We compared the syntheses and turnovers of two proteins related to the polyoma small T antigen synthesized in Escherichia coli from plasmids containing polyoma genomic segments joined to lac control elements. A protein with an authentic polyoma N terminus was more unstable than a protein with N-terminal amino acids derived from 0-galactosidase. Both were more unstable than most bacterial proteins.The polyoma and simian virus 40 (SV40) genomes encode early proteins (T antigens) which are involved in virus replication and neoplastic cell transformation (see reference 6 for review). Polyoma encodes three T antigens (large, medium [or middle], and small); SV40 encodes two (large and small). It would be useful to be able to synthesize the T antigens in large amounts to facilitate their purification and characterization. Approaches to doing this are being developed both in animal cells and in bacteria. The SV40 large T antigen has been expressed at high levels in mamalian cells, under the control of an adenovirus late promoter (5). The SV40 small T antigen has been expressed at high levels in Escherichia coli, under the control of a lac promoter (4).We previously described the synthesis in E. coli of a protein related to the polyoma small T antigen (1). This protein is expressed by a plasmid containing a segment of the polyoma genome joined to a lac control element in such a way that a "fused" protein results. The fused protein is initiated in 0-galactosidase sequences in the lac control element and is continued into the polyoma small-T-antigen coding region in the correct frame for translation. We have now constructed additional plasmids containing another lac control element which permitted the synthesis of a nonfused protein, initiated at the small-T-antigen initiation codon. We describe here the syntheses of the two proteins and a comparison of their stabilities in E. coli.We joined the coding region for the polyoma small T antigen to a lac operon control element in such a way that translation would begin at the initiation codon for the small T antigen. We used the general approach described by Roberts et al. (3), seeking to place the ribosome binding sequence of the lac control element at an optimum distance from the initiation codon. The optimum distance is not precisely predictable but is often seven to nine nucleotides. Therefore, we constructed a group of plasmids having the ribosome binding site at various distances from the small-T-antigen initiation codon and tested their abilities to produce a protein related to the small T antigen.The procedure is summarized in Fig. 1. We cleaved a plasmid, pgt-6, containing the small-Tantigen coding region, at an EcoPJ restriction enzyme site 16 nucleotides upstream from the initiation codon. We digested the plasmid DNA briefly with exonuclease III and then with S1 to produce molecules having ends at various distances less than 16 nucleotides from the initiation codon, as shown in Fig. 2. We then inserted a lac control fragment into the digested plasmid molecules. The...