Abstract
Cryo-electron microscopy (cryo-EM) has produced a number of structural models of the SARS-CoV-2 spike, already prompting biomedical outcomes. However, these reported models and their associated electrostatic potential maps represent an unknown admixture of conformations stemming from the underlying energy landscape of the spike protein. As with any protein, some of the spike's conformational motions are expected to be biophysically relevant, but cannot be interpreted only by static models. Using experimental cryo-EM images, we present the energy landscape of the glycosylated spike protein, and identify the diversity of low-energy conformations in the vicinity of its open (so called 1RBD-up) state. The resulting atomic refinement reveal global and local molecular rearrangements that cannot be inferred from an average 1RBD-up cryo-EM model. Here we report varied degrees of “openness” in global conformations of the 1RBD-up state, not revealed in the single-model interpretations of the density maps, together with conformations that overlap with the reported models. We discover how the glycan shield contributes to the stability of these low-energy conformations. Five out of six binding sites we analyzed, including those for engaging ACE2, therapeutic mini-proteins, linoleic acid, two different kinds of antibodies, switch conformations between their known apo- and holo-conformations, even when the global spike conformation is 1RBD-up. This apo-to-holo switching is reminiscent of a conformational preequilibrium. We found only one binding site, namely that of AB-C135 remains in apo state within all the sampled free energy-minimizing models, suggesting an induced fit mechanism for the docking of this antibody to the spike.
Original language | English |
---|---|
Pages (from-to) | 68-77 |
Number of pages | 10 |
Journal | Current Research in Structural Biology |
Volume | 4 |
DOIs | |
State | Published - Jan 2022 |
Externally published | Yes |
Funding
This work was supported by the US National Science Foundation ( NSF ) under award DBI2029533. The development of underlying techniques was supported by the US Department of Energy , Office of Science , Basic Energy Sciences under award DE-SC0002164 (underlying dynamical techniques), and by the US NSF under awards STC 1231306 (underlying data analytical techniques) and 1551489 (underlying analytical models). A.S. acknowledges an NSF CAREER award MCB-1942763, and the NIH award R01GM095583. J.V. acknowledges support from the NSF Graduate Research Fellowship Grant 2020298734. We thank J. McLellan and collaborators for providing the cryo-EM data set. We are grateful to E. Seitz and S. Maji for preprocessing the data for ManifoldEM analysis. We acknowledge extensive discussions at an early stage of this work with F. Acosta, S. Maji, E. Seitz and J. Frank.
Keywords
- Cryo-EM
- Free energy landscape
- Manifold machine learning
- Molecular dynamics
- Spike protein