Diversification of transcription factor paralogs via noncanonical modularity in C2H2 zinc finger DNA binding

Abstract

A major challenge in obtaining a full molecular description of evolutionary adaptation is to characterize how transcription factor (TF) DNA-binding specificity can change. To identify mechanisms of TF diversification, we performed detailed comparisons of yeast C2H2 ZF proteins with identical canonical recognition residues that are expected to bind the same DNA sequences. Unexpectedly, we found that ZF proteins can adapt to recognize new binding sites in a modular fashion whereby binding to common core sites remains unaffected. We identified two distinct mechanisms, conserved across multiple Ascomycota species, by which this molecular adaptation occurred. Our results suggest a route for TF evolution that alleviates negative pleiotropic effects by modularly gaining new binding sites. These findings expand our current understanding of ZF DNA binding and provide evidence for paralogous ZFs utilizing alternate modes of DNA binding to recognize unique sets of noncanonical binding sites.

Figure 1. Common and TF-preferred DNA-binding specificities of Msn2-family members

(A) Hierarchical clustering of pairwise binding profile comparisons for the Msn2-specificity-group proteins. Comparisons were performed for published datasets ((Badis et al., 2008) (Zhu et al., 2009)) and duplicate PBMs from this study. (B) Comparison of Msn2 and Com2 binding to all possible (32,896) 8-bp sequences. Z-scores are transformed 8-mer median signal intensities (see Experimental Procedures). TF-preferred sites bound preferentially by Msn2 or Com2 are in magenta or orange, respectively. Common sites bound significantly (PBM E-score > 0.48) by both proteins are highlighted. 8-bp sequences (labeled A, B, C and D) assayed for binding in EMSAs are in red. (C) Comparison for Msn2 and Usv1 (details as in B, except that Usv1-preferred sites are in orange). (D) Binding schematic for Msn2-family proteins and binding site motifs for TF-preferred and common sites. Canonical recognition residues are indicated (grey dots). Proposed interaction map for canonical residues and DNA bases of STRE is shown (see also Figure S1). (E) Dissociation binding constants (mean and standard deviation) for Com2, Usv1 and Msn2 to selected DNA sequences are listed. (F) Protein sequence alignment for Msn2 specificity group proteins. Highlighted are conserved residues (*), canonical ZF DNA-contacting residues (grey bars, canonical numbering scheme), Com2 N-terminal RGRK motif (bold), and ZF secondary structure elements: beta-strands (empty arrows) and alpha-helices (empty boxes).

Figure 2. Binding specificities of ZF mutants

(A), (B) Binding profile comparisons (as in Figure 1) for Com2 and Com2 RK→EE mutant relative to Msn2. Com2-preferred sites (as in Figure 1B) are highlighted (orange). 8-bp sequences (labeled A, B, C and D) assayed in EMSAs are in red. (C) Comparison of Com2 and Com2 RK→EE mutant binding to select binding sites. DNA base differences from the common site TCAGGGG (Figure 1D) are in red. Scores are mean z-scores for the eight different 8-mers containing the 7-mer sites shown (error bars = 1 standard deviation (SD)). (**) P < 10⁻³, (***) P < 10⁻⁴, unpaired Student’s t-test (D), (E) Binding profile comparisons for Usv1 two-ZF and 4-Res mutants shown relative to Usv1. Usv1-preferred sites (as in Figure 1D) are highlighted (orange); the subset of sites preferentially bound by Usv1 relative to Usv1 4-Res mutant are highlighted (magenta). Approximate positions of canonical recognition residues (grey dots) and mutated residues (red dots) are illustrated in ZF cartoons. (F) Comparison of Usv1 and Usv1 4-Res mutant binding to two binding sequences: (i) TCAGGGG common site (Figure 1D), and (ii) AGGTAC – a Usv1-preferred site (Figure 1D) that was bound poorly by Usv1 4-Res mutant. Scores are mean z-scores for the eight different 8-mers containing TCAGGGG (columns 1 and 3) and for the 48 different 8-mers containing AGGTAC (columns 2 and 4) (error bars = 1 SD). (****) P < 10⁻¹⁵, unpaired t-test. (G)–(J) Binding specificities of Msn2 wild-type and mutant proteins. Binding profile comparison for Msn2 mutants relative to Msn2. Usv1-preferred sites (as in Figure1D) are highlighted (orange). (K) Dissociation binding constants (mean and standard deviation) for Com2 (as in Figure 1E, for comparison) and Com2 RK→EE mutant to select DNA sequences are listed. Probe sequences are as in Figure 1E.

Figure 3. Binding specificity for Adr1 specificity group proteins

(A)–(B) Pairwise binding profile comparisons for Adr1 and Ypr022C relative to Ygr067c. Z-scores are as in Figure 1B,C. Common and TF-preferred sites are highlighted. Binding motifs are shown for highlighted TF-preferred and common sites (base numbering as in Figure 1D, with base congruence defined by ZF binding schematic (see Figure S1)).

Figure 4. Models for TF-specific binding modes

Binding schematics depict key features of the canonical ZF binding mode and the TF-specific binding modes that facilitate the binding to the Msn2-family common and TF-preferred sequences. Models are presented for (A) Com2 and (B) Usv1 (same model applies for Rgm1). Key features proposed for the TF-specific binding modes are highlighted. Residues mutated in Msn2 and Usv1 constructs (Figure 2) that affected DNA binding are highlighted (red dots).