p300 along the way.

A second model came from a patient. Tumors grown in mice from that patient’s cancer received the molecule for 50 days and regressed throughout, while enzalutamide (a drug that blocks the same receptor) slowed growth only modestly. After dosing stopped on day 50, the molecule-treated tumors showed no evident recurrence over the next 16 days, whereas tumors given a drug that destroys the receptor regrew. In safety tests on 4 to 9 mice per group, treated animals showed no significant changes in body weight, liver and kidney function, or blood counts.

Every result comes from cultured cells, organoids or immunodeficient mice (14 to 16 tumors per treatment arm in the main study, 11 to 12 tumors per arm in the patient-derived one), and the chromatin maps and tumor-tissue stains came from single experiments, so none of it says what the molecule would do in a person. The contact points between the joined proteins come from computer simulation, backed by mutating a few predicted residues, and not from a measured structure. Wang and Chinnaiyan co-founded Medsyn Biopharma and hold equity in it. The University of Michigan has licensed its patents on these molecules to the company, which has also given the university a research contract, with Wang as principal investigator.

If any of it translates, the practical prize is a pill. In mice, 18.5% of a swallowed dose reached the bloodstream. The authors frame the approach as a template for designing molecules that rewire other disease-relevant protein complexes, but that is a design claim, not a result. For now the job is to pick a candidate for human trials: “Our teams are working closely to identify a drug candidate suitable for testing in human clinical trials,” Wang said.

FAQ