The immunoglobulin gene system has fascinated molecular biologists for over two decades, displaying a diverse array of the genetic gymnastics in which eukaryotes are wont to indulge, some in common with other genetic systems, others (so far) unique. Much attention has been paid to the DNA rearrangements ~-hat Ig genes undergo during B-lymphocyte ontogeny, to the genetic elements that confer tissue specificity on Ig gene expression, and to the mechanisms that diversify the repertoire of variable-region (antigen-binding) sequences. It is the purpose of this essay to focus on a different, but no less fascinating phenomenon: the process by which a single heavychain gene can give rise, in a regulated manner, to two mRNAs differing at their 3' ends and encoding polypeptides destined for secretion or cell-surface display, and on the roles played by transcription termination, polyadenylation and RNA splicing in this process. In all genes of higher eukaryotes studied so far, RNA polymerase II transcribes across the polyadenylation site(s) and terminates downstream of the DNA sequences coding for the 3' terminus of the mRNA (for review, see Ref. I). So far, the sequences at which transcription actually terminates are ili-defined, and indeed there often appear to be multiple sites extending over hundreds or even thousands of bases. However, transcripts with heterogeneous 3' ends can be edited accurately by the 3' processing events of endonucleolytic cleavage at a defined site, followed by polyadenylation; it is these latter mechanisms that appear to be chiefly responsible for determining the 3' end of an mRNA. What then of immunoglobulins? The heavy chains of membrane. bound IgM differ from those of the secreted form, being somewhat larger and with extremely h~drophohic C termini. Since there l~ only a single copy of the Cit gene per haploid mouse genome, this posed a ~aradox until it was demonstrated that the two tt polypeptides are encoded by two distinct mRNAs that differ at their 3' termini and that are derived from the same Cit gene by alternative RNA processing pathways ~-4. One form, the Dec mRNA, encodes the It chain of secreted IgM and the other, the itmem mRNA, encodes the corresponding poiypeptide of membrane-bound IgM. The fourth exun of the Cit gene (Cp4) encodes not only the fourth domain of the heavy chain, but also a C-terminal extension (S in Fig. 1) that is specific for the carhoxyl terminus of secreted it chains. A further 1.85 kb downstream are two exons (M1 and M2) encoding the C terminus of membrane-bound it chains. The mRNA specifying the secreted ~t chain incorporates the S sequence, but not the M1 or M2 exons. Conversely, the mRNA specifying membrane-bound it chains contains the M1 and M2 sequences, having excluded S by an RNA splice between Cp4 and M1. (Similar arrangements of S, M1 and M2 exons 3' to each of the Cy, Ce and Co~ genes specify the 3' ends of corresponding secretory and membrane mRNAs, with a somewhat more complicated ~;.tuation for C85.) During B-lymphocyte development there are striking changes in the ratios of the two tt mRNAs (e.g. Ref. 6). In early developmental stages (pre-B cells) the itsec and pmem mRNAs are concurrently produced, with lanem usually being the predominant species. Later, in IgM-secreting plasma cells, the itsec mR_NA becomes the predominant species. Two general models have been invoked to explain the control of production of these two mRNAs. One mechanism would yield the ltsec terminus by specific cleavage and polyadenylation of a common precursor transcript encompassing both ~ e c and Itmem exons, the other by termination oI transcription before the proem exons are reached. Recently, a considerable amount of data bearing on these models has TIG --September 1987, Vol. 3, no. 9