With the continuing spread of the COVID-19 pandemic, researchers are stepping up their search for effective pharmaceutical interventions. This has taken various forms, including computational modeling, drug screens, and structural studies.
Now, a new study published on the preprint server bioRxiv* in July 2020 describes the use of in silico modeling of potential drug targets on the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) protein structures, using data retrieved from the Protein Databank (PDB).
The SARS-CoV-2 Spike Protein
The SARS-CoV-2 is part of the coronavirus family, with characteristic spike proteins that stud the surface and mediate viral recognition by the host cell ACE2 receptor, membrane fusion and entry into the host cell. This glycoprotein exists as an anchored homotrimeric protein. The part outside the cell membrane, or the ectodomain, has an S1 subunit at the N-terminal end, with two domains, the SA and SB domains. There is a subsequent S2 subunit like a stalk.
Each monomer thus has two subunits and two segments, one which spans the cell membrane and one which is a short C-terminal cytoplasmic tail. The S1 recognizes the ACE2 receptor and can be in the open (up) or closed (down) configuration. In the latter, all the domains are closely packed with other domains from the neighboring monomers.
Recognition and receptor binding occur in the up state. This occurs only when one of the three SB domains breaks loose from its partners and tilts upwards, like a flap or hinge. As ACE2 binds to this site, the spike protein undergoes cleavage at the S1-S2 interface, triggering the post-fusion changes that lead to viral entry into the host cell by endocytosis.
This protein has been the target of intense research because drugs that inhibit this SB domain movement can well prevent infection.
The SARS-CoV-2 Main Protease
Other proteins that can act as drug targets include the main protease, which carries out the essential proteolysis that yields the functional proteins required for viral replication and transcription. This is a highly conserved protein across all coronaviruses, and thus a drug directed at this could potentially be used to prevent or treat multiple coronavirus infections.
Conventional Structural Studies
Many studies have described the crystal structure of these proteins, among others, and structural studies are vital to designing drugs using computational biology, to find the best matches among new compounds and to optimize existing molecules. However, crystal structures don’t capture protein dynamics.
Dynamic changes in protein configuration are important since some drugs may bind to protein receptors in different conformations. Secondly, flexibility is an essential determinant of drug binding but is not often evaluated because of the long time needed for this measurement with methods like molecular dynamics (MD) simulations.
Advanced Flexibility Modeling
In the current study, researchers aimed to use an advanced modeling approach to simulate protein flexibility, where the protein is split into a network of units, some rigid and some flexible, called FIRST, coupled with a method of assessing the elastic modes of motion within this nodal system by geometric modeling, called FRODA. The analysis proceeds from the rigid boundaries of the crystal structure, through many conformational changes, to represent the possible motions along the most relevant elastic modes of low-frequency motion.
This helped the scientists to outline the rigidity imparted by the protein structure combined with the potential range of motion that is possible within the existing chemical bonds and steric parameters. They covered all of the more than 200 protein structures that have been published in the PDB. The computational time is markedly reduced, and hence this method could be used for screening protein mobility on a large scale.
Possible motion along mode m7 at Ecut = 1 kcal/mol with (a) side and (b) top view of the open spike ecto domain (6vyb)9. Colors are chosen identical to Fig. 4. As in Fig. 3, the arrows in each panel show the range of motion for chain B (blue shades) in (a+b) and also for chain C (reds) in (b).
Different Types of Rigid Structure
The study describes different types of rigidity patterns shown by the FIRST method, such as the brick-like where the largest part remains rigid until finally, the whole protein becomes flexible. The second is domain-like, where the largest rigid cluster breaks into two structures, also rigid, and then the newly formed four domains remain rigid but open more and more bonds until they finally dissolve to become completely flexible.
Another pair of more common types of rigidity they found was a rapid falling apart of the largest rigid structure, with complete flexibility of the original rigid cluster even while the new clusters remain stable to the opening of more bonds – called first order; and second-order where many parts of the original rigid structure break off, leaving a no longer dominant original cluster at the end. The most complex flexibility is expected with second-order rigidity.
The FIRST analysis resulted in a map of the rigid and flexible regions, which is interpreted as a tendency to move by allowing the perturbation-induced movement of the flexible parts. This allows for possible trajectories to be charted.
S1 Ectodomain Opens and Closes
One striking finding is the possibility that the S1 subunit of the spike protein opens and closes still further by exploiting the movement of the SB domain. In the open state, the structure of the SB domain is clearly seen as being a separate rigid cluster from the rest of the trimer, indicating its markedly greater flexibility. The hinge motion of this domain is seen when the protein is in the open state, to move it back into the closed state. In the opposite direction, it opens further through a total distance of 57Å.
The range of movement of the domain is extensive with significant flexibility within the domain as the hinge movement occurs, all of which goes to say that the SB domain is very flexible and thus able to bind to the ACE2 molecule.
The opening is not visible when beginning of the closed state, which may be due to stronger chemical bonds and steric hindrances. These must, therefore, be overcome to facilitate opening. The distance from open to closed state is 69 Å. The use of motion simulation thus distinguishes the open and closed structures more clearly and confirms the transition between these states, to mediate viral entry into the human cell.
The current study is the first to predict the hinge movement of the SB domain from the dynamic structure alone. The availability of structural data on the researchers’ website can stimulate docking studies using various sizes of ligand, as well as to set up more MD-based dynamics assessments. In short, FRODA and FIRST can be used to know more about the structure of a protein after simulation of its structure and dynamics.
The analysis may help to cover all the unknown 200-plus proteins known so far, and this structural classification will shed light on the extent of flexible motion. This, in turn, is crucial to drug discovery.
bioRxiv publishes preliminary scientific reports that are not peer-reviewed and, therefore, should not be regarded as conclusive, guide clinical practice/health-related behavior, or treated as established information.