Fig. 5

DBF dictates the binding vocabulary of C2H2-ZFs. a Contribution of both base-contacting residues (highlighted) and non-base-contacting residues to the DBF model. The letter height represents the average value that each amino acid at each position contributes to the DBF score. b Evaluation of non-base-contacting residue DBF sub-model. The x-axis represents 228 pairs of DNA-binding/non-binding ZFs with identical base-contacting residues [16]. The histogram of ΔDBF for 228 pairs of DNA-binding and non-binding C2H2-ZFs with identical base-contacting residues (the non-base-contacting DBF sub-model score of the DNA-binding C2H2-ZF minus that of the non-binding C2H2-ZF). The ZF pairs shown here were excluded from the training set for the full DBF model, which was then deconvoluted to obtain the sub-model that was used for evaluation of performance. c Schematic illustration of in silico recombination experiments. The base-contacting residues of a natural C2H2-ZF are swapped with those of another C2H2-ZF and both the original and recombinant C2H2-ZFs, in complex with their respective optimal binding sites, are evaluated by structural modeling. The optimal binding site of each C2H2-ZF is determined by PBM [16]. Modeling results are used to reveal the structural basis of the contribution of base-contacting and non-base-contacting residues to the DBF model, based on comparisons showed by red and green arrows, respectively. d Scatter plot showing the change in the base-contacting DBF score after mutation of base-contacting residues vs. the change in the number of hydrogen bonds with DNA bases. Both original and mutated base-contacting residues are paired with their respective optimal binding sites for structural modeling. e Scatter plot showing the change in the non-base-contacting DBF score vs. the change in the number of hydrogen bonds with DNA phosphate backbone. f Tukey boxplot of DBF scores for base-contacting residue combinations with that are present/absent in fungi C2H2-ZFs (left) or in metazoan C2H2-ZFs (right). The P value corresponds to two-sided Mann–Whitney U test. g Tukey boxplot of DBF scores for base-contacting residue combinations that bind to each DNA triplet. The greatest DBF score that is within 1.5× the interquartile range of the upper quartile (the upper whisker) is considered the obtainable DBF for each triplet. h Obtainable DBF correlates with frequency of C2H2-ZFs that recognize each triplet in fungi. The Pearson correlation of logarithm of count vs. obtainable DBF score is shown. i Heatmap showing enrichment of metazoan C2H2-ZFs among those with low base-contacting DBF score and high non-base-contacting DBF scores. The distribution of C2H2-ZFs as a function of base-contacting and non-base-contacting DBF scores was estimated for metazoans and non-metazoans separately – the color gradient represents the logarithm of ratio of the probability density function for metazoan C2H2-ZFs relative to non-metazoan C2H2-ZFs