Canonical text (the exact bytes fingerprinted)

TRUVACE RECORD VERSION
record: TRV-2026-1644
version: 1
kind: certified
reason: Certified into the record
timestamp: 2026-10-17T01:09:51.063710Z
status: published
lens: trace
sector: health
headline: Proteome-guided drug discovery maps and mitigates therapeutic degrader toxicity
dek: Heterobifunctional targeted degraders (HBDs) enable potent removal of protein targets but their clinical success can be hindered by de novo toxicity inherent to their bivalent chemistry. Here we introduce a drug discovery framework that exploits high-throughput proteomics to map and mitigate toxicity mechanisms of emerging drug modalities. Exposing androgen receptor (AR)-negative cells to a library of experimental AR-HBDs indicated for treatment-resistant prostate cancer linked widespread proteomic responses to…
gain_title: Machine learning trained on proteomic responses traced degrader toxicity to complex I inhibition and guided a small linker change that reduced liver toxicity while keeping antitumor activity in mouse xenografts.
problem_title: Heterobifunctional degraders for treatment-resistant prostate cancer showed hepatotoxicity linked to phthalimide degraders, a toxicity that can hinder their clinical success.
trace_subject: Androgen receptor degraders for treatment-resistant prostate cancer
gain_reading: Machine learning trained on proteomic responses traced degrader toxicity to complex I inhibition and guided a small linker change that reduced liver toxicity while keeping antitumor activity in mouse xenografts.
gain_evidence: Proteome-optimized degraders displayed decreased hepatotoxicity, enhanced specificity and antitumor activity
problem_reading: Heterobifunctional degraders for treatment-resistant prostate cancer showed hepatotoxicity linked to phthalimide degraders, a toxicity that can hinder their clinical success.
problem_evidence: clinical success can be hindered by de novo toxicity inherent to their bivalent chemistry | linked widespread proteomic responses to hepatotoxicity of phthalimide degraders
quick_read: Researchers exposed androgen receptor-negative cells to a library of experimental degraders and measured broad proteome changes. These responses were tied to liver toxicity in phthalimide-based compounds, and a machine learning model trained on the proteomes mapped the main mechanism to inhibition of electron transport chain complex I.

The work matters because toxicity unique to the two-part chemistry of degraders can derail otherwise potent cancer drugs, and the approach offers a way to find and fix such problems early. The abstract reports results in cells and xenografts only, so safety and benefit in patients remain untested.
limitation: 
tag: Dual reading
key_points: The screen used androgen receptor-negative cells so that observed proteomic changes reflected off-target effects rather than the intended target. | Machine learning on proteome data identified inhibition of electron transport chain complex I as the primary toxicity mechanism. | A minor change in the linker region of the degraders reduced off-target engagement. | Optimized compounds showed lower hepatotoxicity and antitumor activity against treatment-resistant prostate cancer xenografts.
rundown: The study addresses heterobifunctional targeted degraders, which remove protein targets rather than merely blocking them. The authors note that their bivalent chemistry can produce new toxicities that threaten clinical success.

The team built a drug discovery framework around high-throughput proteomics and tested a library of experimental androgen receptor degraders designed for treatment-resistant prostate cancer. Widespread proteomic responses were linked to hepatotoxicity in phthalimide degraders.

Using the machine learning model, the group found safer analogs by modifying the linker region. The proteome-optimized compounds had decreased hepatotoxicity and greater specificity, and the authors describe the framework as versatile for developing safer medicines.
sources:
- peer_reviewed | Nature Chemical Biology | https://doi.org/10.1038/s41589-026-02347-2 | 2026-10-09
prev: 0000000000000000000000000000000000000000000000000000000000000000

sha256

c7b7e277f05c7f8c3c4682bf865b5cff3f8a5a60d67b2c39a31fe0e852c9887b

previous

0000000000000000000000000000000000000000000000000000000000000000