Fig. 4: Modeling ZIKV infection dynamics identifies salivary gland traversal as the primary driver of differences in transmissibility.

A Conceptual model for in vitro viral dynamics. The dynamics are described using a logistic growth curve as per Eq. (1) (see Methods). B Conceptual model for in vivo viral dynamics. The mosquito image was created using BioRender: TORII, S. (2025) https://biorender.com/dbcfk4s. After ingestion of a blood meal containing infectious virus (G_v), the virions are degraded in the blood meal according to a clearance rate (μ). The probability that at least one virion infects the midgut epithelium (β) determines whether infection is established. If infection occurs in the midgut (M_v), the virus replicates at a growth rate (r) constrained by the carrying capacity (k). The virus may then disseminate to the hemocoel according to an ‘escape’ rate (λ). Virus in the hemocoel (H_v) undergoes similar replication dynamics as in the midgut and can ‘escape’ to infect the salivary glands (S_v), which eventually enables virus release into saliva. The simplest model assumes fixed parameter values (r, k, λ) across tissues and no between-mosquito variation in probabilities or rates. These assumptions are relaxed stepwise to evaluate processes underlying experimental observations. C Model outputs for viral dissemination. The proportion of simulations reaching the hemocoel is shown for five model scenarios (Table 1). Across all scenarios, except scenario five, transmission occurs in 100% of simulations. Scenario five introduces random variation in the transfer of virions between the hemocoel and salivary glands, enabling a reduced proportion of simulations with transmission, consistent with experimental findings. D Proposed hypothesis for difference between chimeric viruses of the first set. Results from scenario five suggest that differences in chimeric viruses may arise from variability in the rate at which virions infect the salivary glands and/or are released into saliva. This variability can be represented by a Gamma distribution, with variance adjusted between simulations to reflect distinct virus-mosquito interactions. Example Gamma distributions were modeled with variances of 10⁻⁷^.⁵, 10⁻⁷^.³, 10⁻⁷, and 0⁻⁶^.⁵ to show the effects of this variation.