Transcription factors involved in A–P patterning. TFs involved in A–P patterning function in a hierarchical network to subdivide the embryo into 14 segments. The early maternal factors are expressed in broad gradients, with subsequent TF groups expressed in patterns that are increasingly refined. TFs involved in early segmentation or expressed in early patterns that were characterized in this study are grouped according to their initial stage of expression and they are color-coded to indicate the type of DNA-binding domain [Cys₂His₂ zinc fingers = blue, homeodomains = green, bHLH = gray, bZip = red, winged helix = pink, nuclear hormone receptor = orange, POU motif = light blue, paired motif = yellow, and HMG = lavender. Runt is black and was characterized as an alpha fusion (37)].

Overview of the omega-based B1H system. (A) Cartoon depicting recruitment of direct omega fusions (left) and omega–Zif12–HD fusions (right) to the weak promoter driving the HIS3 and URA3 reporters used in this system. The 28-bp library is positioned 7 bases upstream of the –35 box allowing the TF to bind to a recognition element up to three turns upstream of the promoter. The ZF10 library has the binding site for Zif12 (TGGGCGG) positioned 21 bases upstream of the promoter and the 10-bp randomized region is located immediately 5′ to this site. (B) Overview of Bait and Prey plasmids used in this system. Bait plasmids are constructed by cloning the TF of interest as a C-terminal fusion to omega (omega-TF hybrid). Homeodomains are cloned into a modified bait plasmid (pB1H2ω2-12) that results in their expression as an omega-ZF12-HD hybrid (Noyes, et al. manuscript in preparation). (C) Binding site selection procedure. A bait plasmid and the appropriate prey plasmid are transformed into the selection strain. Transformants are grown on minimal media lacking histidine and challenged with various concentrations of 3-AT. Surviving colonies represent a complementary interaction between the bait plasmid (TF) and a single member of the prey library. The library region from ∼20–25 surviving colonies are amplified by colony PCR and sequenced. The resulting sequences are analyzed by MEME (42) to recover the TF's recognition motif.

Specificities of the 35 TF in the A–P regulatory pathway characterized by the omega-B1H system. These TFs are grouped by DNA-binding domain family and color-coded as described in Figure 1.

Comparison of B1H-generated recognition motifs to previously published data. (Left panels) The B1H recognition motifs for Bcd, Kr and Tll are very similar to the motifs generated from DNase footprinting data (FlyREG) (27). In the case of Bcd, the high stringency data (10 mM 3-AT) is most similar to the previously described SELEX data (53), whereas the lower stringency data is more similar to the FlyREG data. (Right panels) The B1H recognition motifs for Cad, Gt and Kni differ significantly from the FlyREG data. For Cad, the B1H-generated data is similar to SELEX data on the chicken homolog (55), but provides better definition of the overall sequence preference.

Motifcount plots over 48 CRMs from D. melanogaster that drive early patterned expression. Comparison of the Motifcount plots based on the DNaseI (17) and B1H PWMs for Bcd, Cad, Hb, Hkb, Gt, Kni, Kr and Tll. In each plot, the red line indicates the TF expression profile over the embryo length (x-axis, 0 = anterior pole; y-axis, arbitrary units). The average number of binding sites for each factor over the CRMs that drive expression in each region (the Motifcount, see Materials and Methods for calculation) is indicated by a blue line plotted as a function of the z-score, where 0 is the genome-wide mean (indicated by the magenta line). Solid bars at the top of each graph indicate the window regions over which correlations between the factor expression profile and the Motifcount profile were calculated with the exception of Bcd and Cad, which were calculated over the entire region. P-values for strongest correlation/anticorrelation between the TF expression profile and its Motifcount along the A–P axis are listed to the right of the plots where the bold value indicates the most significant correlation.

Genome Surveyor display interface. A 20-kb region surrounding the ‘eve’ locus is displayed. Annotations for the D. melanogaster genome are shown at the top of the browser window. The predicted transcripts and genes in the D. melanogaster genome are indicated within the genomic region. Immediately below is a line indicating the regions where a high confidence alignment with the D. pseudoobsura genome has been assembled onto the melanogaster scaffold. Annotations for identified CRMs [downloaded from REDfly (62)] can also be displayed within this region. The user-configurable tracks for individual factors or groups of factors are displayed below the annotations. Multiple factor combination tracks can be displayed simultaneously. These tracks represent the average of the z-scores for each factor plotted over this genomic region for the combination of TFs selected by the user, where the factors included are indicated above each track (i.e. Kr, Bcd, Hkb, Tll and Hb, which were the anterior factor search set used to generate the list of hits in Table 1). The numbers in the upper left-hand corner indicate the maximum value (z-scores) for each plot, the estimated genome-wide mean and the mean + 2 SD, respectively. The positions of the genome-wide mean and the mean + 2 SD are also indicated on the plot by horizontal lines of the same color that transect the plot. In this view the two combination tracks (red) for the anterior factor search set are shown across D. melanogaster genome (mel) and the average over the D. melanogaster and D. pseudoobsura genomes (melpse). Both of these factor combinations contain a strong peak within the ‘eve’ stripe 1 CRM. Two other Combination tracks for other groups of factors (a different gap set and a pair-rule set) are also shown. These groups display significant peaks that overlap with other CRMs. Below the five Combination tracks are a number of tracks for individual factors. These tracks provide a rapid assessment of the individual factors that are potentially contributing to each combination track. For example, significant peaks for Bcd, Hkb, Kr and Tll all overlap with the stripe 1 CRM (blue box).