The mutational landscape of SARS-CoV-2 nucleocapsid protein

National Institute Of Biomedical Imaging And Bioengineering, Bethesda

Investigators

Linked publications, trials & patents

Abstract

The COVID19 pandemic has spawned an unprecedented worldwide collaboration to collect SARS-CoV-2 genome sequences in the GISAID database. It currently has 15 million viral genomes, which is more than an order of magnitude greater than databases for any other virus. This genomic database has been instrumental for enhancing our understanding of viral evolution and geographic spread, such as identifying and predicting variants of concern. Less attention has been devoted to the fact that the database provides unique information on viral sequence diversity and on seemingly inconsequential fluctuations of viral protein sequences that do not result in fixed mutations. Because the database consists of consensus sequences from viral samples taken from infected patients, all sequences represent alternate viral species, each of which is successfully replicating. Charting the scope of observed amino acid substitutions along the protein sequence we obtain a mutational landscape that we believe can be interpreted in the context of biophysical protein properties. We have focused on the SARS-CoV-2 nucleocapsid (N-)protein, which is critical for viral assembly and scaffolding of viral RNA in ribonucleoprotein particles, serves essential functions to suppress intracellular antiviral defense pathways, and presents as a major antigen. N-protein consists of large intrinsically disordered regions, which flank and link the two folded N-protein domains. We have previously shown that >80% of all amino acids in N-protein can be substituted by, on average, 4-5 other amino acids without impacting viability. In the reporting period, we have updated our previous mutational landscape, and were able to show that the three major groups of early genomes, Delta-variant, and the new Omicron-variant species reproduce the same mutation frequencies for N-protein, as in independent replicates of vast, natural deep mutational scans. Furthermore, we have exploited the landscape for the identification of a transient helical oligomerization site in the disordered linker region of N-protein, which we believe is critical for viral assembly, and were able to show that nearly all mutations in the helical region satisfy the structural requirements of coiled-coil oligomers. This supports the interpretation of the mutational landscape as a source of information on biophysical constraints and protein functions. A major challenge in the study of mutational landscapes is the relationship between the genotype and phenotype spectrum. It has been shown recently that non-local physicochemical properties of intrinsically disordered regions can be functionally important and evolutionarily conserved features. Thus, we have exploited the rich database of mutant sequences to calculate the range of physicochemical properties of different N-protein regions. We observed a surprising spread of biophysical parameters such as charge, polarity, and hydrophobicity across the mutant spectrum, but with very distinct and non-overlapping indices in different N-protein regions which point to their functional roles. We have combined this with an experimental study of representative mutants and observed variation in thermodynamic stability, oligomerization, and phase behavior. Overall, it appears the biophysical parameter space is unexpectedly flexible, which may allow for the large sequence space of viable viral species. We expect to publish this study in the coming months. In recent years a hypothesis has emerged that intrinsically disordered regions of viral proteins utilize short linear motifs (SLiMs) to interact with host proteins. SLiMs are recognition modules consisting of 3-10 disordered amino acids that can dock to specific recognition sites and mediate protein-protein interactions in many eukaryotic pathways. Viral proteins, by virtue of their high mutation frequency, are thought to efficiently evolve such motifs to hijack host interaction networks. It has remained unclear so far what role the sequence diversity and quasispecies nature of RNA viruses has in this process. The SARS-CoV-2 mutational landscape offers a unique opportunity to study this, through analysis of SLiMs that are created across the mutational spectrum. To this end, we have analyzed the dynamics of motif mimicry in the observed N-protein sequence space. This showed that the intrinsically disordered regions of N-protein indeed display a highly diverse set of motifs. While only few motifs in the ancestral sequence are conserved, many new motifs appear in different subsets of mutant sequences. Unexpectedly, we found that the theoretically accessible sequence space can form a large fraction of all known eukaryotic motifs. This suggests that motif mimicry may be more ubiquitous than previously thought, and potentially a salient component of the virus-host interface in RNA viruses. We have just submitted this work for publication and released a preprint.