Initiation complexes formed by E. coli ribosomes in the presence of 32P-labeled A protein initiator region from R17 bacteriophage RNA have been treated with colicin E3 and disassembled by exposure to 1% sodium dodecyl sulfate. Electrophoresis on 9% polyacrylamide gels reveals a dissociabte complex containing the 30-nucleotide-long messenger fragment and the 50-nucleotide-long colicin fragment, which arises from the 3' terminus of the 16S RNA. The complex is a pure RNA-RNA hybrid; it is apparently maintained by a seven-base complementarity between the two RNA fragments.Detection of this mRNA rRNA complex strongly supports the hypothesis that during the initiation step of protein biosynthesis the 3' end of 16S RNA base pairs with the polypurine stretch common to initiator regions in E. coli and bacteriophage mRNAs. The implications of our findings with respect to the molecular mechanism of initiation site selection and mRNA binding to ribosomes, the role of rRNA in ribosome function, and species specificity in translation are explored.Shine and Dalgarno (1) originally suggested that a sequence near the 3' terminus of Escherschla coli 16S ribosomal RNA participates directly in the initiation of protein biosynthesis by forming several Watson-Crick base pairs with the messenger RNA. Indeed, one of the few common features of all ribosome-protected initiator regions analyzed so far is a polypurine stretch of 3 to 8 nucleotides located about 10 bases 5' to the initiator codon (Table 1). From 3 to 7 contiguous bases within this region of each mRNA can potentially pair with some portion of the polypyrimidine sequence found in the 3'-terminal Ti oligonucleotide of 16S RNA. Although the relevant 16S RNA sequence as determined by Shine and Dalgarno (1) conflicted with previous reports (2), its validity has now been confirmed in three additional laboratories (3-5).