Quantitative modeling of SARS-CoV-2 replication reveals phase-specific bottlenecks and antiviral targets

Abstract

SARS-CoV-2 replication depends on a tightly coordinated series of intracellular processes that remain incompletely quantified. Here, we integrated high-resolution time-resolved measurements of viral RNA, protein expression, and infectious virion production with mechanistic mathematical modeling to obtain a quantitative description of the viral replication cycle in human lung cells. Using transcriptomic, proteomic, and infectivity data collected over the first 24 hours of infection, we calibrated an ordinary differential equation model that captures genomic and subgenomic RNA synthesis, viral protein production, virion assembly, and virus release. The model accurately reproduced the observed replication dynamics and enabled estimation of kinetic parameters that are difficult to measure experimentally. Sensitivity analysis identified viral RNA replication and non-structural protein maturation as dominant determinants of viral replication efficiency. To assess predictive power, the model was challenged with independent antiviral perturbation experiments using remdesivir, nirmatrelvir, and montelukast. Model predictions closely matched experimentally observed treatment responses and correctly reproduced drug interaction effects during combination therapy. Furthermore, comparison of alternative mechanistic hypotheses supported NSP5 rather than NSP1 as the primary antiviral target of montelukast. Together, these results establish a predictive framework for dissecting intracellular coronavirus replication and evaluating antiviral intervention strategies.