Mechanistic modelling of within-mosquito viral dynamics: Insights into infection and dissemination patterns

Abstract

Vector or host competence can be defined as the ability of an individual to become infected and subsequently transmit a pathogen. Assays to measure competence play a key part in the assessment of the factors affecting mosquito-borne virus transmission and of potential pathogen-blocking control tools for these viruses. For mosquitoes, competence for arboviruses can be measured experimentally and results are usually analysed using standard statistical approaches. Here we develop a mechanistic approach to studying within-mosquito virus dynamics that occur during vector competence experiments. We begin by developing a deterministic model of virus replication in the mosquito midgut and subsequent escape and replication in the hemocoel. We then extend this to a stochastic model to capture the between-individual variation observed in vector competence experiments. We show that the dose-response of the probability of mosquito midgut infection and variation in the dissemination rate can be explained by stochastic processes generated from a small founding population of virions, caused by a relatively low rate of virion infection of susceptible cells. We also show that comparing treatments or species in competence experiments by fitting mechanistic models could provide further insight into potential differences. Generally, our work adds to the growing body of literature emphasizing the importance of intrinsic stochasticity in biological systems.

Investigating the dose-dependency of the midgut escape barrier using a mechanistic model of within-mosquito dengue virus population dynamics

“…aegypti mosquitoes with DENV. Despite being relatively simple, our model provides quantitative predictions which are consistent with our experimental data about how increasing the infection dose influences the rate at which mosquitoes become infectious—a phenomenon that has been demonstrated in previous experimental studies [ 21 , 22 ] and modelling work [ 29 ]. It also provides a framework that could be used to explore how perturbations to key mechanisms, for example, due to viral mutations or environmental fluctuations, may affect transmission rates.…”

“…To account for this known characteristic, we used a phenomenological statistical model to allow specimens to remain uninfected. Others have shown that stochastic individual-based models of the invasion process can be used to understand the dose-dependency in the probability a mosquito becomes infected [ 29 ]; interestingly, this other work also hints at there being a strong dose-dependency in dissemination.…”

“…Whether wild populations avoid these bottlenecks is unclear since infectious doses in the wild are thought to be considerably lower than those required in the laboratory to produce successful infection [ 53 , 63 ], and midgut infection has been demonstrated to present a particularly strong barrier at low infection doses [ 64 ]. Some have even suggested that these bottlenecks may be evolutionarily advantageous by selecting against defective virus particles early on during infection [ 29 ]. At a given titer, however, asymptomatic patients yielded a higher proportion of infected mosquitoes and higher mosquito viral titers; indicating that while DENV titer is important, other factors such as host immune responses may play a role in determining vector infection [ 65 ].…”

“…As only a handful of virus particles are thought to be responsible for disseminated infection, it seems plausible that some mosquitoes may fail to become infectious because no virus particles manage to escape the midgut; our continuous model does not allow such a possibility [ 64 ]. Extending either our framework to model individual virus particles or extending other stochastic models [ 29 ] to incorporate interactions with the basal lamina layer could help to understand whether purely random phenomena can explain why some mosquitoes become infectious and others do not under the same experimental conditions or whether other mechanisms, such as individual differences in the initial blood meal size, virus inoculum, or variation in immune response account for this difference.…”

“…Mathematical models have been an important tool in understanding mosquito-borne pathogen epidemiology and are now routinely used to inform policy [23][24][25][26]. They have also been extensively used to explore within-host pathogen dynamics [22,[27][28][29]. Despite this work exploring within-host pathogen dynamics, only a limited number of mathematical models have explored viral dynamics in the mosquito [22,[29][30][31], and the majority of existing within-vector modelling work has so far focused on malarial parasites [32][33][34][35][36].…”