Identification and functional analysis of human transcriptional promoters

Abstract

Genomic and full-length cDNA sequences provide opportunities for understanding human gene structure and transcriptional regulatory elements. The simplest regulatory elements to identify are promoters, as their positions are dictated by the location of transcription start sites. We aligned full-length cDNA clones from the Mammalian Gene Collection to the human genome rough draft sequence to estimate the start sites of more than 10,000 human transcripts. We selected genomic sequence just upstream from the 5' end of these cDNA sequences and designated these as putative promoters. We assayed the functions of 152 of these DNA fragments, chosen at random from the entire set, in a luciferase-based transfection assay in four human cultured cell types. Ninety-one percent of these DNA fragments showed significant transcriptional activity in at least one of the cell lines, whereas 89% showed activity in at least two of the lines. We analyzed the distributions of strengths of these promoter fragments in the different cell types and identified likely alternative promoters in a large fraction of the genes. These data indicate that this approach is an effective method for predicting human promoters and provide the first set of functional data collected in parallel for a large set of human promoters.

Figure 1.

Distribution of promoter strengths. The distribution of the promoter strengths for the 138 positive clones of the 152 we tested is shown at top on a log scale in the 4 cell types tested. The number of promoters that fall within each bin is shown on the Y axis, and bin boundaries are denoted on the X-axis. We calculated promoter strength as a fold increase of luciferase activity over the negative controls in a given cell type. The black bars indicate promoters that fall above our threshold value for a functional promoter, and the white bars indicate those below that threshold (see Methods). Bins that contain both positive and negative promoters have boundaries that span the threshold value.

Figure 2.

Distribution of the variance of promoter strength. We calculated the coefficient of variance for each individual promoter in the four cell types to estimate the variance of promoter strength. By measuring the standard deviation relative to the mean, we compared the variation between both strong and weak promoters. The promoters in the left tail (low variance) and right tail (high variance) of the distribution may be more likely to have constitutive and cell type-specific activities, respectively.