Virus Traffic

All known cellular cargos including protein and RNA complexes, vesicles, organelles and pathogens are bidirectionally transported by molecular motors on polarized microtubules to the cell center and periphery. This enables fine tuning and reversibility of transport. However, it is unknown if the motor activity is coordinated or works stochastically. Viruses are simple cargos, and their replication requires dynein and kinesin mediated transport. Here we carried out live cell microscopy, single virus tracking and trajectory segmentation (to extract directed runs length and velocity distributions) as a basis to computationally model bidirectional transport of human adenovirus type 2 on microtubules. Directed runs are the fastest intracellular motions observed for adenoviruses and are shown to be microtubule dependent.
Virus motions on microtubules, A. Helenius Laboratory (left), Animation of motor protein stepping, D.Bolinsky (right)

A stochastic model for virus traffic
We used a tug of war model in which the motors dynein (D) and kinesin (K) step [1-2] along a 1D microtubule,  bind [3-5] to and unbind [4-6] from the cargo. The number of available binding sites on the virus is indicated with ρ. Given the parameters r and ρ, trajectories are stochastically simulated and then segmented to extract length and velocity distributions for the directed (fast) motion runs.

Model parameters identified from experimental data using Evolution Strategies
The model parameters are identified by using a Covariance Matrix Adaptation Evolutionary Strategy to minimize the distance (Kullback-Leibler Divergence) between simulated and biological length and velocity distributions.

Model predictions for virus binding sites
Given the predicted range of 6-16 binding sites and the viral capsid structure, interfaces between protein hexon and protein IX (pIX) are left as candidates to harbor the motor binding sites. We imaged pIX deficient adenoviruses and the corresponding run length and velocity distributions show that bidirectionality is preserved.

Summary of Findings
  • The proposed model accurately reproduces motor activity
  • Found an optimal range of 6-16 binding sites
  • Virus dynamics are characterized by low number of bound motors
  • Strong correlation between velocity and number of motors
  • High dependence on the unbinding rates
  • We predict that hexon provides the motor binding sites

People: Mattia Gazzola, Basil Bayati

Collaborators: Urs Greber, Christoph Burkhardt (University of Zurich, Institute of Zoology)

Funding: SystemsX

  • Gazzola M, Burckhardt CJ, Bayati B, Engelke M, Greber UF, Koumoutsakos P (2009) A stochastic model for microtubule motors describes the in vivo cytoplasmic transport of human adenovirus. PLoS Computational Biology, 5(12). (pdf)