tetano
Editor, Senior Moderator
Protein Sci
. 2026 Aug;35(8):e70706.
doi: 10.1002/pro.70706.
Deep mutational scanning of SARS-CoV-2 nucleocapsid reveals regional stability patterns
Alasdair D Keith[SUP] 1 [/SUP], Meredith M Keen[SUP] 1 [/SUP], Filipp Frank[SUP] 1 [/SUP], Eric A Ortlund[SUP] 1 2 [/SUP]
Affiliations
Point mutations within the SARS-CoV-2 nucleocapsid protein (NP) have the potential to impact not only detection but the viral life cycle, and therefore pathogenicity. Leveraging deep mutational scanning (DMS) we determine the effect of 7876 (98.9%) of all possible point mutations on NP surface expression as a proxy for stability, which we term surface-expression-based folding fitness (SEBFF). We show NP susceptibility to SEBFF changes varies with region, with mutations to the structured RNA-binding domain (RBD) and dimerization domain (DD) being more likely to cause decreases in SEBFF. We show that the R203K mutation in the linker region increases the folding fitness of NP, suggesting thermostability may have been one factor driving the adoption of the R203K/G204R set of mutations arising in the alpha, gamma and omicron lineages. Our DMS results appear consistent with the limited recombinant protein and biophysical data for NP variants available from the literature. We tested three computational methods to determine whether any could accurately predict the DMS data; per-variant predictions were poor, but per-site predictions were good, with machine learning methods outperforming the physics-based strategy, and ThermoMPNN-D performing most effectively with the RBD and ProteinMPNN with the DD.
Keywords: SARS‐CoV‐2; deep mutational scanning; machine learning; multiplexed assay; nucleocapsid; thermostability.
. 2026 Aug;35(8):e70706.
doi: 10.1002/pro.70706.
Deep mutational scanning of SARS-CoV-2 nucleocapsid reveals regional stability patterns
Alasdair D Keith[SUP] 1 [/SUP], Meredith M Keen[SUP] 1 [/SUP], Filipp Frank[SUP] 1 [/SUP], Eric A Ortlund[SUP] 1 2 [/SUP]
Affiliations
- PMID: 42423125
- PMCID: PMC13347322 (available on 2027-07-09)
- DOI: 10.1002/pro.70706
Point mutations within the SARS-CoV-2 nucleocapsid protein (NP) have the potential to impact not only detection but the viral life cycle, and therefore pathogenicity. Leveraging deep mutational scanning (DMS) we determine the effect of 7876 (98.9%) of all possible point mutations on NP surface expression as a proxy for stability, which we term surface-expression-based folding fitness (SEBFF). We show NP susceptibility to SEBFF changes varies with region, with mutations to the structured RNA-binding domain (RBD) and dimerization domain (DD) being more likely to cause decreases in SEBFF. We show that the R203K mutation in the linker region increases the folding fitness of NP, suggesting thermostability may have been one factor driving the adoption of the R203K/G204R set of mutations arising in the alpha, gamma and omicron lineages. Our DMS results appear consistent with the limited recombinant protein and biophysical data for NP variants available from the literature. We tested three computational methods to determine whether any could accurately predict the DMS data; per-variant predictions were poor, but per-site predictions were good, with machine learning methods outperforming the physics-based strategy, and ThermoMPNN-D performing most effectively with the RBD and ProteinMPNN with the DD.
Keywords: SARS‐CoV‐2; deep mutational scanning; machine learning; multiplexed assay; nucleocapsid; thermostability.