The evolution of combinatorial gene regulation in fungi

Abstract

It is widely suspected that gene regulatory networks are highly plastic. The rapid turnover of transcription factor binding sites has been predicted on theoretical grounds and has been experimentally demonstrated in closely related species. We combined experimental approaches with comparative genomics to focus on the role of combinatorial control in the evolution of a large transcriptional circuit in the fungal lineage. Our study centers on Mcm1, a transcriptional regulator that, in combination with five cofactors, binds roughly 4% of the genes in Saccharomyces cerevisiae and regulates processes ranging from the cell-cycle to mating. In Kluyveromyces lactis and Candida albicans, two other hemiascomycetes, we find that the Mcm1 combinatorial circuits are substantially different. This massive rewiring of the Mcm1 circuitry has involved both substantial gain and loss of targets in ancient combinatorial circuits as well as the formation of new combinatorial interactions. We have dissected the gains and losses on the global level into subsets of functionally and temporally related changes. One particularly dramatic change is the acquisition of Mcm1 binding sites in close proximity to Rap1 binding sites at 70 ribosomal protein genes in the K. lactis lineage. Another intriguing and very recent gain occurs in the C. albicans lineage, where Mcm1 is found to bind in combination with the regulator Wor1 at many genes that function in processes associated with adaptation to the human host, including the white-opaque epigenetic switch. The large turnover of Mcm1 binding sites and the evolution of new Mcm1-cofactor interactions illuminate in sharp detail the rapid evolution of combinatorial transcription networks.

Figure 1. Mcm1 cis-Regulatory Motifs in Three Species

The four cis-regulatory motifs identified by searching a high-confidence set of Mcm1-bound regions in the indicated species. In C. albicans, a noncanonical motif was found in addition to the canonical Mcm1 motif.

Figure 2. Comparison of Mcm1-Bound Target Genes in Three Species

(A and B) Mcm1 targeted gene sets are compared in a pairwise fashion between species. (A) The number of genes mapped from species A and also found to be in the Mcm1 bound gene set of species B, as a fraction of the total genes bound in species A that can be mapped to species B. (B) The significance (hypergeometric p-value) of each pairwise overlap. (C) The inference of gain and loss rates (green and red, respectively) along each branch of the rooted three species phylogeny. The inferred number of genes added and removed from the Mcm1 regulon is listed at the top and bottom of an arrow flanking each branch. The total counts for each of the eight possible occurrence patterns used as input to the inference algorithm are presented below the tree.

Figure 3. The Ancestral Mcm1-Bound Genes

These twelve genes are targets of Mcm1 in all three species. For each gene, the cell-cycle phase of increased expression [65] (if applicable), the relevant Mcm1 cofactor (if known), and a brief functional annotation is listed. Cell-cycle– and mating-type–regulated genes are shaded orange and blue, respectively.

Figure 4. Comparison of Mcm1–Cofactor Regulons across Species

(A) An example schematic of the Mcm1 homodimer and its cofactor, Yox1, binding in close proximity upstream of an M/G1-specific cell cycle gene. (B) Mcm1 associated cis-regulatory motifs discovered across the three species in this work. Each row of the table specifies an Mcm1–cofactor regulon and each column a species. The total number of Mcm1-bound regions in each species is listed in the header row. The number of Mcm1 bound regions assigned to each Mcm1–cofactor regulon in each species is listed in the upper right corner of each cell of the table; numbers colored black are based on Mcm1 ChIP data, whereas those in blue are not and are therefore more tentative. Mcm1 binds or is predicted to bind the consensus sequence denoted by the orange bar in each cell. The known or predicted cofactor motif is denoted by a blue bar in each cell. Motif graphics were generated with WebLogo [66]. (C) The three-way overlap of target genes in the Fkh2-Mcm1, Yox1-Mcm1, and asg (Mcm1-a2 or Mcm1-α2) regulons in the three species (Sc = S. cerevisiae, Kl = K. lactis, and Ca = C. albicans).

Figure 5. Evolution of Mcm1 Binding Sites at Ribosomal Genes in the Ascomycete Lineage

(A and B) Convergent evolution of Mcm1 motifs at ribosomal genes. (A) Four Mcm1-like cis-regulatory motifs discovered in a MEME search of the ribosomal gene promoters of 32 fully-sequenced ascomycete genomes. The motifs were discovered in the species indicated by the colored circles and oval in (B). The Mcm1-like motif of the A. nidulans branch has a tandem cofactor motif that is nearly identical to that derived from Snt2 ChIP-Chip experiments in S. cerevisiae [43]; we therefore predict that the Snt2 orthologs of the A. nidulans lineage are the Mcm1 cofactors at the ribosomal genes of this lineage. (B) The Mcm1 motifs from K. lactis (green circle) and the A. nidulans lineage (lavender oval) were used to score ribosomal promoters across the ascomycete lineage and thus to verify that presence of the Mcm1 motifs is limited to the four lineages in which Mcm1-like motifs were found de novo by MEME. The significance of motif enrichment at the ribosomal promoters of each species was determined by comparison to genome-wide background frequencies of occurrence using the binomial distribution. See Text S1 for description of the ascomycete phylogeny reconstruction [24,67]. (C) An additional motif similar to that recognized by Rap1 in S. cerevisiae was discovered in the MEME search of K. lactis ribosomal promoters. (D) In K. lactis, the positioning of Rap1-like motif instances is constrained relative to Mcm1 motif instances.

Figure 6. Substitutions within the MADS Box Domain of Mcm1

There are a few substitutions to the MADS box domain of Mcm1 (orange) against a background of strong conservation (white) within the hemiascomycete and euascomycete lineages. The shaded box indicates Mcm1 orthologs from species that also have an Mcm1 duplicate (named Arg80 in S. cerevisiae). Mcm1 residues forming contacts with α2, Mcm1, or DNA in the crystal structure of the α2-Mcm1-DNA ternary complex (Protein Databank ID: 1mnm) are indicated above the alignment with squares, triangles, and circles, respectively. Note the strong correlation between those species having substitutions at the α2 interacting residues, those species with an Mcm1 duplicate, and those species thought to be using a purely negative mode of asg regulation by Mcm1 and α2[24].

Figure 7. Recent Evolution of Noncanonical Mcm1 Binding Sites at White-Opaque Genes

(A–C) C. albicans cell types. (A) White cells. (B) Opaque cells. (C) A white colony (Wh) and an opaque colony (Op). (D) The noncanonical and canonical Mcm1 motif matrices of C. albicans (Figure 1) were used to score promoters for two sets of genes (genes where Mcm1 is found at the noncanonical motif in C. albicans and genes where Mcm1 is found at the canonical motif in C. albicans) across the ascomycete lineage. The significance of motif enrichment at the two mapped gene sets of each species was determined by comparison to genome-wide background frequencies of occurrence using the binomial distribution. (E) ChIP-Chip profiles for Mcm1 and Wor1 in regions flanking four key regulators of the white-opaque switch [45]. Dark blue, aqua, and red lines indicate the Mcm1 ChIP of white cells, Mcm1 ChIP of opaque cells, and Wor1 ChIP of opaque cells, respectively. Yellow circles indicate a noncanonical Mcm1 motif.