SARS-CoV-2: Possible recombination and emergence of potentially more virulent strains

Abstract

COVID-19 is challenging healthcare preparedness, world economies, and livelihoods. The infection and death rates associated with this pandemic are strikingly variable in different countries. To elucidate this discrepancy, we analyzed 2431 early spread SARS-CoV-2 sequences from GISAID. We estimated continental-wise admixture proportions, assessed haplotype block estimation, and tested for the presence or absence of strains' recombination. Herein, we identified 1010 unique missense mutations and seven different SARS-CoV-2 clusters. In samples from Asia, a small haplotype block was identified, whereas samples from Europe and North America harbored large and different haplotype blocks with nonsynonymous variants. Variant frequency and linkage disequilibrium varied among continents, especially in North America. Recombination between different strains was only observed in North American and European sequences. In addition, we structurally modelled the two most common mutations, Spike_D614G and Nsp12_P314L, which suggested that these linked mutations may enhance viral entry and replication, respectively. Overall, we propose that genomic recombination between different strains may contribute to SARS-CoV-2 virulence and COVID-19 severity and may present additional challenges for current treatment regimens and countermeasures. Furthermore, our study provides a possible explanation for the substantial second wave of COVID-19 presented with higher infection and death rates in many countries.

Fig 1. Detection and classification of mutations from GISAID SARS-CoV-2 genome sequences.

Illustration of the distribution of synonymous and nonsynonymous variants for each gene in the raw dataset, MAF ≥ 0.5%, and MAF ≥ 1% thresholds are shown. A set of 72 variants in total was observed with MAF ≥ 0.5% threshold and utilized in subsequent analysis. MAF- Minor Allele Frequency.

Fig 2. Principal Component Analysis using 2352 GISAID sequences.

Principal Component Analysis of 2352 SARS-CoV-2 sequences shows three distinct clusters of color-coded samples (see the legend for their continent of origin). All three clusters diverge from a single point (red circle). The North American cluster (black oval) shows least variance among the three. The European cluster (orange oval) is well-defined with few interspersed Asian samples, an indication of its origin. The third cluster (Blue oval) shows the most variance and includes samples from Oceania, Asia, and others.

Fig 3. Identification of SARS-CoV-2 genetic clusters in different continents.

Illustration of the seven (C1 to C7, color-coded) genetic subdivisions of SARS-CoV-2 sequences across continents using variants with MAF ≥ 0.5%. Differential proportions of strong LD (C), weak LD (D), haplotype block (E), nonsynonymous (F), and synonymous (G) variants across continental datasets are shown.

Fig 4. Estimation of haplotype blocks in continental samples.

Haplotype block estimation and extent of linkage disequilibrium observed between variants in Asia, Europe, and North America, identified a single block with different lengths in Asia and Europe, while in North America two blocks were identified.

Fig 5. Correlation between LD and physical distance across continental SARS-CoV-2 genomes.

(A) Upward peaks show strong LD and downward peaks show weak LD. North America and Oceania showed strong LD in the region overlapping the S protein. Strong LD among variants suggested that these de novo mutations were not broken by recombination events, (B) Smooth lines showed clear differences in LD over physical distance across continental datasets, (C) The genomic structure of SARS-CoV-2 is depicted.

Fig 6. 3D modelling of SARS-CoV-2 Spike protein.

(A) Trimeric structure of SARS-CoV-S spike like protein (PBD:6VSB). (B) Overlay of the SARS-CoV-S spike like protein (PBD ID: 6VSB, blue) with the modelled SARS-CoV-2 S protein (PDB ID: 6M71, magenta). (C) The surface of the modelled S protein with the RRAR furin cleavage site (blue).

Fig 7. 3D modelling of SARS-CoV-2 Spike protein showing suggested bonds for D614.

(A) Suggested hydrogen bonds (red dashed lines) of D614 (S1 domain chain A) with T859 (S2 domain chain B) and D614 and A646 of S1 domain chain A. (B) The suggested hydrogen bond can be disrupted with the D614G mutation altering the activity of the protein.

Fig 8. 3D modelling of G614 mutation.

(A) S protein monomer 6VSB with D614G mutation, the red region of the protein depicts the more flexible region of the protein due to the D614G mutation with a decrease in stability of ΔΔG: -0.086 kcal/mol and an increase in vibrational entropy to ΔΔSVib 0.137 kcal.mol^-1.K^-1. (B) A zoomed-in structure of the N-terminal domain (NTD) and the G614 mutation in close vicinity to the RARR furin cleavage site.

Fig 9. SARS-CoV-2 RNA-dependent RNA polymerase structure in complex with nsp7 and two nsp8.

The viral RNA template and NTP entry is shown in black arrow heads. The active site is a large groove with several structural pockets. (A) Wild type RdRp complex P323 is shown in pink (B) L323 mutation is shown in green. RdRp-RNA-dependent RNA polymerase.

Fig 10. 3D modelling of P323L mutation.

(A) Suggested bonding network of P323 where the COO- group might form H-bonds with the backbone NH group of T324 and S325 and the side chain of S325. The grey dashed lines depict the hydrophobic interactions between P323 and W268 and F275. (B) The mutated L323 forms a H-bond with the side chain of S325 and forms a hydrophobic interaction with L270, which is at the curve of the loop making that region more compact.

Fig 11. 3D depiction of the less relaxed loop caused by L323 mutation.

(A) 3D structure of the RNA-dependent RNA polymerase where the blue region depicts a more rigid structure due to P323L mutation with an increase in stability of ΔΔG: 0.717 kcal/mol and a decrease in vibrational entropy to ΔΔSVib ENCoM: -0.301 kcal.mol^-1.K^-1. (B) A zoomed-in structure showing the less flexible loop region caused by the tight hydrophobic interactions between L323 and the hydrophobic moieties.