P01-02
Generation of Structural Ensemble of Linear Diubiquitin Based on PCS Experiments
Yoshiki YUGAMI *1, Xue-Ni HOU2, Sotaro FUCHIGAMI3, Hidehito TOCHIO2, Kei MORITSUGU1
1 Department of Science, Osaka Prefecture University
2 Graduate School of Science, Kyoto University
3 Graduate School of Pharmaceutical Sciences, Shizuoka University
( * E-mail: sfc05146@st.osakafu-u.ac.jp )
Ubiquitin is a protein consisting of 76 amino acids, and the polyubiquitin modification of proteins is involved in various functions including protein degradation, DNA repair, and signal transduction. Polyubiquitin chains can form diverse structures through the polymerization of ubiquitin via seven Lys residues (K6, K11, K27, K29, K33, K48, K63) and the N-terminal residue (M1; linear), and can function by binding to linkage-specific substrates. In our previous work, pseudo contact shifts (PCS) using lanthanoids were applied to linear diubiquitin and the PCS data was found to be reproduced by using a series of representative structures. This study aims to establish a method for deriving the structure ensemble and the associated free energy landscape of linear diubiquitin that is in good agreement with the PCS data by (1) all-atom molecular dynamics (MD) simulations and more simply, (2) coarse-grained MD with machine learning technique.
In (1), the initial model of linear diubiquitin was taken from PDB: 3b0a, and fully solvated in a rectangular box. Using eight replicas with varying interactions between the two ubiquitin, structural sampling simulation was performed for 500 ns using gREST module of GENESIS. The PCS data were calculated for each of the resulting structural ensemble. In this study, a weight was assigned to each structure and optimized using the steepest descent method to match the PCS experimental data with Tm tags on both D39 and G47 distal ubiquitin, thus extracting the structure ensemble consistent with the experiment. In (2), coarse-grained MD was carried out using cafemol to generate an extensive structural ensemble by connecting extended (PDB: 2w9d) and compact (PDB: 3axc) structures. The obtained Cα-atom structures were converted to all-atom models using the software cg2all, and PCS data were calculated. Similar to (1), the weights for all the structure were calculated and the 2D free energy landscape was derived along the distance and the torsion angle of the two ubiquitin.
While the PCS data from a single crystal structure did not match with the experiment, the agreement from the PCS data using the 400,000 gREST structures was also insufficient. The optimization of the weights for the MD structures was found to drastically improve the agreement. The free energy landscape was also changed considerably, showing that combining PCS experiment can avoid the convergence problem of the structure sampling and remove the artifacts from the MD force field. It was also demonstrated that combining coarse-grained MD with reconstructing the all-atom model can also generate the accurate structure ensemble to match the PCS experiment with much less computational cost.