Patterns of polyadenylation site selection in gene constructs containing multiple polyadenylation signals

We have constructed a series of plasmids containing multiple polyadenylation signals downstream of the herpes simplex virus type 1 (HSV) thymidine kinase (tk)-coding region. The signals used were from the simian virus 40 (SV40) late gene, the HSV tk gene, and an AATAAA-containing segment of the SV40 early region. This last fragment signals polyadenylation poorly in our constructs and not at all during SV40 infection. All plasmids contained the SV40 origin of replication. Plasmids were transfected into Cos-1 cells; after 48 h, cytoplasmic RNA was isolated and the quantity and 3'-end structure of tk mRNAs was analyzed by using S1 nuclease protection assays. In all constructs, all polyadenylation signals were used. Increasing the number of poly(A) signals 3' to the tk-coding region did not affect the total amount of polyadenylated RNA produced, even with the weakest signal. Increasing the distance between two signals caused an increase in the use of the 5' signal and a decrease in the use of the 3' signal. Changing the distance between the 5' cap and first signal did not affect signal use. Analyses of cytoplasmic mRNA stability, nuclear RNA distribution, and transcription in the polyadenylation signal region indicated that the distribution of tk RNAs ending at different poly(A) sites was the result of poly(A) signal choice, not other aspects of RNA metabolism. Four possible mechanisms of polyadenylation signal recognition are discussed.