Impacts of the ubiquitous factor Zelda on Bicoid-dependent DNA binding and transcription in Drosophila

Abstract

In vivo cross-linking studies suggest that the Drosophila transcription factor Bicoid (Bcd) binds to several thousand sites during early embryogenesis, but it is not clear how many of these binding events are functionally important. In contrast, reporter gene studies have identified >60 Bcd-dependent enhancers, all of which contain clusters of the consensus binding sequence TAATCC. These studies also identified clusters of TAATCC motifs (inactive fragments) that failed to drive Bcd-dependent activation. In general, active fragments showed higher levels of Bcd binding in vivo and were enriched in predicted binding sites for the ubiquitous maternal protein Zelda (Zld). Here we tested the role of Zld in Bcd-mediated binding and transcription. Removal of Zld function and mutations in Zld sites caused significant reductions in Bcd binding to known enhancers and variable effects on the activation and spatial positioning of Bcd-dependent expression patterns. Also, insertion of Zld sites converted one of six inactive fragments into a Bcd-responsive enhancer. Genome-wide binding experiments in zld mutants showed variable effects on Bcd-binding peaks, ranging from strong reductions to significantly enhanced levels of binding. Increases in Bcd binding caused the precocious Bcd-dependent activation of genes that are normally not expressed in early embryos, suggesting that Zld controls the genome-wide binding profile of Bcd at the qualitative level and is critical for selecting target genes for activation in the early embryo. These results underscore the importance of combinatorial binding in enhancer function and provide data that will help predict regulatory activities based on DNA sequence.

Keywords: DNA binding; Drosophila; enhancer; pattern formation; pioneer factor; transcription.

Figure 1.

Zld-binding sites are overrepresented in active fragments. (A–D) Four typical reporter constructs driven by different DNA fragments show either Bcd-dependent patterns (A,B) or no expression in early embryos (C,D). Schematics below each embryo represent the distribution of Bcd-binding sites in corresponding DNA fragments. Bcd sites were identified by the ClusterDraw program (Lifanov et al. 2003) using the B1H PWM (Noyes et al. 2008). The cutoff P-value for Bcd sites was 5 × 10⁻³. (E) Actual counts of eight canonical Zld-binding sites in active and inactive fragments. P-values were calculated by a χ² test (Robin et al. 2007) and reflect the enrichment of each heptamer in active fragments. (F) Distribution of Zld-binding sites in active fragments (hatched bars) and inactive fragments (white bars).

Figure 2.

Adding Zld sites activates the inactive fragment HC_45 and fine-tunes enhancer activity. (A–E) Maps of HC_45 variants that carry different numbers and combinations of extra Zld sites. Blue bars represent Bcd-binding sites (P < 0.005), and red bars represent Zld sites introduced to the HC_45 element. The number of extra Zld sites increases from 0 to 4, corresponding to A–E. Within each box, different combinations of Zld sites are shown. (F–J) lacZ RNA expression in a wild-type embryo carrying HC_45-lacZ transgenes with increasing numbers of additional Zld sites. (K) lacZ expression driven by HC_45 with scrambled sequences in the nucleotides that were mutated to add new Zld sites in J. (L,M) lacZ expression driven by HC_45_4Z in zld⁻ (L) and bcd⁻ (M) embryos. Note that there is still some expression of the HC_45_4Z construct in zld mutants. This suggests that another unknown factor can bind the inserted sites. (N) Zld (in red) and Bcd (in blue) ChIP-qPCR with HC_45 (blank bars) and HC_45_4Z (hatched bars). Primers were chosen to amplify only the transgenic fragment. A flanking region near the landing site was used as negative control (see the Supplemental Material for sequences). Zld- and Bcd-binding signals were normalized with the negative control and expressed as fold enrichment (see the Materials and Methods). Error bars represent standard deviations. Significant differences (P < 0.05, Student's t-test) are shown by double asterisks.

Figure 3.

Zld regulates Bcd-dependent enhancers by affecting Bcd-binding. (A) A wild-type (wt) Bcd gradient at nc13. (B) A schematic shows the definition of type 0, type I, and type II enhancers. Numbers represent positions within the embryo (anterior is 100, and posterior is 0). (C–P) lacZ expression patterns driven by type I and type II enhancers in wild-type (C,F,H,J,M,O) and zld⁻ (D,G,I,K,N,P) embryos and enhancers with mutated Zld sites (E,L). (Q) Bcd ChIP_qPCR with different enhancers. Binding signals were normalized to the negative control defined in the legend for Figure 2. Error bars represent standard deviations from the mean. Significant differences (P < 0.05, Student's t-test) are shown by double asterisks.

Figure 4.

Three groups of Bcd-binding peaks behave differently in wild-type (wt) and zld⁻ embryos. (A–C) ChIP-seq read profiles for Bcd binding in wild-type (top panel) and zld⁻ (bottom panel). Bcd binding to 1079 peaks was abolished in zld⁻, as shown in A. (B) Six-hundred-forty other Bcd-binding peaks were called in both wild type and zld⁻, although peak heights may differ in the two genetic backgrounds. (C) Finally, 279 peaks were called by MACS only in the zld mutant samples. Read densities were calculated on tiling windows of 10 bp and expressed as the normalized number per million reads sequenced. To compare enrichment levels between wild type and zld⁻, wild-type read counts were scaled down 0.4474-fold to match the scale of zld⁻. The University of California at Santa Cruz (UCSC) gene annotation is shown at the bottom of each panel. The red boxes indicate the positions of enhancers. Genomic loci coordinates are shown next to the X-axis. (D–F) Distributions of Zld-binding sites in three groups of Bcd-binding peaks—wild-type-only peaks (D), shared peaks (E), and zld⁻-only peaks (F)—in a 500-bp window around peak summits. (G–I) Correlations between enrichment ratios (wild type/zld⁻) and number of Zld sites in wild-type-only peaks (G), shared peaks (H), and zld⁻-only peaks (I) in a 500-bp window around peak summits. Red lines represent medians. The bottom and top lines of the blue boxes represent the first and third quartiles, respectively. Error bars represent standard errors. Red crosses represent outliers that are beyond 1.5 times the interquartile range. (r) Pearson's correlation coefficient. P-values were calculated using a permutation test. Overlapping peaks of two biological replicates were used in D–I.

Figure 5.

Zld strengthens target gene responses to the Bcd gradient. (A) Correlation between Bcd-binding changes and mRNA expression changes in zld mutants. Each Bcd-dependent enhancer known to associate with a specific target gene is represented as a circle, and the different colors represent different numbers of Zld sites per kilobase. When multiple enhancers were associated with one gene, the average of Bcd-binding enrichment ratios of all enhancers was used. One-hour to 2-h mRNA expression data were obtained from Liang et al. (2008). (r) Pearson's correlation coefficient. The P-value was calculated using a permutation test. (B) Change of Bcd binding to type 0, type I, and type II Bcd-dependent enhancers in wild-type (wt) and zld⁻. Red lines represent medians. The bottom and top lines of the blue boxes represent the first and third quartiles, respectively. Error bars represent standard errors. Red crosses represent outliers that are beyond 1.5 times the interquartile range. (C) Correlations between Bcd-binding strength and posterior boundary positions (PBPs) of 66 enhancers. (D) Correlation between Zld-binding strength and posterior boundary positions of 66 enhancers. Zld-binding peak heights were obtained from Harrison et al. (2011). Zld-binding peak height was assigned 0 if a given enhancer was not associated with any peak. (r) Pearson's correlation coefficient. The P-value was calculated using a permutation test.

Figure 6.

Redistributed Bcd binding and precocious activation of target genes in zld⁻ embryos. (A) Relative levels of Bcd binding to basal promoter regions increase in zld⁻ compared with wild-type (wt) embryos. The percentages of peaks located within certain distances from peak summits to the nearest basal promoters are shown for different peak groups. TSSs were determined by looking for 5′ untranslated region (UTR) annotation from modENCODE submission: annotation of the developmental transcriptome of Drosophila melanogaster (ID: modENCODE_4057). P-values were calculated using a χ² test. (B) Ectopic Bcd binding correlates with up-regulated transcription. Numbers in the second column represent peaks that are associated with up-regulated genes. To identify genes associated with Bcd-binding peaks, we located coding sequences within 1 kb upstream of and downstream from the peak summit and filtered out genes with the TSS outside this region. Up-regulated genes were identified by comparing 1- to 2-h wild-type and zld⁻ microarray data from Liang et al. (2008). P-values were calculated by χ² test between the zld-only peaks and wild-type-only or shared peaks. (C–J) Expression of endogenous genes in wild-type (C,E,G) and zld⁻ (D,F,H) embryos. (I,J) lacZ expression of a dve enhancer in wild-type (I) and zld⁻ (J) embryos. (K) Bcd ChIP-qPCR of three enhancer regions. Error bars represent standard deviations from the mean. (*) P < 0.05; (**) P < 0.005.