Polyadenylation site selection cannot occur in vivo after excision of the 3'-terminal intron

Splicing of 3'-terminal introns and polyadenylation of pre-mRNAs can be coupled in an appropriate cell-free system. However, definitive evidence has been lacking as to whether these events are coupled in vivo and whether the order of these two processing events is obligatory. Here, we investigated these questions by examining the in vivo processing of transcripts that differ solely by the precise insertion of an intron within the first of two polyadenylation signals. Quantitative S1 nuclease mapping and PCR techniques were utilized to analyze the processed RNAs that accumulated in monkey cells transfected with plasmids encoding these transcripts. We found that, whereas all of the primary transcripts that lacked the inserted intron were processed via utilization of the 5'-proximal polyadenylation signal, none of the transcripts initially disrupted in this signal were processed this way even though the disrupting intron had been properly excised and excision sometimes preceded polyadenylation. In addition, deletion of the second polyadenylation signal resulted in failure of spliced transcripts to accumulate. We conclude that selection of, but not necessarily cleavage at the polyadenylation site precedes excision of the 3'-terminal intron in vivo; although coupling exists during selection of the sites to be used for polyadenylation and excision of the 3'-terminal intron, the actual order of the subsequent enzymatic reactions is probably simply a reflection of their relative kinetics.