Dissecting the functional specificities of two Hox proteins

Abstract

Hox proteins frequently select and regulate their specific target genes with the help of cofactors like Extradenticle (Exd) and Homothorax (Hth). For the Drosophila Hox protein Sex combs reduced (Scr), Exd has been shown to position a normally unstructured portion of Scr so that two basic amino acid side chains can insert into the minor groove of an Scr-specific DNA-binding site. Here we provide evidence that another Drosophila Hox protein, Deformed (Dfd), uses a very similar mechanism to achieve specificity in vivo, thus generalizing this mechanism. Furthermore, we show that subtle differences in the way Dfd and Scr recognize their specific binding sites, in conjunction with non-DNA-binding domains, influence whether the target gene is transcriptionally activated or repressed. These results suggest that the interaction between these DNA-binding proteins and the DNA-binding site determines the architecture of the Hox-cofactor-DNA ternary complex, which in turn determines whether the complex recruits coactivators or corepressors.

Figure 1.

Analysis of modC from Dfd. (A) Schematic of 2.7-kb EAE-lacZ and 570-bp modC-lacZ. Sequence of Dfd-, Exd-, and Hth-binding sites in modC are color-coded. Mutant versions of modC are shown below the wild-type sequence. The Dfd–Exd composite binding site (site I) and nearby Hth-binding site is labeled “I.” Other combinations of Dfd-, Exd-, and Hth-binding sites are indicated as II, III, and IV. Note that Exd-binding sites often overlap with Hth-binding sites. (B,C) Epidermal expression of EAE-lacZ (B) and modC-lacZ (C) in maxillary and mandibular segments of the embryo. In these and all subsequent images, costaining for Scr in the right panels is used to identify the adjacent Dfd-expressing segments. Anterior is always to the left, and all embryos are at stage 13. (D,E) Mutagenesis of Dfd–Exd (D) or Exd half-site (E) of site I in modC results in a modest decrease in reporter gene activity (DE-lacZ and E-lacZ, respectively). In these and subsequent images, the yellow arrowheads point to the maxillary segments. (F) Mutagenesis of four Dfd-binding sites in the modC enhancer significantly decreases enhancer activity (4D-lacZ). (G) Mutagenesis of the Dfd–Exd site from site I and four Dfd-binding sites abolishes enhancer activity (DE-4D-lacZ). (H) Mutagenesis of the Exd half-site from site I and four Dfd-binding sites abolishes enhancer activity (E4D-lacZ). (I) modC-lacZ is induced in response to ectopic Dfd by AG11-Gal4.

Figure 2.

Dfd-specific functions depend on Exd–Hth. (A,B) modC-lacZ expression is normal in the maxillary segment (yellow arrowheads) of hth^P2 heterozygotes (A; visualized by hb-lacZ marked balancer), but is abolished in hth^P2 homozygotes (B). The ectopic expression driven by modC in hth^P2 embryos suggests that Hth may have a repressive role in thoracic and abdominal segments. (C) modC-lacZ requires direct Exd–Hth input. Mutagenesis of three Exd–Hth-binding sites in modC (3EH-lacZ) (Fig. 1A) results in a large decrease in enhancer activity. Note that Dfd monomer binding to this element is weakly affected by this mutation. (D) Hth binds to modC in vivo. Anti-Hth antibody was used to precipitate embryonic chromatin. Immunoprecipitated chromatin was assayed for the presence of modC (spanning site I) and a negative control locus, PDH. Data are presented as a percentage of the signal obtained relative to input chromatin and represent the averages and standard deviation of two independent ChIPs. (E, lanes 7,8) Dfd^WT binds cooperatively to site I with Exd–HM. (Lanes 4,5) Dfd^YPAA fails to bind cooperatively with Exd–HM to site I. (Lanes 12,13) Dfd^WT fails to bind cooperatively with Exd–HM to site I that has a mutant Exd half-site (site I^exd-mut). Hox monomer and Hox–Exd–HM trimer complexes are indicated by the pink and black arrowheads, respectively. (F,G) Ectopic expression of Dfd^WT (F) results in ectopic activation of EAE-lacZ, while ectopic expression of Dfd^YPAA (G) does not (see Fig. 3F for quantification). (H) Head and thorax of a wild-type embryo showing the normal position of cirri in the head segment (indicated by pink arrowhead). Ubiquitous expression of Dfd^WT (H′) results in ectopic cirri (yellow arrowheads, see inset in panel) and mouth hooks in thoracic segments (see Fig. 3G for quantification), but expression of Dfd^YPAA failed to induce any Dfd-specific structures (H″).

Figure 3.

Dfd specificity module residues His-15 and Arg3 are required in vitro and in vivo for Dfd-specific functions. (A) Schematic showing Hox specificity module, homeodomain, YPWM motif, and linker. Below are alignments of specificity modules from Dfd orthologs, showing the conserved histidine and an arginine (underlined) at positions analogous to those in Scr's specificity module (Joshi et al. 2007). (B) Dfd^{His-15A, Arg3A} binding to site I is weaker both as a monomer (lane 6) and as cooperative trimer (lanes 7,8) with Exd–HM in comparison with Dfd^WT (lanes 3–5). Hox monomer and Hox–Exd–HM trimer complexes are indicated by pink and black arrowheads, respectively. (C) Kd measurements for Dfd^WT–Exd–HM and Dfd^{His-15A, Arg3A}–Exd–HM to site I show an eightfold difference in binding affinity. (D,E) Ubiquitous expression of Dfd^WT results in ectopic activation of EAE-lacZ (D), while Dfd^{His-15A, Arg3A} (E) is severely compromised in its ability to induce this reporter gene (see F for quantification). (F) Graph showing percentage of total embryos that activate EAE-lacZ ectopically when Dfd^WT, Dfd^{His-15A, Arg3A}, or Dfd^YPAA is expressed ubiquitously using AG11-Gal4 (n ≥ 56 for each genotype, n = 2). (G) Graph showing the quantification of ectopic cirri made in thoracic segments of embryonic cuticles when Dfd^WT or Dfd^{His-15A, Arg3A} is expressed ubiquitously using AG11-Gal4. Error bars represent standard deviations.

Figure 4.

Dfd represses fkh250-lacZ in an Exd-independent manner. (A) Wild-type expression patterns of fkh250-lacZ and prd-GAL4 (driving GFP) to show overlap between these two expression domains. The strongest expression of fkh250-lacZ is in PS2. (B) Ectopic expression of Scr^WT activates fkh250-lacZ (yellow arrowheads). (C) Ectopic expression of Dfd^WT represses fkh250-lacZ (red arrowhead). (D) Ectopic expression of Dfd^{His-15A, Arg3A} represses fkh250-lacZ (red arrowhead), showing that repression of fkh250-lacZ by Dfd is independent of these residues. (E) Ectopic expression of Dfd^AAAA represses fkh250-lacZ (red arrowhead), suggesting that repression of fkh250-lacZ by Dfd does not require recruitment of Exd by its YPWM motif. (F) Coexpression of both Dfd^WT and Scr^WT represses fkh250-lacZ in PS2 and largely blocks its activation in posterior segments (white arrowheads), showing that fkh250-lacZ repression by Dfd^WT is direct. (G) Coexpression of both Dfd^AAAA and Scr^WT represses fkh250-lacZ in PS2 and largely blocks its activation in posterior segments (white arrowheads), showing that repression of fkh250-lacZ by Dfd is direct and unlikely to require Exd recruitment.

Figure 5.

Enhancing Exd recruitment by Dfd to fkh250 is not sufficient for activation. (A) Schematics of Dfd^ScrSM and Dfd^NScrSM chimeras. Extent of specificity modules and homeodomains are indicated (for details, see the text). (B) Dfd^WT binds less cooperatively with Exd–HM to fkh250 (lanes 4–6) compared with Scr^WT (lanes 8–10). (Lanes 12–14) Dfd^ScrSM binds very cooperatively with Exd–HM to fkh250. (Lanes 16–18) Dfd^ScrSM binds very cooperatively with Exd–HM on site I. Hox monomer and Hox–Exd–HM trimer complexes are indicated by pink and black arrowheads, respectively. (C,D) Ectopic expression of Dfd^ScrSM represses fkh250-lacZ (C; red arrowhead) and activates EAE-lacZ (D), suggesting that an enhancement of Exd recruitment is not sufficient for fkh250-lacZ activation.

Figure 6.

Changing the functional specificity of Dfd. (A) Wild-type expression pattern of fkh250-lacZ. (B,C) Ectopic expression of Dfd^NScrSM (B) or Dfd^ScrSMΔ23 (C) activates fkh250-lacZ (yellow arrowheads), showing a change in specificity. (D,E) Ectopic expression of DfdΔ23 (D) or Dfd^ScrSMΔ23^AAAA (E) represses fkh250-lacZ in PS2 (red arrowhead), emphasizing the importance of Scr's SM- and Exd-mediated cooperative binding in activation of fkh250-lacZ. (F–H) Ectopic expression of Dfd^NScrSM (F), Dfd^ScrSMΔ23 (G), or DfdΔ23 (H) activates EAE-lacZ. These proteins retain the ability to activate Dfd, like Dfd^WT. (I) Ectopic expression of Dfd^ScrSMΔ23^AAAA does not activate EAE-lacZ, emphasizing the importance of Exd recruitment in EAE-lacZ activation. (J–O) Cuticle phenotypes resulting from ubiquitous expression of Dfd^WT (J), Scr^WT (K), Dfd^NScrSM (L), Dfd^ScrSMΔ23 (M), DfdΔ23 (N), and Dfd^ScrSMΔ23^AAAA (O). Scr-specific structures (ectopic T1 beard) are indicated with green arrowheads, and Dfd-specific structures (cirri) are indicated by white arrowheads. Dark-field and phase-contrast images for the T2 segment are shown for all cuticles.

Figure 7.

Interplay between DNA-binding specificity and transcriptional output. (A) Specific in vivo functions of Scr and Dfd. Green arrows indicate specific functions, the repression bar is indicated in red, and the black line indicates no effect. (B–F) Schematized is a subset of the various regulatory scenarios examined in this study. Activation is indicated by a green arrow, and the repression bar is indicated in red. A comparison between Dfd^ScrSM–Exd bound to fkh250 (C), which represses transcription, and Dfd^ScrSMΔ23–Exd bound to fkh250 (E), which activates transcription, suggests that motifs 2 and 3 are required to either recruit a corepressor or block the recruitment of a coactivator. A comparison between DfdΔ23–Exd bound to fkh250 (D), which represses transcription, and Dfd^ScrSMΔ23–Exd bound to fkh250 (E), which activates transcription, suggests that differences in the specificity module, and thus how these proteins bind to fkh250, determines the sign of the regulation. Finally, a comparison between DfdΔ23–Exd bound to fkh250 (D), which represses transcription, and DfdΔ23–Exd bound to EAE (F), which activates transcription, suggests that differences between binding sites also contribute to differences in regulatory output.