In vitro and in vivo footprint analysis of the promoter of carcinoembryonic antigen in colon carcinoma cells: effects of interferon gamma treatment.

The analysis of the carcinoembryonic antigen (CEA) promoter in the colon carcinoma lines HT-29 and SW403, using HeLa as a control, was performed using gel mobility shift assays and in vitro and in vivo footprinting before and after IFN-gamma treatment. Using a 332-bp probe extending from the start of translation (+1 to -331), we detected 4-5 specific complexes that increased with intensity with IFN-gamma treatment as measured by a gel mobility shift assay. In contrast, no complexes were observed for probes covering the regions -500 to -1000 and -1000 to -1500. DNase I in vitro footprinting with the 332-bp probe revealed three footprints, FP-I to FP-III, none of which changed during the course of IFN-gamma treatment. Using probes corresponding to each footprint, 6-7 specific complexes were observed by gel mobility shift assays. Although minor changes were observed on IFN-gamma treatment, no consistent pattern was observed for all cell lines tested. Several of the proteins involved in the promoter complexes were identified by antibody super shifts, UV cross-linking, and Southwestern blotting. FP-I bound an Sp1-like protein, binding to a GT box sequence, and USF (upstream regulatory factor). FP-II and FP-III bound Sp1, binding through the consensus sequence for a GC box. Lower molecular weight complexes of an unknown nature were observed with sequence specificity for both single- and double-stranded DNA. DNase I in vivo footprints confirmed the boundaries of FP-I to FP-III and revealed a fourth but weaker footprint, FP-IV. The strongest in vivo footprints were observed for SW403 cells, with weaker and no footprints observed for HeLa cells, thus correlating with the degree of CEA transcriptional activity (HeLa cells make no CEA mRNA). DNase I hypersensitive sites correlated well with the boundaries of the footprints and also revealed activity around the start of transcription (-110). The specific pattern for DNase I hypersensitivity for Sp1 in the CEA promoter was the same as observed for the SV40 early promoter. In vivo footprinting with dimethyl sulfate revealed protein binding at the Sp1 consensus sequences in FP-II and FP-III and at the USF consensus sequence in FP-I. We conclude that in vivo footprinting is the most accurate predictor of the state of transcriptional activity of the CEA gene. It is also likely that Sp1 and USF play a major role in CEA transcriptional activation and that the majority of IFN-gamma effects are at the posttranscriptional level.