The papillomaviruses cause benign and malignant lesions of squamous epithelia in higher vertebrates. The complete lytic cycle of these viruses (including late gene expression) occurs only in the differentiated cells of the squamous epithelium. Malignant lesions and infected cells in culture do not produce virus. We have used bovine papillomavirus type 1 (BPV-l) as a model system for the study of papillomavirus late gene expression and its control. Transcriptional mapping data indicates that the late mRNAs which encode the major and minor capsid proteins are expressed from a strong viral transcriptional promoter (called the late promoter) which is active only in productively infected epithelium. Late mRNAs are almost undetectable in nonproductively infected (transformed) cells. Analysis of BPV-1 mutants with deletion of most of the late region demonstrated that the BPV-1 late polyadenylation site can be efficiently utilized even in transformed cells. Eukaryotic expression vectors have also been used to show that the late poly(A) site is a more efficient poly(A) site than the early poly(A) site. The inefficient use of the late poly(A) site when it is positioned in its normal context 3 kb downstream of the early poly(A) site may be due partially to transcription termination or pausing which has been shown to occur between the early and late poly(A) sites in transformed cells and partially to a short inhibitory element in the late 3'UTR. This late 3'UTR element has been shown not to function by destabilizing late mRNAs as was previously thought and may function at an RNA processing or transport level. We also have evidence that BPV-1 poly(A) site choice could be regulated by splicing factors. BPV-1 late promoter transcripts can be spliced into acceptors at nt 3225 or 3605 and this splice site choice is differentially regulated in nonproductively and productively infected cells. We have shown, using transfection studies, that exon sequences between these two splice sites modulate the use of the splice donor at nt 3764. In addition, we have shown that efficient use of the nt 3764 splice site in the absence of these exon sequences can suppress polyadenylation at the early poly(A) site.
