Formation of the 3' end of Ul and U2 small nuclear RNA (snRNA) precursors is directed by a conserved sequence called the 3' box located 9 to 28 nucleotides downstream of all metazoan Ul to U4 snRNA genes sequenced so far. Deletion of part or all of the 3' box from human Ul and U2 genes drastically reduces 3'-end formation. To define the essential nucleotides within this box that direct 3'-end formation, we constructed a set of point mutations in the conserved residues of the human Ul 3' box. The ability of the various mutations to direct 3'-end formation was tested by microinjection into Xenopuws oocytes and transfection into HeLa cells. We found that the point mutations had diverse effects on 3'-end formation, ranging from no effect at all to severe inhibition; however, no single or double point mutation we tested completely eliminated 3'-end formation. We also showed that a rat U3 3' flank can effectively substitute for the human Ul 3' flank, indicating that the 3' boxes of the different U snRNA genes are functionally equivalent.The Ul, U2, and U3 small nuclear RNA (snRNA) genes are thought to be transcribed by RNA polymerase II since their transcription is sensitive to low levels of a-amanitin (10, 35) and the snRNAs are capped. However, there are two striking differences between the synthesis of other RNA polymerase II transcripts (mRNAs) and that of the snRNAs. First, snRNA promoters lack sequence homology with the TATA and CAAT boxes found in most mRNA promoters but share several unique conserved sequences among themselves (1,28,35,(40)(41)(42)(43). Second, the 3' ends of snRNAs are not polyadenylated and seem to be formed by a different mechanism from that of mRNA 3' ends.The 3' ends of most mRNAs are formed by endonucleolytic cleavage of a long precursor RNA followed by polyadenylation. The cleavage reaction is directed both by a highly conserved AAUAAA sequence found 10 to 30 nucleotides (nt) upstream of the poly(A) addition site, as well as by sequences located downstream of the cleavage site (13, 15, 32; for a review, see reference 2). Evidence suggests that this cleavage and polyadenylation step may involve a small nuclear ribonucleoprotein (16,33,34). The 3' ends of histone mRNAs are not polyadenylated but are also formed by an RNA processing event. This processing is directed by a highly conserved GC-rich hairpin at the 3' end of the mRNA followed by a purine-rich block about 15 nt further downstream. This processing event is known to require U7 snRNA, at least for H3 mRNA (11,12,39; for a review, see reference 2).The 3' ends of mature Ul, U2, and U3 snRNAs are formed in at least two steps. First, precursors are formed with a few additional nucleotides at their 3' end. These precursors are then processed into the mature snRNA (3,7,18,27,41,44,45). Two pieces of evidence now indicate that the 3' ends of the snRNA precursors are formed by either termination of transcription or an RNA processing event directly coupled to transcription. First, run-on transcription in isolated nuclei does not proceed fu...