One of the most puzzling questions in the field of galaxy evolution is how a
galaxy manages to grow a supermassive black hole of 10^6-10^9 solar masses at
its center and how this growth affects the galaxy itself. In those galaxies
hosting an active galactic nucleus (AGN), a dusty toroidal structure that
surrounds the central engine is thought to be the reservoir of molecular gas
that feeds the AGN and obscures the nucleus. However, black holes of 10^9 solar
masses already existed when the Universe was only ~1 Gyr old. To reach this mass
in such time, they must have started out as lower-mass seed black holes of
100-10^6 solar masses (the so-called intermediate-mass black holes, IMBHs) and
grow very fast via accretion and merger processes. Such seed IMBHs should be
present in the nucleus of low-mass galaxies and in the halos of large galaxies
(e.g., after tidal stripping of merging satellite low-mass galaxies).
Nevertheless, observational evidence of their existence is still scarce.

In this talk I will show dust maps of the central few kpc of a sample of AGNs
that reveal the presence of dust filaments just crossing in front of the
nucleus. The extinction of these nuclear dust lanes is sufficient to hide the
optical nucleus of the low-luminosity AGNs, thus casting doubts on the presence
of a dusty torus. This supports the finding, from molecular gas kinematics, that
the gas reservoir vanishes at low luminosities. Instead, the nuclear dust
filaments are observed to connect with kpc-scale dust structures, suggesting
that these are the feeding channels through which material flows from galactic
scales to the nucleus.
To probe the existence of the initial seed IMBHs from which AGNs grow, we
carried out a program aimed at studying jet radio emission in extreme
ultraluminous X-ray sources (ULXs), whose high X-ray luminosities can difficulty
be explained by stellar evolution models but make them compelling IMBH
candidates. The radio observations reveal compact radio emission from two ULXs,
which become potential IMBH candidates, as well as the first detection of
possible steady jet emission from an IMBH. With a total size of ~650 pc, this
source could be the largest non-nuclear extragalactic jet ever discovered. Its
location in the arm of a spiral galaxy undergoing a minor merger event indicates
that the nuclei of minor mergers remain amongst the best candidates for IMBHs.