Exploring the Binding Mechanism between Human Profilin (PFN1) and Polyproline-10 through Binding Mode Screening

Leili Zhang; David R. Bell; Binquan Luan; Ruhong Zhou

In Figure S9 , we compare how the 5 properties (average RMSDs from MD over the first 50ns, maximal forces from two sets of SMD, and rupture works from two sets of SMD) rank the 11 binding modes as compared with PMFs. The correlation coefficients between the ranks are similar compared to the correlation coefficients between the absolute values ( Figure S9 A-E), ruling out the "clustering effect" of the data (meaning the ranks may be interchangeable when absolute numbers are close). In Figure S9 F, we list the most stable binding modes predicted from the 5 metrics. Interestingly, mode7 is the most stable binding mode for 4 out of 5 metrics except for the rupture work (from 1 nm/ns SMD) where mode4 (in Category III along with mode7) is predicted to be the most stable. The ranks of mode11 are reliable except for maximal forces (from 0.1 nm/ns SMD). The ranks of mode9 are also high in the list but may be affected by its structural flexibility seen in the MD simulations. Notably, the top ranks derived from the rupture works from SMD correspond well to the most stable binding structures obtained from the umbrella sampling, indicating that this technique could provide a fast and reliable way of ranking the PPI binding modes from the protein docking programs such as GRAMM-X.