Prediction and Evolution of the Molecular Fitness of SARS-CoV-2 Variants: Introducing SpikePro

Abstract

The understanding of the molecular mechanisms driving the fitness of the SARS-CoV-2 virus and its mutational evolution is still a critical issue. We built a simplified computational model, called SpikePro, to predict the SARS-CoV-2 fitness from the amino acid sequence and structure of the spike protein. It contains three contributions: the inter-human transmissibility of the virus predicted from the stability of the spike protein, the infectivity computed in terms of the affinity of the spike protein for the ACE2 receptor, and the ability of the virus to escape from the human immune response based on the binding affinity of the spike protein for a set of neutralizing antibodies. Our model reproduces well the available experimental, epidemiological and clinical data on the impact of variants on the biophysical characteristics of the virus. For example, it is able to identify circulating viral strains that, by increasing their fitness, recently became dominant at the population level. SpikePro is a useful, freely available instrument which predicts rapidly and with good accuracy the dangerousness of new viral strains. It can be integrated and play a fundamental role in the genomic surveillance programs of the SARS-CoV-2 virus that, despite all the efforts, remain time-consuming and expensive.

Keywords: COVID-19; SARS-CoV-2; deep mutagenesis; immune escape; protein binding affinity; protein stability; spike protein variants; viral fitness.

Figure 1

Receptor binding domain of the SARS-CoV-2 spike protein (PDB code 6M0J). The two views are related by a 180${}^{\circ }$ rotation with respect to the plane (shown as a vertical line) representing the ACE2 binding interface. (a) The ensemble of residues that bind to ACE2 are colored in red spheres and the other residues are in blue. (b) The ensemble of epitope residues targeted by at least one nAb and less then ten nAbs of the ${\mathcal{D}}^{\mathrm{nAb}}$ set are in light pink spheres, those targeted by ten or more nAbs are in dark pink spheres and the other, non-epitope, residues are in blue. The list of epitope residues and ACE2 binding sites can be found in our GitHub repository (https://github.com/3BioCompBio/SpikeProSARS-CoV-2/blob/main/Structures/Epitope.dat, accessed on 10 April 2021).

Figure 2

Schematic representation of the three steps of our computational pipeline: in silico mutagenesis experiments to compute the change in stability of the spike protein, and its change in binding affinity for ACE2 and for the 31 nAbs from ${\mathcal{D}}^{\mathrm{nAb}}$. The spike protein is in blue, ACE2 in red and the antigen-binding fragment of a nAb in orange. The structures used for the pictures on the left, center and right have the PDB codes 6VXX, 6M0J and 7B3O, respectively.

Figure 3

Mutation rate $R_i$ of SARS-CoV-2 spike protein variants i observed in the GISAID database [46] as a function of (a) the predicted change in folding free energy $\Delta \Delta G_i^{\mathrm{S}}$ (in kcal/mol) and (b) the predicted fitness contribution ${\varphi }_i^{\mathrm{S}}$.

Figure 4

HeatMap of the predicted $\Delta \Delta G_i^{\mathrm{nAb}}$ values for each of the 31 nAb/spike protein complexes from the set ${\mathcal{D}}^{\mathrm{nAb}}$. The color scale is shown on the right (in kcal/mol). Light blue corresponds to variants that slightly stabilize the complex and orange, to mutations that destabilize the complex. The most destabilizing mutations are likely to lead the virus to escape from the immune system.

Figure 5

Average per-residue ${\varphi }^{\mathrm{nAb}}$ fitness contributions related to the ability of the virus to escape from the immune system, mapped onto the 3D structure of the spike protein RBD (PDB code 6M0J). As shown in the color bar on the right, residues whose mutation lead to highest average ${\varphi }^{\mathrm{nAb}}$ and thus to viral escape from nAbs are shown in white; residues with fitness values of one or lower and not allowing viral escape are in dark blue. The ${\varphi }^{\mathrm{nAb}}$ fitness values for all residues in the RBD are given in our GitHub repository (https://github.com/3BioCompBio/SpikeProSARS-CoV-2/blob/main/phi_nAb.dat, accessed on 10 April 2021). The ACE2 binding interface is shown in red in the two small pictures at the top (see also Figure 1). The left and right pictures are related by a 180${}^{\circ }$ rotation with respect to the vertical plane (shown as a line) representing the ACE2 binding interface.

Figure 6

Time evolution of the predicted overall fitness $\Phi$. (a) Average fitness $\langle \Phi \rangle$ per month for the SARS-CoV-2 strains collected from the GISAID database [46] as a function of time; (b) Probability distribution of $\Phi$ for the SARS-CoV-2 strains collected from GISAID during a given month.

Figure 7

Probability distribution of the predicted fitness $\Phi$ for ${10}^6$ simulated viral strains obtained by inserting (a) three and (b) five random mutations in the spike protein; (c) probability distribution function of the predicted fitness $\Phi$ for all the strains collected in the GISAID database.