I offer this as an echo to Elio's post from last October, Teaching E. coli to Endocytose. There Elio reported the recent education of E. coli by the heterologous expression of a mammalian gene. I tell of a bacterium instructed by a phage.
Apparently Pseudomonas phage φ6 missed that classic 1952 paper by Hershey and Chase. You know, the one where they radiolabeled either T2 or T4 phages with P32 and S35, mixed them with susceptible E. coli as hosts, let the phages adsorb and begin the infection, then whirred them in a Waring blender. The P32-labeled DNA entered the host cells, while the S35-labeled capsid proteins remained outside. This became the paradigm for phage infection, slightly tempered later to accommodate the entry, along with the genome, of various 'internal proteins' that had been packaged inside the capsid. However, for eukaryotic viruses that have double-stranded RNA (dsRNA) genomes, it's a whole different story. Literally. The whole capsid enters the cell. I discussed some of the benefits of this strategy in my post about the 'viral turtles' that infect yeast. Most dsRNA viruses rely on the host to take up the virion by endocytosis. Since Bacteria, it is widely believed, don't endocytose, this would seem to preclude phage virions from playing this game. But apparently phage φ6 missed this injunction, too.
Click "source" to read more.