Although it is generally accepted that the chromosome is divided into
elementary subunits, the structural and functional domains, the organi
sation of these structures at the molecular level is not well understo
od. In particular, the domain boundaries are not easily identifiable.
Several possible candidates such as MARs/SARs, insulators, LCRs, palin
dromic sequences, or easily melting sequences have been found in the r
egions having properties one would except for boundaries. None of thes
e elements, however, has been found in all of the constructs functioni
ng as boundaries in tests in vivo. Recent work suggests that the commo
n denominator might be the presence of GC-rich oligonucleotide streche
s and the formation of the chromatin hypersensitive sites. A model is
discussed in which ''unusual'' structures, in particular the four-stra
nded DNA sequence elements containing unpaired bases, play the role of
domain boundaries.