Predictive modeling of non-viral gene transfer.

In non-viral gene delivery, the variance of transgenic expression stems from the low number of plasmids successfully transferred. Here, we experimentally determine Lipofectamine- and PEI-mediated exogenous gene expression distributions from single cell time-lapse analysis. Broad Poisson-like distributions of steady state expression are observed for both transfection agents, when used with synchronized cell lines. At the same time, co-transfection analysis with YFP- and CFP-coding plasmids shows that multiple plasmids are simultaneously expressed, suggesting that plasmids are delivered in correlated units (complexes). We present a mathematical model of transfection, where a stochastic, two-step process is assumed, with the first being the low-probability entry step of complexes into the nucleus, followed by the subsequent release and activation of a small number of plasmids from a delivered complex. This conceptually simple model consistently predicts the observed fraction of transfected cells, the cotransfection ratio and the expression level distribution. It yields the number of efficient plasmids per complex and elucidates the origin of the associated noise, consequently providing a platform for evaluating and improving non-viral vectors.

How many trimers? Modeling influenza virus fusion yields a minimum aggregate size of six trimers, three of which are fusogenic.

Conflicting reports in leading journals have indicated the minimum number of influenza hemagglutinin (HA) trimers required for fusion to be between one and eight. Interestingly, the data in these reports are either almost identical, or can be transformed to be directly comparable. Different statistical or phenomenological models, however, were used to analyze these data, resulting in the varied interpretations. In an attempt to resolve this contradiction, we use PABM, a brane calculus we recently introduced, enabling an algorithmic systems biology approach that allows the problem to be modeled in a manner following a biological logic. Since a scalable PABM executor is still under development, we sufficiently simplified the fusion model and analyzed it using the model checker, PRISM. We validated the model against older HA-expressing cell-to-cell fusion data using the same parameters with the exception of three, namely HA and sialic acid (SA) surface densities and the aggregation rate, which were expected to be different as a result of the difference in the experimental setup. Results are consistent with the interpretation that a minimum aggregate size of six HA trimers, of which three undergo a conformational change to become fusogenic, is required for fusion. Of these three, two are free, while one is bound. Finally, we determined the effects of varying the SA surface density and showed that only a limited range of densities permit fusion. Our results demonstrate the potential of modeling in providing more precise interpretations of data.