De novo protein design has not yet achieved the creation of proteases capable of selectively cleaving any desired peptide bond within a native protein with high precision. Here, we report the use of the flow-based generative model Proteus2 to design metalloproteases by generating enzyme–substrate complexes conditioned on a target peptide sequence and a predefined catalytic motif. Our approach employs a two-step encapsulation strategy to create clamp-like metalloproteases that bind the target peptide in a manner that maximizes substrate sequence specificity. The generative process simultaneously optimizes the precise positioning of the target peptide bond in a catalytically competent configuration and accurately scaffolds the transition state catalytic residues—both essential for specific and efficient catalysis. Using this strategy, we designed zinc metalloproteases targeting three distinct cleavage sites within the aggregation-prone regions of amyloid-β (Aβ), a key pathogenic factor in Alzheimer’s disease. Experimental characterization validated five enzymes, each capable of precise cleavage at the intended sites with high specificity and minimal or undetectable activity on non-cognate substrates. On average, these enzymes accelerated peptide bond hydrolysis by more than 107-fold relative to the uncatalyzed reaction, and enabled efficient digestion of the Aβ peptide into smaller segments when enzymes targeting different sites were combined. Cryo-EM structures of three designed enzymes in complex with Aβ peptide, each targeting a distinct cleavage site, revealed close agreement with the design models. Together, these results demonstrate the potential of sequence-guided, generative approaches for developing programmable, sequence-specific proteolysis and lay the groundwork for future applications in basic research and therapeutic development.
