Tight clustering of human hepatitis B virus integration sites in hepatomas near a triple-stranded region.

The open circular genome of human hepatitis B virus (HBV) is known to contain a partially double-stranded DNA with a single-stranded gap region of variable length. This circular structure of the genome is maintained by base-pairing of the 5' ends of the two DNA stands, the long or L(-) strand and the short or S(+) strand. By cloning, mapping, and sequencing studies, we have localized three recombinational junctions of the integrated HBV in two hepatoma samples, HT14 and FOCUS. Breakpoints of recombination derived from these results and those of others appear to be clustered and coincidental with the identified 5' or the deduced 3' end of the long-strand DNA, respectively. Statistical analysis of these results supports the hypothesis that integration preferentially occurs in an extremely narrow region on the HBV genome. This site-specific recombinational mechanism appears to be conserved among different HBV subtypes. No extensive sequence homology was found between each pair of the recombining parental molecules; however, at the site of crossover, 2- to 3-base-pair junctional homology was consistently observed. Examination of the patterns of the integrated HBV DNAs allowed us to categorize these various patterns into four different groups according to their end specificity and strand polarity. The molecular form of relaxed circle is proposed to be one major substrate for HBV integration. The effect of free strand in the integration of HBV is emphasized in this model. Unlike any other known DNA animal viruses, the site specificity of HBV integration appears to be similar to that of the retroviruses.