De novo ligases put ubiquitin on a chosen lysine: PCNA K164 is picked out among 45 off-target lysines
Synopsis
A team used RFdiffusion3 to design ubiquitin ligases from scratch around a cysteine catalytic triad, conjugating an activated ubiquitin donor directly to a chosen lysine on a target protein. Against a 19-residue model peptide with a single lysine, 35 of 48 designs showed measurable ubiquitylation activity, and four reached near-quantitative transfer without experimental optimization. Against the homotrimeric sliding clamp PCNA, only four of 48 designs were active, and PL4 modified K164 alone; a single N148A substitution then let PL4.1 convert 90% of PCNA to ubiquitylated product in 3 hours.
Interpretation
Which lysine gets ubiquitylated can now be set by design rather than by a natural E3 ligase: the designed enzyme recognizes one lysine on a target protein and attaches ubiquitin through an isopeptide bond. Natural ubiquitylation runs through an E1-E2-E3 cascade, with substrate recognition carried mainly by more than 600 human E3 ligases, so redirecting ubiquitin transfer to a chosen lysine has been difficult; chemical and semisynthetic routes can make defined ubiquitin conjugates but often need elaborate synthesis and protein refolding. On a 19-residue model peptide with a single lysine, 35 of 48 designs showed measurable ubiquitylation activity, spanning 10 of 11 backbones; MPL16, 21, 29 and 40 gave near-quantitative transfer of Ub-MESNa to the peptide overnight with only minimal extra donor hydrolysis.
The same design pipeline extends from peptide recognition to a globular native protein and discriminates one lysine from 45 others. The PCNA trimer carries three symmetry-related K164 sites plus 45 other lysines, and the highly polar surface around K164 lacks a hydrophobic patch to anchor an enzyme, while an adjacent loop is partly unresolved in the crystal structure, making recognition harder than for the model peptide. All 48 PCNA designs expressed solubly and four ubiquitylated PCNA; PL4 showed no detectable ubiquitylation of PCNA K164R, whereas PL3, PL14 and PL19 retained some off-target activity.
Coupling the designed ligase to E1-dependent donor activation allows site-specific ubiquitylation directly from unactivated ubiquitin and installation of defined-length ubiquitin chains. Preparing site-specifically ubiquitylated PCNA has required substantial experimental effort, and the engineered promiscuous E2 Ubch5c S22R modified PCNA more slowly under similar E1-coupled conditions while generating multiple ubiquitylated species. With 200 µM ubiquitin, 100 µM PCNA and 10 µM ligase, PL4.1 converted about 90% of PCNA to ubiquitylated product over 24 hours versus 60% for Ubch5c; PL4.1 also transferred linear M1-linked di-, tri- and tetraubiquitin chains to K164 with near-quantitative PCNA conversion.
Perspective
The demonstrated scope is in vitro: a model peptide, a SUMO-peptide fusion and human PCNA, using Ub-MESNa or E1 with ATP and MESNa to generate the activated donor in situ, plus a 1.6 Å crystal structure of PL4.1; the authors propose delivering designed ligases into cells as a next direction but report no intracellular results.
The authors report that only four of 48 PCNA designs were active and attribute the bottleneck to precise substrate placement across the enzyme-PCNA interface; in PL4 the designed N148 side chain is mispositioned in the predicted structure so neither designed oxyanion-hole interaction forms, and removing that residue raised activity about 26-fold. PL4.1 showed no detectable activity on yeast PCNA, indicating that recognition depends on specific interface features rather than overall fold.