Recombinant adenoviruses were constructed in which the viral ElA gene was deleted and the E1B promoter was replaced by the rat albumin, mouse p-major globin, or mouse immunoglobulin heavy-chain promoter.After infection of human or rat hepatoma cells, ElB-containing mRNAs could be detected only from the virus containing the albumin promoter. Conversely, only the immunoglobulin promoter was active in virus-infected myeloma cells. However, in hepatoma cells transcription from the albumin promoter in the virus was much less than that of the endogenous cellular albumin gene or of other viral genes. In primary mouse hepatocytes endogenous albumin gene transcription was high immediately after plating but declined within 24 h. Expression of the albumin promoter in the virus paralleled that of the cellular albumin gene. From these results it appears that cell-specific expression of albumin depends on the presence of tissue-specific trans-acting factors, but the presence of such factors does not suffice for a maximal rate of transcription, a conclusion that requires direct comparison within a differentiated cell of a newly introduced and preexisting active cell gene.Tissue-specific gene expression in vertebrates appears most often to result from regulation of transcriptional initiation (11,12,36,41,44). In our previous experiments on gene control in mouse hepatocytes and cultured mouse and rat hepatoma cells, we have noted two aspects of transcriptional control: specificity among cell types (12, 36) and a higher differential rate of transcription for liver-specific genes by adult hepatocytes on the one hand and fetal hepatocytes, hepatoma cells, and cultured adult cells on the other (6-8, 36).While cell-specific expression of genes introduced by transfection into cultured differentiated cell lines from the pancreas, lens, the P-lymphocytic series, and liver has been demonstrated, (5,13,15,16,19,34,37,42), the direct measurement of transcription rates following transfection is difficult to determine. Moreover, transfection experiments do not afford a comparison between the rate of transcription of endogenous and newly introduced genes because only a small fraction of the exposed cells actually take up and express the plasmid DNA. A further difficulty in transfection experiments is that all cell types are not equally receptive to uptake or the expression of newly introduced DNA, making comparisons between cell types very difficult.Finally, the gene expression assay in most reported transfection experiments relies on the steady-state amount of an mRNA or protein product, often the levels of bacterial chloramphenicol acetyltransferase (CAT) activity produced from a fusion of the regulatory region of the tissue-specific gene with the gene encoding the CAT protein. Such steadystate measurements are usually made only after prolonged incubation time (24 to 72 h). Use of this assay to study transcriptional control requires the assumption that both an unusual mRNA and protein product are equally stable in different cells under diff...