Abstract
Spliceosome is often assembled across an exon and undergoes rearrangement to span a neighboring intron. Most states of the intron-defined spliceosome have been structurally characterized. However, the structure of a fully assembled exon-defined spliceosome remains at large. During spliceosome assembly, the pre-catalytic state (B complex) is converted from its precursor (pre-B complex). Here we report atomic structures of the exon-defined human spliceosome in four sequential states: mature pre-B, late pre-B, early B, and mature B. In the previously unknown late pre-B state, U1 snRNP is already released but the remaining proteins are still in the pre-B state; unexpectedly, the RNAs are in the B state, with U6 snRNA forming a duplex with 5'-splice site and U5 snRNA recognizing the 3'-end of the exon. In the early and mature B complexes, the B-specific factors are stepwise recruited and specifically recognize the exon 3'-region. Our study reveals key insights into the assembly of the exon-defined spliceosomes and identifies mechanistic steps of the pre-B-to-B transition.