1) That was Rutherford's discovery, not Bohr's.
2) The Bohr model (i.e. electrons traveling in discrete orbits around the atom) was disproven long ago. Electrons in atoms are described probabilistically, not as discrete particles traveling in planetary orbits, which would violate the Heisenberg Uncertainty principle.
3) Virtual particles are only a manifestation of a particular kind of mathematical treatment (i.e. perturbation theory). It is not clear that they have any independent "reality" .. many people would claim that they do not.
However, virtual particles can be a useful mnemonic for qualitative rationalization of how the "communication" between the electron and nucleus obeys relativity. Namely, the electromagnetic force between electron and nucleus can be mathematically construed as an exchange of virtual photons which carry the momentum between the real particles. Thus if the nucleus suddenly disappeared instantaneously, there would be a slight delay before the electron was aware of it, because "virtual particles" emitted by the nucleus before it disappeared would still cause the electron to behave as if the nucleus were still there.
Note that this is description is HIGHLY qualitative (verging on a pop-sci description). However, that is basically what the Feynman diagrams (which are a visual representation of the mathematical perturbation theory treatment) indicate happens. The reason this picture gained traction is because the QED description of the electron-nucleus interaction is very hard to solve exactly, and the perturbation theory approximation gave excellent agreement with experiment. In fact, QED successfully reproduced the relativistic shift of the H-atom energy levels called the "Lamb shift". This can be rationalized as follows: the virtual photons carrying the EM force between the electron and nucleus have a very small probability of becoming a positron-electron pair. These two cases would have different energies, and the Lamb shift is the manifestation of that effect ... QED calculations of the Lamb shift give incredibly good agreement with experiment.
So, are there virtual particles in the empty space between electron and nucleus? Well, it kind of depends on your point of view. Personally, I find the QED description in terms of Feynman diagrams quite useful for qualitatively describing what is going on. It is also clearly responsible for providing incredibly accurate calculations in some cases. However, I also think it is important to recognize that this is still just based on a mathematical description ... the physical reality could be quite different. However, that last comment can be applied to pretty much any phenomenon in quantum mechanics and atomic physics.