Zelda potentiates morphogen activity by increasing chromatin accessibility

Abstract

Zygotic genome activation (ZGA) is a major genome programming event whereby the cells of the embryo begin to adopt specified fates. Experiments in Drosophila and zebrafish have revealed that ZGA depends on transcription factors that provide large-scale control of gene expression by direct and specific binding to gene regulatory sequences. Zelda (Zld) plays such a role in the Drosophila embryo, where it has been shown to control the action of patterning signals; however, the mechanisms underlying this effect remain largely unclear. A recent model proposed that Zld binding sites act as quantitative regulators of the spatiotemporal expression of genes activated by Dorsal (Dl), the morphogen that patterns the dorsoventral axis. Here we tested this model experimentally, using enhancers of brinker (brk) and short gastrulation (sog), both of which are directly activated by Dl, but at different concentration thresholds. In agreement with the model, we show that there is a clear positive correlation between the number of Zld binding sites and the spatial domain of enhancer activity. Likewise, the timing of expression could be advanced or delayed. We present evidence that Zld facilitates binding of Dl to regulatory DNA, and that this is associated with increased chromatin accessibility. Importantly, the change in chromatin accessibility is strongly correlated with the change in Zld binding, but not Dl. We propose that the ability of genome activators to facilitate readout of transcriptional input is key to widespread transcriptional induction during ZGA.

Figure 1. The number of Zld binding sites determines the spatial extent of Dl target gene expression

Wild-type (A, C, E, G, H, I, K, M, O, P) and zld⁻ (B, D, F, J, L, N) embryos in nc 14 were hybridized with RNA probes synthesized against cDNA sequences for sog (A, B), brk (I, J) or lacZ (C–H, K–P) for transgenic embryos. Embryos are oriented anterior to the left and dorsal up here and in subsequent figures. A schematic representation of the enhancer that drives lacZ expression is shown below transgenic embryos (C–H, K–P). Green triangle = Dl site; Dark purple diamond = canonical Zld site; Light purple diamond = non-canonical Zld site; Red diamond = mutagenized Zld site. (C–F) Mutation of all three Zld sites in the sog NEE caused a reduction in the expression domain it drives. (G, H) Elimination of one (H) or two (G) Zld sites in sog NEE resulted in a step-wise narrowing of expression domain. (K–N) Addition of 3 Zld sites to the brk5′ enhancer led to a Zld -dependent expansion in expression. (O) Mutation of the 3 added Zld sites gave an expression similar to that driven by brk wt enhancer. (P) Removal of all Zld sites in brk5′ enhancer led to sporadic and thin expression pattern. Anterior-posterior modulation seems to be in play for the expression of brk, which could be explained by the presence of two Bcd sites in this enhancer [15]. (Q) Scatter plot showing the width of expression domain (in the number of cells it spans) driven by different forms of the sog NEE that contains 0 (0TAG), 1 (1TAG), 2 (2TAG) or 3 (3TAG) Zld sites. Each dot represents the average from at least 20 embryos. The width of expression domain correlates with number of Zld sites (linear regression R²=0.66). (R) Bar chart showing the width of expression domain driven by brk wt, brk +3a, brk +3m, sog 0 and sog wt enhancers. Data are represented as mean ± standard error of the mean (SEM) (*** means t-test p-value < 0.005). See also Figures S1–S3.

Figure 2. The number of Zld binding sites determines the timing of Dl target gene activation

nc 11 (A, C, E, G) and 12 (B, D, F, H) embryos carrying sog(A–D) or brk(E–H) transgenes were hybridized with RNA probes synthesized against intronic sequences of the yellow reporter gene. Dl antibody staining (not shown) was used to orient embryos. DAPI stained nuclei expressing yellow reporter gene are pseudo-colored in yellow. Compared to the sog wt enhancer (A, B), mutation of all Zld sites in the sog NEE (sog 0; C, D) results in delayed and sporadic expression. (E–H) Embryos carrying the brk enhancer with added Zld sites (brk +3a; G, H) have advanced initiation of transcription compared to embryos carrying the brk wt enhancer (brk wt; E, F). (I) Table showing the number of embryos carrying the four transgenic lines that display any expression from nc 9 to nc 12.

Figure 3. Zld promotes Dl binding to target enhancers and increases chromatin accessibility

Bar charts showing Zld (A, D, G) and Dl (B, E, H) ChIP-qPCR results and DNase I digestion-qPCR results (C, F, I) performed on 1.5–3 hr embryos carrying transgenic enhancers. (A, B, C) Embryos carrying the sog enhancer with mutated Zld sites (sog 0) has reduced Zld (A) and Dl (B) binding, as well as lower sensitivity to DNase I digestion (C) on the reporter region compared to embryos carrying the sog wt enhancer. (D, E, F) the brk transgenic enhancer with added Zld sites (brk +3a) has higher Zld (D) and Dl (E) binding, and higher sensitivity to DNase I digestion (F) than the brk wt enhancer. (G, H, I) the brk transgenic enhancer with mutated Dl sites (brk 0Dl) has reduced Zld (G) and Dl (H) binding, and slightly decreased sensitivity to DNase I digestion (I) than the brk wt enhancer. Shown are ChIP enrichment or DNase I hypersensitivity relative to an unrelated genomic region (see Experimental Procedures). Error bars indicate SEM from three biological replicates (*: p<0.05, **: p<0.01 and ***: p < 0.005).

Figure 4. The change in chromatin accessibility correlates with the change in Zld binding on target enhancers

Zld and Dl ChIP enrichment and DNase I hypersensitivity on the transgenic region were normalized to endogenous enhancer locus, then the fold change was calculated for the two lines under comparison (sog 0 vs. sog wt, brk +3a vs. brk wt, and brk 0Dl vs. brk wt). Blue dots and green dots represent Zld and Dl, respectively. The change in Zld binding between lines strongly correlates with the change in DNase I hypersensitivity (linear regression R²=0.98) while the change in Dl binding does not (R²=0.02).