The role of binding site cluster strength in Bicoid-dependent patterning in Drosophila

Abstract

The maternal morphogen Bicoid (Bcd) is distributed in an embryonic gradient that is critical for patterning the anterior-posterior (AP) body plan in Drosophila. Previous work identified several target genes that respond directly to Bcd-dependent activation. Positioning of these targets along the AP axis is thought to be controlled by cis-regulatory modules (CRMs) that contain clusters of Bcd-binding sites of different "strengths." Here we use a combination of Bcd-site cluster analysis and evolutionary conservation to predict Bcd-dependent CRMs. We tested 14 predicted CRMs by in vivo reporter gene assays; 11 show Bcd-dependent activation, which brings the total number of known Bcd target elements to 21. Some CRMs drive expression patterns that are restricted to the most anterior part of the embryo, whereas others extend into middle and posterior regions. However, we do not detect a strong correlation between AP position of target gene expression and the strength of Bcd site clusters alone. Rather, we find that binding sites for other activators, including Hunchback and Caudal correlate with CRM expression in middle and posterior body regions. Also, many Bcd-dependent CRMs contain clusters of sites for the gap protein Kruppel, which may limit the posterior extent of activation by the Bcd gradient. We propose that the key design principle in AP patterning is the differential integration of positive and negative transcriptional information at the level of individual CRMs for each target gene.

Fig. 1.

Regulation of otd and hb by the Bcd morphogen. (A–D) Lateral views of blastoderm-stage embryos showing bcd mRNA (A), Bcd protein expression (B), and the mRNA expression patterns of the Bcd target genes otd (C) and hb (D). All embryos are oriented with anterior to the left and dorsal up. (E) Schematic summary of the data shown in B–D. Thresholds for Bcd-mediated activation of otd (T1) and hb (T2) and the percentage of EL on the x axis are shown. (F–I) Computational analysis of the Bcd-binding site clusters in the otd and hb loci. (F and G) Graphical representations of Bcd clusters in 4-kb sequences from the otd (F) and hb P2 (G) genomic regions. (Lower) Each predicted Bcd site is shown as an individual dot in the scatter plots, with PWM scores on the y axis. (Upper) Colinear cluster view representations of the data in the scatter plots (see Materials and Methods). Cluster significance is represented by a color code (far right): cool colors (light blue and green) indicate low-significance clusters; hot colors (orange and red) represent highsignificance clusters. (H and I) Results of mutating one (H) or two (I) highaffinity sites (red asterisks) in the hb P2 region. [The image in B is reproduced with permission from the embryo tu9 entry of the FlyEx database (Copyright 1998, David Kosman and John Reinitz).]

Fig. 2.

Expression patterns directed by Bcd-regulated CRMs. Wild-type mRNA expression patterns (Left) are shown for the indicated genes, as well as lacZ RNA patterns in transgenic embryos carrying reporter constructs driven by computationally predicted CRMs (Right). All embryos are oriented with anterior to the left and dorsal up. (F, J, L, P, and T) Several CRMs direct patterns that do not perfectly coincide with the patterns of their respective endogenous genes, suggesting that other sequences are required for the wild-type expression pattern.

Fig. 3.

Cluster analyses of all known Bcd-dependent CRMs. (A–C) Correlations between PBPs (expressed as the percentage of EL) for each CRM (x axes) and three different CRM features: site number (A), average PWM score at >5.0 (B), and highest PWM score for a single site (C). (D) cluster view representations of the 4-kb genomic regions containing known Bcd-dependent CRMs (see Fig. 1 legend). CRMs are classified into groups according to ranges of PBPs. The average PBP for each CRM is shown above each image.

Fig. 4.

Analysis of Bcd, Hb, and Kr binding-site clusters in Bcd-dependent CRMs. Each graph depicts individual scans (Bcd, left; Hb, center; Kr, right) of experimentally verified CRMs, which are grouped according to their major transcriptional inputs (shown on left). The average PBP for each CRM is shown above each panel. Color codes representing cluster significance for each regulatory factor are also shown.