Altered dinucleotide content within the latently transcribed regions of the DNA of alpha herpes viruses--implications for latent RNA expression and DNA structure.

The alphaherpesviruses establish latent infections in sensory neuronal cells, during which a number of latency-associated RNA transcripts (LATs) are produced. The reasons for the production of the LATs, however, and the mechanism by which the LAT region of the genome remains active during latency are unknown. Here we have analysed alphaherpesvirus sequences in an attempt to assess any differences between latently active regions of the genomes and the genomes as a whole which might be necessary for this differential expression pattern or for the LAT function, whatever that might be. We show that the LAT regions of all of the four alphaerpesviruses examined exhibit a previously unidentified and marked increase in CC + GG dinucleotides within the region and that three of the four viruses examined also show a marked decrease in CG + GC dinucleotides. The CC + GG increase was shown to occur throughout the LAT regions of these viruses, but the GC + CG decrease was shown to be confined only to LATs which are thought to be expressed as introns, and also to the intron sequences of the IE1 genes (at least for herpes simplex viruses 1 and 2), which overlap the LAT region. Bovine herpesvirus, the virus which did not show the GC + CG reduction in its LAT region but did show the GG + CC increase, is thought not to express LATs by a mechanism involving splicing, again indicating that the CG + GC reduction may be an intron-related phenomenon and also further suggesting that the two effects of CG + GC decrease and GG + CC increase may be independent of one another. Possible reasons for these unusual dinucleotide frequencies within the latently active regions of the alphaherpesviruses relating to DNA and RNA structure and to LAT function are discussed.