The regional integration of retroviral sequences into the mosaic genomes of mammals

We have reviewed here three sets of data concerning the integration of retr oviral sequences in the mammalian genome: (i) our experimental localization of a number of proviruses integrated in isochores characterized by differe nt GC levels; (ii) results from other laboratories on the localization of r etroviral sequences in open chromatin regions and/or next to CpG islands; a nd (iii) our compositional analysis of genes located in the neighborhood of integrated retroviral sequences. The three sets of data have provided a ve ry consistent picture in that a compartmentalized, isopycnic integration of expressed proviruses appears to be the rule ('isopycnic' refers to the com positional match between viral and host sequences around the integration si te). The results reviewed here suggest that: (i) integration of proviral sequenc es is targeted initially towards 'open chromatin regions'; while these exis t in both GC-rich and GC-poor isochores, the 'open chromatin regions' of GC -rich isochores are the main targets for integration of retroviral sequence s because of their much greater abundance; (ii) isopycnicity is associated with stability of integration; indeed, even non-expressed integrated retrov iral sequences tend to show an isopycnic localization in the genome; (iii) transcription of integrated viral sequences (like transcription of host gen es) appears to be associated, as a rule, with an isopycnic localization, as indicated by transcribed sequences that show an isopycnic integration and act in trans; (iv) selection plays a role in the choice of specific sites w ithin an isopycnic region; in exceptional cases [such as mouse mammary tumo r virus (MMTV) activating GC-rich oncogenes], selection may override isopyc nicity. (C) 1998 Elsevier Science B.V. All rights reserved.