The involvement of different structural domains of the coat protein (CP) of turnip mosaic virus, a potyvirus, in establishing and/or maintaining particle assembly was analyzed through deletion mutants of the protein. In order to identify exclusively those domains involved in protein–protein interactions within the particle, the analysis was performed by agroinfiltration “in planta”, followed by the assessment of CP accumulation in leaves and the assembly of virus-like particles lacking nucleic acids, also known as empty virus-like particles (eVLP). Thus, the interactions involving viral RNA could be excluded. It was found that deletions precluding eVLP assembly did not allow for protein accumulation either, probably indicating that non-assembled CP protein was degraded in the plant leaves. Deletions involving the CP structural core were incompatible with particle assembly. On the N-terminal domain, only the deletion avoiding the subdomain involved in interactions with other CP subunits was incorporated into eVLPs. The C-terminal domain was shown to be more permissive to deletions. Assembled eVLPs were found for mutants, eliminating the whole domain. The C-terminal domain mutants were unusually long, suggesting some role of the domain in the regulation of particle length. The identification of the CP domains responsible for eVLP formation will allow for new approaches to protein stretch replacement with peptides or proteins of nanobiotechnological interest. Finally, specific cases of application are considered.