Structural and computational investigation of MMP‐3 affinity improvements with an engineered TIMP‐1 inhibitor

Abstract

Matrix metalloproteinases (MMPs) are a family of zinc‐dependent endopeptidases that function to mediate tissue remodeling; however, they are also implicated in pathological conditions such as cancer, inflammatory and degenerative diseases. There are 23 human MMPs and significant sequence and structural homology exists among them and with other metalloproteinases. This represents a substantial challenge in developing therapeutically viable selective inhibitors due to off‐target effects. Tissue inhibitors of metalloproteinases (TIMPs) are a family of four endogenous protein protease inhibitors that regulate MMP activity. Our lab and others have utilized TIMPs as scaffolds, engineering them to selectively target specific MMPs for potential therapeutic applications. We previously employed a yeast surface display platform to identify TIMP‐1 variants possessing enhanced MMP‐3 affinity. In the current study, we used structures of MMP‐3 co‐crystallized with either WT TIMP‐1 or an engineered TIMP‐1 variant as seeds for molecular dynamics simulations. Simulations were also conducted on unbound WT and engineered TIMP‐1 to examine MMP‐independent dynamics that distinguish the engineered variant. The simulation data were analyzed with root‐mean‐square deviation/fluctuation, time‐dependent secondary structure, temporally resolved hydrogen bonding, conformational clustering, and dynamical network algorithms to discern structural and dynamical determinants of the increased affinity. Key observations that demonstrate the increased effectiveness of engineered TIMP‐1 as an MMP‐3 inhibitor were secondary structural changes in the MMP‐binding loops in proximity to mutations, attenuated global and mutation‐adjacent dynamics, decreased conformational dynamics, and increased persistence of MMP‐3‐compatible conformations. There were also indications of greater interdomain cooperativity within the engineered TIMP‐1 variant: increased interdomain hydrogen bonding, coordinated motions of the N‐terminal region with the C‐terminal domain, and increased intercommunity information flow/motion transfer. From these observations, we hypothesize the increased interdomain interactions represent a mechanism of dynamical stabilization such that mainly cooperative motion between domains presides, rather than random independent motions. Our findings suggest propagated motions between domains may influence the MMP affinity of TIMPs. Furthermore, these data suggest future strategies to design next‐generation mutant TIMP libraries by targeting regions dynamically coupled to the MMP binding loops for variation, thus exploiting natural protein dynamics to achieve desired selectivity.