Progressive and accurate assembly of multi-domain protein structures from cryo-EM density maps

Abstract

Progress in cryo-electron microscopy (cryo-EM) has provided the potential for large-size protein structure determination. However, the solution rate for multi-domain proteins remains low due to the difficulty in modeling inter-domain orientations. We developed DEMO-EM, an automatic method to assemble multi-domain structures from cryo-EM maps through a progressive structural refinement procedure combining rigid-body domain fitting and flexible assembly simulations with deep neural network inter-domain distance profiles. The method was tested on a large-scale benchmark set of proteins containing up to twelve continuous and discontinuous domains with medium-to-low-resolution density maps, where DEMO-EM produced models with correct inter-domain orientations (TM-score >0.5) for 98% of cases and significantly outperformed the state-of-the-art methods. DEMO-EM was applied to SARS-Cov-2 coronavirus genome and generated models with average TM-score/RMSD of 0.97/1.4Å to the deposited structures. These results demonstrated an efficient pipeline that enables automated and reliable large-scale multi-domain protein structure modeling with atomic-level accuracy from cryo-EM maps.

Figure 1

Flowchart of DEMO-EM illustrated with a 3-domain protein from the iron-dependent regulator of mycobacterium tuberculosis (PDBID: 1fx7A). Starting from the query sequence, domain boundaries and models of each domain are first predicted by FUpred and I-TASSER, respectively. Meanwhile, inter-domain distances are predicted with a deep convolutional neural-network predictor DomainDist. Second, each of the domain models is independently fit into the density map by quasi-Newton searching. Third, the initial full-length models are optimized by a two-step rigid-body REMC simulation to minimize the DCS between the density map and full-length model (Eq. 1). Fourth, the lowest DCS model selected from the rigid-body assembly simulations undergoes flexible assembly with atom-, segment-, and domain-level refinements using REMC simulation guided by the DCS and inter-domain distance profiles coupled with a knowledge-based force field, with the resulting decoy conformations clustered by SPICKER to obtain a centroid model. Finally, the flexible assembly simulation is performed again for the full-atomic model with constraints from centroid models adding to the energy, and the final model is created from the lowest energy model after side-chain repacking with FASPR and FG-MD.

Figure 2.

Summary of full-length structural models constructed by different approaches on 357 multi-domain proteins using synthesized density maps. (a) Mean and distribution of TM-score for models by DEMO-EM, MDFF and Rosetta, respectively. (b) Boxplot and distribution for RMSD of models by DEMO-EM, MDFF, and Rosetta, respectively. (c) Head-to-head comparison between TM-score of initial models by domain matching and that of final models after rigid-body assembly, flexible assembly and refinement. (d) Comparison between TM-score of individual domain models by I-TASSER and that in final full-length models by DEMO-EM.

Figure 3

Representative examples showing the process of DEMO-EM. Density maps are shown in gray shadow where cartoons are DEMO-EM models with different colors indicating different domains. (a) 2e1qC, a protein with 10 domains (8 continuous domains and 2 discontinuous domains) using a simulated density map with a resolution of 5.3 Å. (b) 1q25A, a protein with 3 domains using a simulated density map with a resolution of 9.9Å.

Figure 4.

Summary of structures constructed by different approaches using experimental density map. (a) Distribution of NDO score of domain boundaries predicted by FUpred. (b) TM-score of full-length models constructed by DEMO-EM, MDFF, and Rosetta. (c) Head-to-head TM-score comparison of the initial individual models by I-TASSER and that in final full-length models by DEMO-EM. (d, e) The deposited model in PDB (PDBID: 6eny) (d) and reconstructed model (e) by DEMO-EM for human PLC editing module, where different color represents different domains. (f, g) The deposited model in PDB (PDBID: 5fj6) (f) and reproduced model (g) by DEMO-EM for the P2 polymerase inside in vitro assembled bacteriophage phi6 polymerase complex.

Figure 5.

Overlay of structural models by DEMO-EM on the cryo-EM density maps for the six proteins in SARS-CoV-2 genome. (a) Spike protein (density map from EMD-21375). (b) NSP8 (EMD-11007). (c) Helicase/NSP13 (EMD-22160). (d) ORF3a (EMD-22136). (e) NSP7 (EMD-11007). (f) RNA-directed RNA polymerase/NSP12 (EMD-11007). (g) Comparison of the Spike RBD domain by DEMO-EM (cyan) with the X-ray structure (red, PDB 7bz5A).