# AI-Designed Enzymes Cut Selected Protein Sequences in Lab Tests

By Simon Yoon

Canonical URL: https://www.tokenpost.com/news/technology/32018
Published: 2026-10-14T19:35:58.000Z
Updated: 2026-10-14T19:35:58.000Z
Section: Technology

> Two studies tested cysteine- and zinc-based designs, including enzymes that cut selected sequences in TDP-43 and amyloid-beta.

Two groups used AI to design enzymes that cut selected protein sequences, including sequences in TDP-43 and amyloid-beta, proteins associated with neurodegenerative diseases. Laboratory results show the designs can catalyze protein breakdown in mammalian cells, but do not establish a treatment for people.

In one study, 13 of 69 cysteine-based designs cut their intended targets. The most active design accelerated peptide-bond cleavage by up to 30 million times compared with the reaction without an enzyme. Peptide bonds connect amino acids to form proteins.

A separate effort tested 135 zinc-based designs. Its strongest initial design increased peptide-bond hydrolysis by more than 100 million times. Four amino-acid substitutions raised that rate enhancement to more than 10 billion times.

The zinc proteases cut selected sequences from TDP-43, amyloid-beta and serum amyloid A in laboratory tests. Across those results, rate enhancements reached up to one billion times; that maximum was not specified for each target.

Both studies found that designed enzymes cut targets in mammalian cells. The cell experiments do not establish that the enzymes can be delivered safely or clear disease-linked proteins in people. Getting the enzymes to the relevant tissues and cells remains a delivery challenge.

The enzymes were designed from scratch around catalytic components and selected target sequences. Their folds differed from those of naturally occurring proteases. The studies also found that binding to a target alone did not ensure strong enzyme activity; mutations that better organized the active site improved performance.