Full-size structural and energetic modelling of the Staphylokinase active complex

Ischaemic stroke is one of the world’s leading causes of death and disability. In spite of this, its primary treatment relies on recombinant tissue plasminogen activator (alteplase and tenecteplase), despite potentially fatal side-effects, limited effectiveness, and high costs. Staphylokinase, a bacteriaderived thrombolytic, holds promise with significantly higher fibrin specificity, potentially reducing side-effects. However, its clinical application is limited by immunogenicity and sub-optimal activity. The persistence of these challenges is partly due to structural studies employing a truncated partner protein. To overcome these challenges, we will develop a full-size 3D model of the protein using template-based AlphaFold2 predictions in conjunction with energy-based optimisation of side-chains. This will be followed by meta-inference simulations integrating experimental restraints, enabling accurate investigation of the dynamics of the full-size active complex. Additionally, to maintain feasibility in experimental validation, we developed a filtering strategy. Benchmarking on 38 staphylokinase variants using Spearman rank correlation revealed poor or inconsistent performance for Rosetta ddG and Cartesian ddG (ρ ≈ −0.25), and BioEmu (ρ ≈ 0.26), while ThermoMPNN (ρ ≈ −0.49) and FoldX (ρ ≈ −0.44) performed best. FoldX was chosen for further optimisation due to improved reproducibility and tunable parameters, yielding a final Spearman ρ of -0.404. Next, saturation mutagenesis will be combined with FoldX and in silico prediction of affinity to identify optimal regions for diversification. Ultimately, this work will enable the efficient development of novel variants, as well as provide an enhanced molecular understanding of the full-size active complex.