Analysis of homeodomain specificities allows the family-wide prediction of preferred recognition sites

Abstract

We describe the comprehensive characterization of homeodomain DNA-binding specificities from a metazoan genome. The analysis of all 84 independent homeodomains from D. melanogaster reveals the breadth of DNA sequences that can be specified by this recognition motif. The majority of these factors can be organized into 11 different specificity groups, where the preferred recognition sequence between these groups can differ at up to four of the six core recognition positions. Analysis of the recognition motifs within these groups led to a catalog of common specificity determinants that may cooperate or compete to define the binding site preference. With these recognition principles, a homeodomain can be reengineered to create factors where its specificity is altered at the majority of recognition positions. This resource also allows prediction of homeodomain specificities from other organisms, which is demonstrated by the prediction and analysis of human homeodomain specificities.

Figure 1

DNA recognition by the homeodomain family. A) The structure of Msx-1 bound to DNA is representative of homeodomain-DNA interactions (Hovde et al., 2001). B) Detailed view of the recognition contacts (red), where residues at positions 2 and 5 of the N-terminal arm (orange) interact with bases in the minor groove and residues at positions 47, 50, 51 and 54 of the recognition helix (yellow) are positioned to make contacts in the major groove. C) (Top) Sequence logo representation of the diversity in our set of 84 homeodomains. (Bottom) Windows highlighting the diversity in the DNA-recognition regions - the N-terminal arm (red) and recognition helix (yellow). The key recognition positions are indicated with asterisks. D) Cartoon depicting recruitment of omega-Zif12-HD (homeodomain) fusions to the weak promoter driving the HIS3 and URA3 reporters used in the B1H system (Meng et al., 2005; Meng and Wolfe, 2006).

Figure 2

Clustering of the 84 Drosophila homeodomains. (A) Clustering based on the similarity between the recognition motifs of these factors, which we have organized into eleven different specificity groups. (B) The typical and atypical homeodomains are distributed into separate groups. The average specificity of each group is indicated under the Group recognition motif, and to the right is the Sequence logo of the key recognition positions. (C) The specificity groups (colored rectangles) are mapped onto the homeodomain amino acid sequence similarity tree. In instances where neighbors have been assigned to different specificity groups (indicated by red brackets) any difference in residue type at a key recognition position (5, 47, 50, 54 or 55) is noted (ND = No difference).

Figure 3

Atypical homeodomain specificity and correlations with positions 54 and 55. A) (Left) Sequence logos for types of atypical homeodomains (either groups or outliers). (Right) The corresponding amino acid sequences at the key DNA contact positions. Arg at position 54 (magenta) correlates with a preference for Cyt at binding site position 4. Arg at position 55 (cyan) correlates with a preference for Gua at binding site position 2. Notable exceptions are indicated by red circles. B) Structural model of DNA recognition for atypical family members constructed from a superposition of the contacts observed in the MATα2-DNA (Wolberger et al., 1991) and Exd-Ubx-DNA structures (Passner et al., 1999). The arginines potentially specify the contacted Gua and the 5′ Thy due to the favorable van der Waals interaction (~4 Å) with the T-methyl group (silver sphere).

Figure 4

The role of position 8 in organizing the N-terminal arm. A) A large hydrophobic residue at position 8 docks into a pocket formed by the three-helix bundle of the homeodomain fold anchoring the N-terminal arm over the minor groove. B) Surface rendering of the homeodomain (residues 9–60, recognition helix shown in yellow; Msx-1 structure (Hovde et al., 2001)). Phe8 (red) sits in a structural pocket. C) Iroquois family members contain Ala at position 8, allowing the N-terminal arm to sample other conformations that reduce the specificity of the factor. D) Reintroduction of the Phe at position 8 in Caup (A8F) dramatically alters the specificity of the protein at positions 1 and 2 of the binding site.

Figure 5

Catalog of common specificity determinants for Asn51-containing homeodomains. Amino acid positions that are most likely to influence the sequence preference at a particular position are indicated in boxes (solid line – major groove, dotted line – minor groove) surrounding the core 6 bp binding element. An arrow points from the box of potential interactions to the base within each base pair that it describes. For simplicity some interactions, such as Lys50 with binding site positions 5 and 6, are described as influencing specificity on the primary strand of the DNA when in reality direct contacts are made to the complementary strand. DNA recognition by residues in the N-terminal arm is also dependent on the type of residue at position 8 as observed for the Iroquois group.

Figure 6

Exploring DNA-binding specificity through mutagenesis. A) Mutational analysis of binding site position 4 in Bcd. Three different mutants (I47N, K50A and I47N with K50A) were characterized to determine the alteration in base preference at this position. The frequency that each base was recovered at position 4 is indicated to the right of the Sequence logo for each factor. B) Conversion of Engrailed (En) into a homeodomain with TGIF-like specificity. (Top) Schematic representation of the critical base contacts responsible for specificity in En and TGIF family members. (Bottom) Flow diagram of the mutations required to complete the specificity conversion. Two intermediate specificity conversions (En^V1 and En^V2) were obtained first, and these mutations were combined along with Q50A to produce TGIF-like specificity.

Figure 7

Comparison of the predicted and determined recognition motifs for 6 human homeodomains. The specificities of the human factors were determined using the B1H system. In each case the “Determined” compares favorably with the “Predicted” motif generated using our algorithm. The p-value for each comparison was calculated from the weight matrices for each motif as described in the Methods with additional metrics of these comparisons in Supplementary Table 9. Of particular note, the specificity of Six3 is consistent with other Six family members; it does not specify TAAT as previously described (Zhu et al., 2002).