Characterization of a structurally tricistronic gene of human cytomegalovirus composed of U(s)18, U(s)19, and U(s)20.

A tricistronic gene mapped between 0.91 and 0.93 map units within the EcoRI D fragment of the human cytomegalovirus unique short region (U(s)) has been cloned, sequenced, and expressed in vitro. Cloned cDNAs of 2.3, 1.8, and 1.1 kb derived from this region were isolated from a lambda gt11 cDNA library made from virus-infected fibroblasts and used for this study. Two major classes of 3'-coterminal mRNAs, 2.8 and 1.1 kb, were transcribed from this region. Sequence analysis of the cDNAs and the upstream genomic DNA revealed three open reading frames (ORFs), U(s)18, U(s)19, and U(s)20, and a common polyadenylation signal located 15 bases upstream of the poly(A) tail of both the 2.85- and 1.1-kb mRNAs. Protein structure analyses predicted the existence of multiple hydrophobic moieties, suggesting that the U(s)18, U(s)19, and U(s)20 polypeptides were transmembrane proteins. The major transcription initiation site, determined by primer extension and S1 nuclease mapping, for the 2.85-kb transcript was located right at the first initiation codon of the U(s)20 ORF. There was no typical TATA box or CAAT box upstream of the 2.85-kb mRNA cap site except for a TATAAGA sequence that was found about 210 bp downstream from the major cap site. The 1.1-kb transcript was initiated 33 bp upstream of the U(s)18 translation initiation site, and an atypical TATA box sequence (GATAAGA) was found 22 bp upstream of the transcription start site. Differences in transcription kinetics and sensitivities to metabolic inhibitors suggest that they were regulated by different mechanisms; the 2.85-kb mRNA belongs to the early (beta) class of transcripts, while the 1.1-kb mRNA is a late (gamma) message. Subgenomic DNA segments derived from the U(s)18, U(s)19, and U(s)20 ORFs were subcloned and overexpressed in Escherichia coli as fusion proteins with glutathione-s-transferase. Western immunoblot analysis with antibodies against the U(s)18, U(s)19, and U(s)20 fusion proteins detected virus-specific polypeptides with molecular sizes of 36, 32, and 43 kDa, respectively. All three antibodies also exhibited a positive immunofluorescence reaction with human cytomegalovirus-infected cells harvested at late stages of infection.