Computing enhancers on the fly. M. Markstein 1,2, K. Senger 1, R. Zinzen 1, K-P. Yee 3, P. Markstein 4, V. Markstein 5, T. Cake 6, M. Levine 1. 1) Dept Molecular & Cell Biol, Univ California, Berkeley, Berkeley, CA; 2) Committee on Developmental Biology, University of Chicago, Chicago, IL; 3) Dept Computer Science, Univ California, Berkeley, Berkeley, CA; 4) Hewlett-Packard Labs, Palo Alto, CA; 5) in silico Labs, Woodside, CA; 6) OpenGenomics.org, Berkeley, CA.
Recent studies have established that genome-wide searches for clusters of transcription factor binding sites can identify bona fide enhancers for known and novel genes in the early fly embryo. However, the success rate of this approach is unclear because in each of these studies, only one or a few of the predicted enhancers were tested by transgenic analysis; the majority of predicted enhancers were instead assumed to be true or false based on the activity of neighboring genes.
Here we present the first comprehensive test of a bioinformatics screen for enhancers. We use transgenics to test each of the 15 predicted enhancers identified in a genome-wide search for clusters of high affinity Dorsal binding sites. Since Dorsal plays a role in both early dorsal-ventral patterning and larval immunity, we analyze the activity of each transgene for both roles. We show that 8 of the 15 clusters can drive expression along the d/v axis of the early embryo: 5 are associated with the d/v genes sog, brk, CG12443, Ady43A and Phm, while the other 3 are not associated with d/v genes. Preliminary results show that at least one of the 15 clusters is sufficient to respond in an immunity assay.
Ultimately we would like to be able to predict different classes of Dorsal responsive enhancers. To this end, we developed a program called MERmaid which has successfully identified novel motifs uniquely shared by each class of Dorsal responsive enhancers. MERmaid is available at www.opengenomics.org.