Proper control of gene expression is necessary, yet it's unclear how regulatory elements interact with only correct genes in the genome. One possibility is that the specialized regulatory regions immediately surrounding transcription start sites, known as promoters, dictate the ability to interact with nearby regulatory elements. If true, we expect genes with similar promoter sequences to be co-expressed, and those with dissimilar promoters to have differential expression.
We determined co-expression indices for all adjacent promoters in D. melanogaster using data from several sources, including bi-hourly embryonic time-course data generated by modENCODE(Nature 459:927), a complete life-cycle time-course from RAMPAGE profiling (Genome Res. 23:169), and GRO-seq-derived promoter activity data for cultured S2 cells (Cell Rep. 2:1025). We then used an alignment-free sequence similarity measure (“N2”) to calculate relatedness scores for every adjacent promoter pair and tested for correlation between sequence relatedness and co-expression index. Distances between paired promoters were also tested for correlation with co-expression.
Resulting analyses of the RAMPAGE dataset showed a positive trend between co-expression and sequence similarity among a small subset of adjacent promoters as compared to randomized data. No correlation appeared for the majority of pairs overall. A smaller minority of modENCODE data showed similar positive correlation. This disparity was likely due to the longer and higher resolution time-course of the RAMPAGE dataset. The co-expressed groups of promoters from S2 cells did not vary substantially from dissimilarly expressed pairs in terms of their sequence similarity.
Although these analyses failed to reveal a strong connection between promoter sequence and coordinated gene activity, other metrics may provide better insight. We are continuing to evaluate other sequence similarity measures, for instance whether discrete sets of core promoter motifs (rather than entire promoter sequences) are adequate in classifying similarities in promoter activity.