Cell-type-specific Hox regulatory strategies orchestrate tissue identity

Abstract

Hox proteins are homeodomain transcription factors that diversify serially homologous segments along the animal body axis, as revealed by the classic bithorax phenotype of Drosophila melanogaster, in which mutations in Ultrabithorax (Ubx) transform the third thoracic segment into the likeness of the second thoracic segment. To specify segment identity, we show that Ubx both increases and decreases chromatin accessibility, coinciding with its dual role as both an activator and repressor of transcription. However, the choice of transcriptional activity executed by Ubx is spatially regulated and depends on the availability of cofactors, with Ubx acting as a repressor in some populations and as an activator in others. Ubx-mediated changes to chromatin accessibility positively and negatively affect the binding of Scalloped (Sd), a transcription factor that is required for appendage development in both segments. These findings illustrate how a single Hox protein can modify complex gene regulatory networks to transform the identity of an entire tissue.

Keywords: Drosophila melanogaster; Hox cofactors; Hox genes; Ubx; Ultrabithorax; appendage development; bithorax; haltere; serial homology; wing.

Figure 1.. Segment specific chromatin accessibility and gene expression in wing and haltere imaginal discs

(A) Schematics of an adult fly highlighting the contributions of the dorsal wing and haltere imaginal discs; the lower panel shows a magnified view of the proximal appendage (hinge) regions. For both the wing-bearing T2 and haltere-bearing T3 segments, blue marks body wall domains (notum (N) and postnotum (PN), respectively) and red marks the appendages (wing and haltere, respectively). The tsh+ domain (blue) gives rise to the body and proximal hinge, while the nub+ domain (red) gives rise to the distal hinge and appendage proper (wing blade and capitellum). The Hox cofactor Hth (yellow), which induces the nuclear localization of Exd (Exd^nuc), is expressed in the body wall, proximal hinge, and distal hinge, but is absent from the appendage proper. (B) Left, immunostain of 3^rd larval instar wing (W) and haltere (H) imaginal discs showing distal (nub+, red) and proximal (tsh+, blue) populations. Also shown is the T3 leg imaginal disc (L). Right, Ubx is expressed throughout the haltere disc, and is absent from the wing disc. Scale bars for this and subsequent panels are 50 μm. (C) Experimental scheme to compare chromatin accessibility using ATAC-seq in homologous distal (nub+, red) and proximal (tsh+, blue) populations of the wing and haltere imaginal discs. Dotted background indicates the presence of Ubx in all haltere imaginal disc cells. (D-F) Genome-wide comparison of wing and haltere ATAC-seq data for whole tissue (D) nub+ cells (E) and tsh+ cells (F). Colored points satisfy a threshold of log2FC>0.5, padj<0.05. Diamond shaped points are ATAC peaks within the Ubx genomic locus. A common set of 24,915 open chromatin regions, generated by merging ATAC-seq peaks in each sorted data set, was used for comparisons. (G-H) ATAC-seq genomic tracks at previously described Ubx target CRMs sal1.1 (G) and knW (H). Cloned fragments driving reporter expression (green) above the genome tracks are indicated by the green bar. (I) Comparison of ATAC-seq scores with transcriptome measurements from sorted nub+ and tsh+ cells. Differentially expressed genes (DESeq padj<0.01) are significantly more likely to have a differential ATAC peak (DESeq −log₁₀pval) compared to genes expressed at similar levels. (See STAR Methods for details.) Median value indicated by horizontal line.

Figure 2.. Ubx regulates chromatin accessibility

(A) Expression of the Ubx target Salm in wildtype and following Ubx knockdown. De-repression of Salm in the haltere pouch is observed (arrow). Loss of Ubx expression (left) and de-repression of Spalt are magnified in insets (yellow box). (B) Genomic tracks showing the salm locus. The sal1.1 CRM is marked by the grey box. The region corresponding to ATAC peak generated by MACS2 and compared using DESeq2 is indicated by a dashed box. For each comparison (WT nub[W vs H] and ubx.RNAi nub[W vs H]) the log2 fold change is indicated as the top number and adjusted p-value in parenthesis. (C) Volcano plot comparing nub+ chromatin accessibility in wing and haltere imaginal discs following knockdown of Ubx. Inset; the same comparison in wild type discs is repeated from Figure 1e for comparison. Note that the loci within the genomic region of Ubx (diamond-shapes) remain differentially accessible, as expected given regulation of Ubx expression is upstream of Ubx activity. (D) de novo motif analysis of the four differential ATAC-seq categories defined in Figure 1D-F. The top ranked motif for each category is shown. Candidate Ubx and Ubx-Hth-Exd motifs resemble motifs derived from SELEX-seq assays (see Figure S2C-D). Heatmaps on the left show the wing and haltere ATAC-seq signals for each of the four categories. (E) Heatmap showing the haltere ChIP signal for Ubx and Hth at loci within the differential ATAC-seq categories. Regions are centered around the closest match to the top-ranked de novo motif for that category (panel D) and sorted by highest-to-lowest scoring match to that motif. (F) Plots showing distribution of average ChIP-signal centered around the same motif as panel E. Each category is split into thirds based on the degree of similarity of motif matching to the top ranked de novo motif for that category. See STAR Methods for details.

Figure 3.. Analysis of Ubx-targeted CRMs

(A) Position of homologous distal hinge and pouch domains based on Hth and Nub expression in wing and haltere discs. The edges of Hth and Nub expression domains are marked with dotted yellow and red lines, respectively. (B) Summary of CRM-reporters. The nub[Hsal1.1, knW, and ana-spot^,,. (C-F) Examples of nub[HW] CRM-reporter genes (green). The upper left panels show genomic tracks for nub+ ATAC-seq wing, nub+ ATAC-seq haltere, Ubx ChIP, and Hth ChIP (Log2Fold difference between wing and haltere indicated next to dashed line); the upper right panels show wing and haltere disc expression patterns for the reporter genes, and the bottom panels show Ubx null somatic clones in the haltere, with a subset of clones magnified in the insets. Clones are marked by the absence of RFP (arrows). ID for reporter examples: C: Rep-3 (left) and Rep-2 (right); D: Rep-18 (left) Rep-17 (right); E: Rep-5 (left) and Rep-6 (right); F: Rep-14 (left) and Rep-12 (right). See Figure S3 for additional examples and Table S1 for a list of all reporter genes.

Figure 4.. Ubx-mediated changes to chromatin accessibility changes where Sd binds.

(A) Homologous patterns of Sd expression in the wing and haltere imaginal discs. In both tissues Sd is expressed in the pouch, in the distal hinge and along the dorsal-ventral compartment boundary. Boundaries of Nub (Red) and Hth (Yellow) expression are indicated with dotted lines as in Figure 3. (B) Volcano plot comparing Sd binding in wing and haltere imaginal discs (Diffbind FDR<0.05). (C) Genomic tracks near the Samuel CRM (green box) and reporter expression driven by this CRM in wing and haltere discs. (D) de novo motif analysis of the disc-specific Sd binding peaks for the Sd H<W and H>W categories. (E) Heatmaps showing the ChIP signal for differential Sd binding, nub+ ATAC-seq signal, Ubx ChIP signal, and Hth ChIP signal. Regions are sorted based on highest-to-lowest W:H ratio of Distal ATAC-seq signal at the peak center. The top set shows the Sd H<W category and the bottom set shows the Sd H>W category as defined in panel B. (F) Summary defining the three domains in T2 and T3, whether Ubx acts as a monomer or Ubx-Hth-Exd complex, whether Ubx opens or closes chromatin, and the effect on Sd binding.