Duplication of a promiscuous transcription factor drives the emergence of a new regulatory network

Abstract

The emergence of new genes throughout evolution requires rewiring and extension of regulatory networks. However, the molecular details of how the transcriptional regulation of new gene copies evolves remain largely unexplored. Here we show how duplication of a transcription factor gene allowed the emergence of two independent regulatory circuits. Interestingly, the ancestral transcription factor was promiscuous and could bind different motifs in its target promoters. After duplication, one paralogue evolved increased binding specificity so that it only binds one type of motif, whereas the other copy evolved a decreased activity so that it only activates promoters that contain multiple binding sites. Interestingly, only a few mutations in both the DNA-binding domains and in the promoter binding sites were required to gradually disentangle the two networks. These results reveal how duplication of a promiscuous transcription factor followed by concerted cis and trans mutations allows expansion of a regulatory network.

Figure 1. Maltose- and isomaltose-specific genes are differentially regulated.

Representative brightfield and fluorescence microscopy images of yeast cells with various MALS or MALT genes fluorescently tagged are shown for wt cells (a) and strains carrying deletions of genes encoding transcriptional regulators (b: MALX3 and c: YFL052W). Cells were grown in presence of either palatinose (α 1–6 disaccharide) or maltose (α 1–4 disaccharide) as indicated above the pictures. Gene names are listed in the first column, and protein activities towards the two types of sugars (maltose or palatinose) are indicated in the second and third columns. Scale bar is included in the upper left image and equals 10 μm. The experiment was repeated at least three times.

Figure 2. Different DNA-binding specificity of different MalR transcription factors.

(a) Sequence logo of Yfl052w DNA-binding site CGG(9N)CGG. (b) Sequence logo of Malx3 DNA-binding site CGC(9N)CGN. (c) Sequence Diversity Diagram conveying differences and similarities between Yfl052w (blue) and Malx3 (red) binding sites. Regions where two groups overlap are shown in purple. Thickness of the line shows the relative proportion of the aligned sequences. Positions that differentiate two groups are identified by a separation of blue and red lines. The rings below are basic heatmaps in circular layout showing the information content of the positions. The more saturated the colour, the higher the information content, thus indicating conserved regions. Mutual information (MI) represents covariance of positions and is shown as grey lines inside the circles. High MI indicates the dependency of one position on another.

Figure 3. Maltose- and palatinose-specific MalR regulators have different DNA-binding sites.

Representative flow cytometry histograms of populations of fluorescent reporter strains carrying different types of DNA-binding sites in the promoters of a maltose-specific gene MAL32 (red block arrow), or a palatinose-specific IMA5 (blue block arrow), grown in maltose or palatinose. CGG-containing binding sites found in promoters of palatinose-specific genes are depicted as blue rectangles, CGC-containing DNA-binding sites found in promoters of maltose-specific gene are depicted as red rectangles. (a) DNA-binding specificity of the palatinose-specific regulator Yfl052w. (1) A fluorescently tagged IMA5 gene under its native promoter shows a normal level of IMA5 expression activated by Yfl052w in palatinose. (2) Deletion of the Yfl052w binding site abolishes the expression of IMA5. (3) A fluorescently tagged maltose-specific MAL32 under its native promoter shows no expression on palatinose and is used to quantify autofluorescence. Introduction of a Yfl052w binding site from the promoter of IMA5 (4) or IMA1 (5) in front of the fluorescently tagged MAL32 gene makes the expression of this gene responsive to palatinose. (6) Deletion of YFL052W abolishes the expression of the MAL32 gene from (4). (b) DNA-binding specificity of the maltose-specific regulator Malx3. (1) A fluorescently tagged MAL32 under its native promoter shows a normal level of MAL32 expression activated by Malx3 in maltose. (2) Deletion of one Malx3 binding site decreases the expression levels of fluorescently labelled MAL32. (3) Deletion of all three Malx3 binding sites abolishes the expression of MAL32. (4) A fluorescently tagged IMA5 gene under its native promoter is used to show autofluorescence. Introduction of one (5), (6), two (7)–(9) or three (10) Malx3 binding sites in the promoter region of a fluorescently tagged IMA5 gene makes the expression of this gene responsive to maltose, with increasing number of binding sites resulting in higher expression levels. (11)–(13) Deletion of the maltose-specific regulator MALX3 abolishes the expression of strains from (7), (8) and (10). Each experiment was repeated at least three times with two biological replicates.

Figure 4. Yfl052w and Malx3 regulators show different DNA-binding specificity.

(a) Two point mutations in a maltose-inducible promoter yield a palatinose-inducible promoter. (1) Histogram of the fluorescence signal of a strain with a yECitrine-tagged MAL32 gene. This maltose-specific reporter gene shows no expression in palatinose and can be used to estimate the background fluorescence levels. (2) Single nucleotide C to G substitution in both CGC motifs in the upstream Malx3 binding site of the of MAL32 promoter leads to expression of this gene in palatinose. (3) Deletion of palatinose-specific regulator Yfl052w abolishes the expression of the mutant promoter. (b) Fluorescence microscopy images of the same strains reported in a. (c) Increasing the number of CGG-containing binding sites in a palatinose-inducible promoter yields a promoter that is responsive to maltose. (d) Representative fluorescence microscopy images of the strains shown in c in maltose. (e) Increasing the number of CGG-containing binding sites does not affect the expression of the palatinose-inducible gene IMA5 in palatinose. (f) Representative fluorescence microscopy images of the strains shown in c in palatinose. Scale bar is included in the upper left image of b,d,f and equals 10 μm. Each experiment was repeated at least three times with two biological replicates.

Figure 5. Differences in the DNA-binding domain of Malx3 and Yfl052w explain their different binding specificity.

(a) Alignment of the Malx3 and Yfl052w DNA-binding domains. Amino acids predicted to interact with the DNA-binding site are indicated with a black rectangle. The key position 12 that differs between Malx3 and Yfl052w is highlighted with a blue arrow. (b) Molecular modelling of the interaction between the Zn-finger domain and its DNA-binding site. Important base pairs are represented as yellow and magenta sticks, important amino acids are represented as blue sticks. The Arg15 is shared between both transcription factors and is responsible for the recognition of the G in the middle of the CGG binding motif. Arg12 in Malx3 does not take part in recognition of the CGG motif, but Cys12 in Yfl052w does interact with the DNA and is responsible for the preference for a G nucleotide in the third position of the motif. (c) A mutated version of the palatinose-specific Yfl052 activator (Cys12Arg and Ile13Val) is able to partly activate the MAL32 promoter in response to palatinose and also retains its capacity to activate the IMA5 promoter. (d) A mutated version of the maltose-specific Malx3 activator (Arg12Cys and Val13Ile) is incapable to activate MAL32 or IMA5 in response to maltose. (e) A mutated version of the maltose-specific Malx3 activator (Arg12Cys and Val13Ile) is capable to partly activate an IMA5 promoter containing an additional Yfl052w binding site. Each experiment was repeated at least three times with two biological replicates.

Figure 6. Possible evolutionary mutational path of MAL regulatory network diversification.

(a) Simplified phylogenetic tree of the fungal lineage. The numbers correspond to the key evolutionary events listed in b. WGD denotes the documented whole-genome duplication event in the fungal lineage. (b) Likely evolutionary path of the MAL regulatory network. The path starts from the common ancestor of L. elongisporus, S. bayanus and S. cerevisiae and ends at the modern day S. cerevisiae. In the common ancestor of L. elongisporus, S. bayanus and S. cerevisiae, maltose and isomaltose enzymatic activities are not separated and coexist in a single ancestral MalS enzyme, which is regulated by the single promiscuous MalR regulator. In the common ancestor of S. cerevisiae and K. thermotolerans, the MALS genes duplicated and neofunctionalized (1, 2), so that both types of target genes (maltose and palatinose specific) are present and are regulated by one promiscuous Malx3-like transcription factor that has an Arg residue at position 12 allowing it to bind both CGG and CGC motifs. The regulation is not specific at this point, that is, palatinose- and maltose-specific genes are equally expressed in presence of their respective substrate as well as a nonspecific disaccharide (as it is in S. bayanus). Two separate regulatory circuits that appear around the deviation of S. bayanus from the Saccharomyces tree. The MALR gene is duplicated (3) and this duplication event is followed by two single-nucleotide mutations in the first positions of the Arg12 and Val13 codons, changing these to Cys and Ile in one of the paralogues (4), thus preventing it from binding CGC motifs in the promoters of maltose-specific genes. Analysis of genomes that carry only one type MALR gene suggests that in the ancestral yeast CGG and CGC motifs were randomly distributed among maltose- and palatinose-specific genes. This implies that these binding sites needed to change in concert with the mutations in the MALR paralogues, so that palatinose-specific genes only contain one CGG site, and maltose-specific genes contain three CGC motifs so that they can still be activated by the weakened Malx3 paralogue (5, 6).