It has long been known that retroviral spreading is more efficient when cells can physically interact with each other 
. Applying 4D imaging and single-particle tracking, we have demonstrated that the murine leukemia virus can redirect virus assembly to sites of cell–cell contact for transmission to neighboring cells. As such, our results support a model of polarized assembly as the primary cause for the accumulation of viral particles at zones of cell–cell contact (Model II in ). Our data contribute to the emerging picture that several steps of the viral life cycle are efficiently coordinated at sites of cell–cell contact. Future work will reveal to what extent our model applies to other viruses and experimental conditions.
Our work is based on the ability of spinning disc confocal microscopy to detect de novo assembly and monitor the subsequent spatial movement of completely assembled particles. Applying a cautious definition of contact zones, our visual approach revealed an approximately 10-fold enhancement of virus assembly at sites of cell–cell contact. In the absence of cell–cell contact, particle release from producer cells into the culture supernatant was observed, consistent with the production of cell-free virus. Yet, in the context of coculture, MLV assembly was strongly directed towards sites of cell–cell contact, followed by efficient transmission to target cells. These data indicated that although assembly occurs randomly at plasma membrane, assembly becomes polarized following the establishment of cell–cell contact.
In an effort to understand the mechanism of the enhancement of assembly at sites of cell–cell contact, we observed no acceleration of assembly. On average, the assembly time observed for MLV in HEK293 cells was approximately 15 min, slower in comparison to the approximately 8 min observed for HIV in HeLa cells 
. MLV assembly was even slower in COS-1 cells, averaging 20.2 min for 79 events, suggesting that assembly time varies depending on the cell type (Figure S1
, unpublished data). Future experiments carried out in the same cell type in parallel are required to address the observed differences between HIV and MLV.
Although virus assembly per se was not accelerated at the sites of cell–cell contact, Gag proteins that drive virus particle assembly were recruited to cell–cell contacts. An elevation of Gag levels at contact sites may increase the frequency of nucleation, thereby enhancing virus assembly. The polarization of assembly required the cytoplasmic tail of the viral Env glycoprotein. Evidence for a communication between the cytoplasmic tail of retroviral Env and Gag proteins has been reported 
. Env expressed in polarized cells such as MDCK cells and neurons can relocalize Gag 
. In this work, we demonstrate that the establishment of cell–cell adhesion following Env/receptor interactions can break symmetry and establish polarity in otherwise nonpolarized fibroblasts. Future work is needed to understand whether the communication between Env and Gag is direct or indirect.
Our results reinforce similarities between virological and biological synapses in that the establishment of cell–cell adhesion is followed by polarization and the directed delivery of ligands towards sites of cell–cell contact 
. Our data suggest that the MLV Env glycoprotein functions analogously to a cellular adhesion protein that establishes cell–cell contact and polarizes cells. Intriguingly, once MLV Env is packaged into virions, during or soon after virus budding, the cytoplasmic tail is cleaved off by the viral protease 
. As such, the viral protease transforms an adhesion protein into a highly fusogenic fusion protein to mediate virus-to-cell fusion. This mechanism represents yet another clever adaptation and utilization of cellular principles by viruses to favor efficient viral spreading.